0:00 In this video, you'll learn everything 0:01 that you need to know about e-commerce 0:03 in 2026. We spent over two hours on 0:06 creative strategy, over an hour on offer 0:08 design. We go into Google Ads technical 0:11 account structure, meta ads technical 0:12 account structure, as well as everything 0:14 that you need to know about finance to 0:17 be able to scale profitably. It doesn't 0:18 matter if you're doing 7, 8, 9, or even 0:20 10 figures in revenue because almost all 0:22 the content throughout this video is 0:24 targeted at eight and nine figure 0:26 brands. Now, before we start, if you're 0:27 a performance marketer, please reach out 0:29 to us at hiringbluensedigital.com 0:31 if you're looking for a role. And if 0:33 you're a brand doing over $5 million a 0:34 year in revenue, click the link in the 0:36 description and reach out to us for a 0:38 free audit where everything that you're 0:39 about to learn in this video will get 0:41 translated and applied specifically into 0:43 your accounts to help you achieve scale. 0:45 With that being said, let's dive in. 0:46 I've completed over a,000 audits on 0:48 seven, eight, and nine figure brands. 0:50 And this bottleneck is the most common 0:52 one that people completely overlook. 0:54 It's data integrity and assurance. What 0:56 this really comes down to is using 0:58 metrics that directly correlate to the 1:01 actual commercial outcomes within the 1:02 business and more using metrics that we 1:04 can trust. So firstly, we're going to be 1:06 going through why data integrity is the 1:08 bottleneck in about 30% of businesses 1:10 that I audit. Then we're going to go 1:12 into platform reported rorowaz, why it's 1:14 lying to you, how to fix it. We'll then 1:16 talk about attribution windows, go into 1:18 better metrics, the best metric. We'll 1:20 then show you how to reconcile the 1:22 platform data against the P&L. We'll go 1:24 into incrementality testing and causal 1:26 inference for those businesses that are 1:27 a little bit bigger. Then we'll talk 1:29 about tracking infrastructure, the nine 1:31 patterns that we see in every Blue Sense 1:33 audit so that you don't make the same 1:35 mistakes. And then we'll wrap it all 1:36 together through the playbook. The best 1:38 way to explain why data integrity is a 1:40 bottleneck in a lot of businesses is 1:41 because every business is just a bunch 1:44 of feedback loops. And so you collect 1:46 data over here. You then make a decision 1:48 with that data that then goes into 1:51 action. Then you collect more data and 1:53 you make another decision. Flywheel 1:54 effects. You get this specifically in 1:56 creative within ecom. And so you will 1:59 look at an ad down here and you will go 2:01 this ad is performing well. It's at a 4x 2:04 return. Amazing. We're going to do two 2:06 things. Number one, we're going to go 2:07 and make more ads like this. We'll tell 2:09 the creative team. Number two, let's put 2:11 way more spend in here. So you'll make 2:12 the decision. Then you'll collect more 2:14 data. Maybe rorowaz falls off a little 2:16 bit and it goes to a 3x at the higher 2:18 spend. And then we go and launch a bunch 2:20 more ads. and these ads are all 2:21 performing well. Amazing. So, we 2:23 reinforce into this decision even harder 2:25 and we continue this feedback loop. Now, 2:27 this entire thing breaks if this number 2:30 right here was wrong. What if it 2:32 actually wasn't at a 4X? But you made 2:34 all these decisions under the assumption 2:35 that it was driving a commercial 2:36 outcome. But once again, what if it 2:38 wasn't true? What if this ad actually 2:40 wasn't driving top ofunnel awareness? 2:42 What if this number was substantially 2:44 overinflated? What if the number didn't 2:46 matter at all? Well, then all of a 2:47 sudden you're making decisions within 2:49 the business that seem to be making an 2:51 impact when we zoom in on this one 2:52 particular metric and this one number. 2:54 But when we zoom out, profit isn't going 2:56 up, revenue isn't going up, the business 2:58 doesn't seem to be getting healthier, 2:59 and we feel we're making decisions. We 3:01 feel we're analyzing data and we're 3:02 constantly iterating, but it's doing 3:04 nothing. And that feeling of being on a 3:06 treadmill and not moving is the feeling 3:09 that your data is wrong. the metrics 3:11 that you're using to make decisions are 3:13 not actually stepstone starting up or 3:14 commercially aligning to correct 3:15 decision- making that's fixing the 3:17 business. And so when you feel like 3:18 you're just spinning your wheels, it 3:19 usually ends up being a metric problem. 3:22 Now, I don't want to blame everything 3:23 within an ecom business on just data 3:25 integrity. Really, there's three pillars 3:27 that we see in an e-commerce business, 3:29 which is that there's the actual 3:30 technical account structure. How are the 3:31 different platforms being structured to 3:33 be able to distribute budgets and learn 3:35 based on the actual commercial 3:36 objectives of the business? Number two 3:38 is creative. What is the creative 3:40 velocity? What's the diversity? What's 3:41 the strategy here? And then number three 3:44 is data integrity and assurance and 3:46 measurement. Now, even though I've put 3:47 it in this order, data integrity 3:49 actually forms the base. So then we can 3:52 actually work on creative and then 3:54 ultimately technical account structure 3:55 sits up the top here. And the reason 3:57 this is so critical is because the 3:58 account structure and creative is 4:00 meaningless unless the metrics actually 4:02 make sense and we have congruent 4:04 decision-making. Then technical account 4:06 structure means nothing unless we have 4:07 creative in place as well. That's it at 4:09 a velocity that's required. It's good 4:11 creative and it's diverse. And then only 4:13 does technical account structure matter 4:15 if these two are actually in place in 4:17 the first place. Now the reason this 4:19 matters more than ever right now is 4:21 because for the last 5 years, iOS 14, 4:23 cookie depreciation, modeled 4:24 conversions, and Meta's increased 4:26 reluctance to actually expose the 4:28 attribution mechanics have all pushed 4:30 brands further away from causal truth. 4:32 Platform rise has become a number that 4:34 lives further and further and further 4:36 away from the P&L. Most brands have 4:39 responded by either ignoring the problem 4:40 or they buy third party attribution 4:42 tools which just end up compounding the 4:44 issue and causing even more confusion 4:46 internally. The brands that are still 4:48 scaling in 2026 are the ones that built 4:50 a measurement layer that they can defend 4:52 internally and that they can reconcile 4:54 against actual commercial outcomes. And 4:56 I cannot stress this enough. A business 5:00 will always not grow due to one 5:02 particular bottleneck. There's always 5:04 one thing that is stopping growth. And 5:05 then as you grow, another bottleneck 5:07 appears. And so growing a business is 5:08 the product of continuously solving 5:10 bottlenecks over and over again. And 5:12 that's what we do for clients because 5:13 ultimately our goal is to grow the 5:15 business primarily through paid media. 5:17 But if there is a bottleneck that is not 5:18 allowing us to push paid, we will go and 5:20 address it. Now the real question 5:21 becomes okay if solving bottlenecks is 5:23 the way to grow if that's really what 5:25 you need to understand and be able to 5:27 identify well then how do we identify 5:29 bottlenecks as fast as possible know 5:32 that that is the bottleneck that's where 5:33 we should put all the resourcing and 5:34 then continue to grow and the way that 5:37 you diagnose and find bottlenecks is 5:40 through data and if the data doesn't 5:42 have integrity if the data is wrong if 5:44 the data is meaningless irrespective to 5:46 our goals if the data is giving us false 5:49 flags then we will never be able to 5:50 diagnose the bottleneck. And so it all 5:52 starts with having accurate, correct, 5:55 measurable data that we can then use to 5:58 go and identify where the bottleneck is 6:00 in the business that we can unlock. If 6:02 this isn't in place, data integrity ends 6:05 up typically being the bottleneck in the 6:06 business. So platform reported rorowaz, 6:08 which has been the gold standard for the 6:10 last 5 to 10 years. is I'm actually 6:11 going into an RFP in a few days time 6:13 with a 9 figureure brand and they are 6:15 telling us how can we improve our 6:17 reported rorowaz and the answer is we're 6:20 not going to do that and I'm going to 6:21 explain why and why this is a terrible 6:23 KPI and if you are northstarring against 6:25 this well you're probably 3 years behind 6:28 competitors who are not operating under 6:30 this model and they will ultimately beat 6:32 you so you will ultimately have to move 6:34 away from this kind of orientation into 6:36 the platforms but hopefully I can speed 6:38 that process up for you by explaining 6:39 why you shouldn't be doing this. So 6:41 number one, all rorowaz 6:44 uh is not equal and this applies through 6:46 a multitude of different factors. So 6:49 number one, this is all revenue isn't 6:51 equal. So what that means is that we can 6:54 have two separate campaigns. We can have 6:56 campaign one over here and campaign two. 6:58 They can both have a full return on ad 7:00 spend on them and they could have both 7:02 driven, let's say, $10,000 in attributed 7:05 revenue. But the margin profile across 7:07 these two different $10,000 of revenue 7:09 could be very different because this one 7:11 could be a discount campaign that has 7:13 eroded margin. This one could have 7:16 products that have a lower margin. This 7:17 one up here could be a retention 7:19 campaign that's full price whereas this 7:20 one's using some kind of discount code 7:22 on entry. The margin profile changes 7:25 across different campaigns. And so 7:27 rorowaz here and rorowaz here doesn't 7:29 actually mean the same amount of profit 7:31 contribution. And so we might go into a 7:33 profit contribution calculation and find 7:34 out that this top campaign is at three, 7:36 this bottom campaign is at four. And so 7:39 if we actually look a little bit deeper 7:40 into the profitability of these 7:41 campaigns, we would say, "Oh, actually 7:43 we want to spend way more money here. 7:45 Even though it's the same rorowaz on the 7:47 surface, this campaign generates a lot 7:49 more profit." That is number one. Number 7:51 two is that all rorowaz isn't equal 7:54 across different platforms when we 7:57 compare them. So if you take a really 8:00 good example of this is Pinterest and 8:02 you go and look at your Pinterest return 8:04 on ad spend and then you go and compare 8:07 this to we could give a really extreme 8:10 example here and we could call this TV, 8:11 right? Cuz TV has a zero rorowaz on it 8:14 because there's no attribution but let's 8:16 keep it within the realm of digital and 8:18 go meta. Maybe your meta rorowaz is a 8:20 2.5. Your Pinterest is a seven. Now you 8:23 look at this objectively and you go well 8:25 Pinterest is where we should put more 8:26 money. Let's not put more money into 8:28 meta. Let's go Pinterest. Issue is the 8:30 way that Pinterest attributes is it 8:33 includes viewthroughs. So if a user sees 8:35 an ad and then buys, they don't have to 8:36 click on it. They don't have to interact 8:38 with it. It will attribute the 8:39 conversion. And then number two, if 8:41 you've ever gone on Pinterest, there's a 8:42 bunch of tiles all over the place. 8:44 There's like six to 12 tiles depending 8:46 on the size of your screen. As long as 8:47 one of them is an ad, it can claim the 8:49 view through conversion. And so what 8:50 Pinterest tends to do is it will just 8:52 reserve to warm audiences that have been 8:54 on the website in the last few days. and 8:56 that person was probably going to buy 8:57 anyway. You get to serve a pin to them. 8:59 It gets to claim credit. You get massive 9:01 overattribution. And so comparing these 9:04 two numbers is a terrible idea. It's not 9:06 a good exercise. Same thing with 9:08 comparing this over to Google or another 9:09 cuz they all have different attribution 9:11 models. They all fundamentally work 9:13 different and they all sit at different 9:15 stages of the funnel. So crossplatform 9:17 rorowaz is not equal and all rorowaz is 9:20 not equal across the funnel which is a 9:23 development of this idea but it applies 9:25 into the platform too. So if we have a 9:28 meta campaign that's a cold targeting 9:30 campaign and the reason it's cold 9:32 targeting is that we have existing 9:33 customer lists excluded. So it is cold 9:36 and we have very top ofunnel creative in 9:38 here too. So we're also getting the 9:40 algorithm to serve our ads to cold 9:41 audiences because of the creative type. 9:43 Then down the bottom here we have let's 9:45 go to the absolute extreme and say that 9:48 this is a bottom offunnel retargeting 9:50 campaign for existing customers. Now 9:52 this campaign down the bottom here might 9:54 have a seven rorowaz and this campaign 9:57 up the top here might have a two. Now 9:59 naturally once again you look at these 10:00 two numbers and you go let's put spend 10:02 here. This is a great return. This is 10:04 going to grow the business. Let's not 10:05 put it up here. But the issue is this 10:07 sits at the bottom of the funnel. So the 10:08 only way the business grows is if more 10:10 spend gets injected at the top 10:12 efficiently. If we start pulling spend 10:14 away from the top and redirecting it to 10:15 the bottom, that's when the business 10:17 starts to collapse cuz we're not 10:18 bringing new cold audiences in to begin 10:20 to convert them. And you see this all 10:22 the time, particularly with something 10:24 like a middle of funnel campaign that 10:26 sits in here that's maybe at a 4x. 10:28 You'll see all the time people will go, 10:29 "Oh, let's just take 30% of spend and 10:32 push it down here. Let's get some more 10:33 spend in middle and bottom." You rarely 10:35 ever need more spend in middle and 10:37 bottom of funnel. Almost everyone 10:39 overspends here. You need more spend on 10:41 like a YouTube cold targeting campaign. 10:43 You need more spend on Tik Tok. You need 10:45 more spend on all of these really 10:46 upperfunnel channels because this is 10:48 what's going to drive conversions down 10:49 the line in these middle of funnel and 10:51 bottom of funnel campaigns. And I'll 10:52 give you an example like a tricky 10:54 example of how this will start to play 10:55 out really confuse your decision-m which 10:58 is that let's take this meta funnel 11:01 right here and then let's say adjacent 11:03 to this you decide as a business that 11:05 you want to test out Tik Tok. So you go 11:07 and launch Tik Tok over here and maybe 11:08 you spend 25% of your meta budget on the 11:11 platform doesn't seem to really do that 11:13 well. It's at like a 1x maybe a 2x. Nah, 11:16 we can't really scale this. But at the 11:18 exact same time, this middle ofunnel 11:21 campaign here on Meta suddenly jumps 11:23 from a 4x to a 12x. Now why did it jump? 11:26 Well, it jumped up because all of this 11:29 Tik Tok traffic that you were driving 11:30 started getting retargeted on Meta and 11:33 then started converting. However, most 11:35 people will look at Meta and go, "Oh, 11:37 our middle of funnel campaign is doing a 11:39 lot better. Let's increase budgets and 11:41 spend more here." They might even take 11:42 it a step further and go, I think that's 11:45 probably because cold targeting is doing 11:46 well. So, if we're going to up budgets 11:47 here, let's up budgets on cold 11:49 targeting, too, because overall meta is 11:50 doing well. This had nothing to do with 11:52 Meta doing well. Increasing the budget 11:53 here is just going to be a waste of 11:54 money. The reason why Meta is 11:56 overattributing now is because of the 11:58 Tik Tok spend. And so looking at rorowaz 12:00 at a platform level, comparing platforms 12:02 is not helpful. Looking at it through 12:04 the funnel, it's going to change. And so 12:07 all rorowaz is not equal. A rorowaz up 12:09 here of a two is much better than a 12:10 rorowaz down here of a seven. And then 12:12 this directional decision-m based on a 12:14 return on ad spend number is going to 12:16 mislead you in b budget allocations 12:18 across the platform. Now there's three 12:19 main sources of overinflation in return 12:22 on ad spend. Number one, which you'll 12:24 hear me say so much throughout this 12:26 video, is view through conversions. So, 12:29 to explain this out of the gates on 12:31 Meta, by default, you use 7-day click, 12:36 one day view. And actually, these days, 12:37 this will be 7-day click, one day view, 12:40 one day engaged. So, what this means is 12:42 that if a user clicks on an ad and then 12:44 buys within 7 days, Meta can claim the 12:46 conversion, which is somewhat 12:48 reasonable. If someone clicked and 12:49 interacted with your ad and then 12:50 purchased, we want that attributed to 12:52 the campaign. One day view means if 12:55 someone simply viewed an ad, they don't 12:57 have to watch through the ad. They don't 12:58 have to click on the ad. They don't have 13:00 to interact with it in any way. It just 13:02 needs to serve into their feed and then 13:03 they have to buy within 24 hours. Meta 13:06 can also claim the conversion. And then 13:09 lastly, one day engaged view is a 13:11 definition of if someone watched more 13:13 than 3 seconds or liked or commented or 13:15 interacted but didn't actually do an 13:17 outbound click. That will also get 13:19 counted over here. Now, 7-day click is 13:20 fine. And one day engaged is fine. But 13:22 the one day view is where a lot of 13:24 overattribution ends up occurring. 13:27 Because if you're uh someone who's 13:29 already visited the website, if you're 13:30 an existing customer, you just get 13:32 served an ad, you are going to buy 13:33 anyway irrespective of whether the ad 13:35 served to you and then metagos and 13:37 claims your conversion or attributes it 13:38 to the ad. And so anytime you have one 13:40 day view in your reporting, you end up 13:42 with a rorowaz number that's way above 13:44 the actual reality of the incremental 13:46 impact of that campaign in the business. 13:48 Number two is existing customer bleed 13:51 over. So this is pretty obvious, but if 13:53 existing customers are getting targeted 13:55 within your cold campaign. Well, you 13:58 will end up with a bunch of 13:59 overattribution because existing 14:01 customers will view the ads. Existing 14:03 customers are probably going to buy 14:04 again anyway and it gets credited in. 14:07 Now let's say you removed user 14:09 attribution from your attribution model. 14:12 This will still overattribute because a 14:15 click from an existing customer on an ad 14:17 doesn't actually infer a causal 14:19 relationship. And we'll go into 14:21 causality in more detail a little bit 14:23 later on. But the reality is that if I'm 14:25 an existing customer, I've bought from 14:26 you before. Let's say I'm buying 14:28 supplements and then I get an ad from 14:30 you for the supplement. Am I purchasing 14:32 because I got the ad or am I purchasing 14:34 because I'm like, "Oh yeah, I only have 14:36 5 days left in the pack and I might as 14:37 well buy right now." And then I click 14:38 through and buy. probably would have 14:39 bought anyway regardless once the actual 14:42 supplement ran out of my kitchen. It's 14:44 just the ad reminding me a couple days 14:45 earlier, oh yeah, let me get that 14:46 through sooner. Is that actually 14:48 genuinely incremental to the business? 14:50 Probably not. And so you still end up 14:52 with a good amount of over attribution 14:54 here. And then lastly is the halo 14:56 effect. And so you see this all the time 14:58 with large retailers. If you're in a big 15:00 business or you've ran ad accounts for 15:02 big brands, which is that they will do 15:04 some kind of activation that sits way 15:06 outside of the ad account, way outside 15:07 of digital. Then everything in digital 15:09 just improves. Okay, there's some kind 15:11 of pop-up activation in New York and 15:13 then suddenly all the performance in New 15:15 York in the ad account triples and 15:16 you're like, "Huh, if you didn't know 15:18 that the popup and the activation was 15:20 occurring, you would look at the ad 15:21 account and you go, I it doesn't make 15:23 sense. The data makes no sense. How is 15:25 rorow so much better?" But it's because 15:27 of this halo effect being caused from 15:28 elsewhere. And so if you then made the 15:30 decision, oh, our rorowes is up in New 15:32 York this week. Let's pump spend. 15:33 probably be a bad decision because the 15:35 rorowaz going up actually has nothing to 15:37 do with the causal impact of the spend, 15:39 but it has to do with a third party 15:41 confounding event that's then making it 15:43 look causal when it's actually 15:44 correlated. Now, it's worth noting here 15:46 too that inflation will compound with 15:49 the maturity of the business. So, this 15:52 becomes a larger and larger problem for 15:54 the bigger the business is. And this is 15:56 why this has become a very large focus 15:58 of ours as we primarily now only work 16:00 with 8, 9, and 10 figure businesses is 16:02 that they pretty much all have this 16:04 problem. This is why this entire video 16:05 exists because data integrity and 16:07 assurance is so critical in an 8 to 9 16:10 figure business, but almost no one is 16:12 doing it actually well. Now, why does 16:14 maturity cause a further overinflation? 16:16 Well, it's because there's way more 16:18 existing customers. So, there's way more 16:20 opportunity for retargeting and 16:21 overinflation. There is way more 16:23 channels that they're spending on. Now, 16:25 this might even just be retail stores. 16:27 You could think of retail stores as 16:28 advertising because the store is there 16:30 as people walk in. So, this also causes 16:32 overinflation in online. And then a 16:34 mature business that's been around for a 16:36 while also likely has much larger 16:39 organic presence. And so, the organic is 16:41 overinflating all of the paid as well. 16:43 So, the bigger you get, the less 16:45 reliable return on ad spend becomes as a 16:48 read at any level within the attribution 16:51 funnel. So, how do we actually start to 16:53 improve return on ad spend then is there 16:55 things that we can do? There is so much 16:57 that you can do. What most teams will do 16:59 once they wrap their head around this 17:00 and I'd say 50% of the market 17:02 understands everything that I've just 17:03 gone through. So, a lot of people are 17:04 across this. And so, the solution 17:06 becomes let's add let's add more. Okay? 17:09 So, let's add in attribution tools. 17:11 Let's add in more dashboards. Let's add 17:12 in more conversion events. Let's add in 17:14 more platform settings to try to fix 17:16 this problem. Addition is not the 17:18 solution. We want to subtract. We want 17:21 to remove as much noise as possible. We 17:24 don't want to add more noise into the 17:25 bucket. We don't want to add more 17:27 metrics. We want to simplify everything 17:29 down to where we have just core KPIs 17:32 that are commercially orientated and we 17:35 can make clear easy decisions. The more 17:37 you add and add and add, the more 17:39 confusion there becomes. That's when you 17:41 sit in a meeting with a seauite and one 17:43 person is going oh but the this 17:45 attribution model on triple whale is 17:47 saying that Pinterest is actually bad. 17:49 And then someone else is going, "Well, 17:50 we ran an incrementality test on 17:51 Pinterest and it was actually good." And 17:53 then someone else is arguing that they 17:54 ran a causal inference test and that it 17:56 was actually bad. And there's all of 17:58 these confounding voices based on all of 17:59 this different data, which is why adding 18:01 more data generally doesn't help. We 18:03 want to strip data away. We want to get 18:05 a priority, a hierarchy of metrics. What 18:07 is the most important? That's the dec. 18:09 If that's good, that's the decide. And 18:11 then as you start to move down the 18:12 hierarchy, it gives an ability for you 18:14 to prioritize what to actually do and 18:16 what not to do. So in regards to 18:17 attribution windows, I think it's really 18:19 important for everyone to understand 18:21 what attribution window your particular 18:24 platform is using. So on Google, there 18:26 will typically be a 30-day click window, 18:29 which means if someone clicks and then 18:30 buys within 30 days, Google can claim 18:32 the conversion plus typically a 7-day 18:34 view, a 7-day engage view, and a 1-day 18:36 view window. On Meta, typically by 18:38 default, it's 7-day click 1-day view. 18:40 Typically on Tik Tok, it's around about 18:42 the same, but you can change it, you can 18:44 increase it, you can decrease it. Same 18:45 thing on Pinterest, Microsoft's a little 18:47 bit different, etc. So, firstly, you 18:48 want to understand what is the 18:50 attribution window on each of our 18:52 platforms because that will change the 18:54 incrementality factor of that number. 18:56 Now, if we just silo into meta because 18:58 that's where about 60 to 70% of our 19:00 spend sits. By default, you will be 7day 19:03 click, one day view, one day engaged. I 19:06 recommend changing this on your 19:08 campaigns moving forward to 7-day click, 19:11 one day engaged. Remove the view 19:13 throughs. Now, you can even do this 19:15 purely from a reporting perspective by 19:17 hitting columns, compare attribution 19:19 settings, and then opening up 7-day 19:21 click reporting. This will correlate 19:23 much more tightly to the actual 19:25 efficiency numbers within the P&L, which 19:27 we'll dive into a little bit later. The 19:29 reason why we have one day engaged in 19:31 here is because the 7-day click 19:33 definition actually changed about 2 19:35 months ago from today. So, right now 19:37 it's the 5th of May 2026. In I believe 19:41 April, maybe March 2026. 7-day click 19:44 used to be defined as someone who would 19:46 click on an ad and buy or someone who 19:49 would interact with a post with a click. 19:51 And so it didn't have to be an outbound 19:53 click to the website. It could be a 19:55 click on the like button. It could be a 19:56 comment. It could be a share. It could 19:58 be a click to the page. As long as it 20:00 was a click on the ad, and it didn't 20:02 have to be an outbound click, it would 20:04 count within the conversions, which I 20:06 think is something that not many people 20:07 knew. Okay? I could comment on an ad and 20:09 then as long as I bought, Meta would 20:11 claim that as a conversion to that 20:13 campaign on a click basis. Now, Meta 20:15 rewrote the definition recently and 20:17 removed it. And so now 7-day click has 20:19 to be an outbound click. And so all of 20:21 the 7-day click rorowaz numbers 20:22 inherently got a little bit worse 20:24 overnight because there was a bunch of 20:26 conversions that were getting counted 20:27 that don't get counted anymore. How they 20:28 then factored in this is that all of 20:31 those engagement clicks. So people that 20:33 liked, comment, share, go to the page, 20:35 this is now called a one day engaged. So 20:38 this is if someone engages. So we want 20:41 to include this in here because this is 20:42 the traditional definition of a click. 20:45 It's just moved into the engaged 20:46 audience. And if someone's engaging with 20:48 an ad and then they buy, typically 20:50 there's probably some kind of causal 20:52 relationship. Now to take this one step 20:53 further in meta, we also have something 20:56 called incremental attribution which 20:58 rolled out about 6 months ago now. Now 21:00 you can look at your incremental 21:01 attribution by once again going and 21:03 hitting on columns, hit compare 21:04 attribution setting, hit incremental 21:06 attribution and break it out. Now this 21:08 is a really good measure because what it 21:11 does is let's say you're targeting 21:13 10,000 people a day in this square. Meta 21:16 will go and hold out 10% of users and 21:18 measure if these people are still buying 21:20 compared to people that are seeing ads. 21:22 And let's say that of the people that 21:23 are seeing your ads, 2% of them are 21:25 purchasing from you. People that aren't 21:27 seeing your ads, 1% of people are still 21:29 purchasing from you. Now, why is this 21:31 the case? Well, typically this is the 21:33 case in larger retail businesses because 21:35 everyone already knows who you are. 21:37 People are already buying from you all 21:38 the time. If you're if no one knows who 21:40 you are, let's say you're doing $10,000 21:42 a month in revenue, obviously this will 21:44 be 0%. People aren't going to buy from 21:45 you unless they see your ad. But if 21:47 people know who you are, they're going 21:49 to buy from you regardless. A percentage 21:50 will be buying from you. Now, we do a 21:52 difference in difference calculation, 21:54 which is just minusing the two. So 2 - 1 21:56 = 1. And we know that the actual impact 21:59 of ads is not 2%, it's 1%. So if this 22:01 was saying a five return on ad spend 22:03 over here, well, no, it's actually a 2.5 22:06 return on ad spend. So this is 22:08 ultimately how uh lift tests work within 22:10 meta. This is called a conversion lift 22:13 test. Then what Meta does, and this is 22:15 what most people don't actually know, is 22:17 that Meta is not doing this in real time 22:19 and then applying it to incremental 22:21 attribution. So there aren't real-time 22:23 hold out groups occurring in your ad 22:24 account. Instead, Meta takes conversion 22:27 lift experiments from your competitors 22:29 and based on the reads that they get of 22:31 how incremental they are, it then goes 22:33 and applies a factor into your ad 22:34 account. And so the incremental 22:35 attribution read isn't a accurate real 22:39 lift experiment outcome within the ad 22:42 account. It is a estimation based on 22:45 your niche and industries conversion 22:47 lift test that competitors have run. So 22:49 it's not a perfect measure but it's good 22:51 directional advice on what kind of 22:53 campaigns are actually doing well using 22:56 proper lift tests as the experimental 22:58 data. Now before we wrap this section up 23:00 I want to quickly mention attribution 23:02 models which is using some kind of 23:03 third-party attribution tool. So, why 23:06 don't we just solve all of these 23:07 problems by adding a third party 23:09 attribution tool in that can track users 23:11 using an external pixel who click on 23:13 Meta and then click on Google and then 23:15 let's say they go over to uh Pinterest 23:18 and then ultimately they buy, right? We 23:20 can just track this using a third party 23:22 attribution tool. Why don't we do that? 23:24 The reason being is that this is an 23:25 unknowable reality. We can never track 23:27 someone's conversion journey perfectly 23:29 and so we just playing guessing games 23:31 and adding more rather than subtracting. 23:34 And so we're introducing more confusion 23:36 into decisionm. The real idea here is 23:39 that the most valuable clicks within or 23:42 the most valuable interactions within a 23:44 consumer journey is the first click and 23:46 the last click ultimately because we 23:49 want to know how do people first find 23:51 out about us that ultimately leads them 23:53 to a conversion down the line and then 23:55 how do they convert? Because then what 23:57 we need is we need to crank [snorts] the 23:59 top of funnel as much as possible on how 24:01 people initially hear about us and then 24:03 we need to make sure we modulate the 24:04 bottom of funnel enough to make sure 24:06 that these people end up converting down 24:08 the line. Now the tough thing with third 24:10 party attribution models is the last 24:11 click we pretty much always have 24:12 accuracy on. We know when people are 24:15 clicking and then buying, but we know 24:16 that in the platform anyway. We don't 24:18 need a third party attribution model to 24:19 tell us where bottom of funnel is. It's 24:20 pretty obvious. Well, what about first 24:22 click? First click is almost unknowable. 24:24 first click attribution doesn't exist 24:26 because the cookies or the attribution 24:29 always breaks at some point in the line. 24:32 And I say this because I've spent 24:34 hundreds of hours in third party 24:36 attribution tools trying to understand 24:38 where the first click is occurring and 24:39 it's like always inaccurate. You go into 24:42 individual users who have purchased from 24:44 you and you can look at their entire uh 24:45 click journey across all the different 24:47 platforms. Half the time it says that 24:49 the first click over here came from 24:51 brand search and you're like how is that 24:54 even possible? How was the first time 24:56 they ever interacted with the business 24:58 from a brand search? Like obviously it 25:00 wasn't obviously probably something 25:02 let's say it was well they probably saw 25:04 their friend wearing the product back 25:06 here and that was the first interaction 25:08 with the brand. And so there was this 25:09 virality component particularly in 25:11 fashion with people wearing the product 25:13 around and that's actually what caused 25:14 the interest that then pushed to the 25:15 brand search that then pushed through 25:18 all of these attribution touch points. 25:20 And so if we saw this we would go oh 25:21 let's put more spend into brand search 25:23 but that's not actually what we need. We 25:25 need more of whatever this is over here. 25:27 All right. So what are better metrics? 25:28 Well, you can think about how accurate a 25:31 metric is correlated to the actual 25:34 commercial outcome in the business based 25:36 on these three layers of the pyramid. 25:39 Down the bottom, you have attributed 25:40 numbers. So, you have stuff like 25:42 rorowaz, CPA, uh, multi-touch 25:44 attribution, etc. Then, as we move up 25:48 one, we look at finance grade level 25:50 metrics. So, rather than relying on 25:52 attribution, we're just looking at 25:54 actual commercial outcomes. And so what 25:56 is the actual profit contribution in the 25:58 business? What is the me of the 26:00 business? What is the acquisition me of 26:02 the business? Then we move all the way 26:04 up the top here to the best causal read 26:06 of what is actually occurring which is 26:08 with studies which is where we take 26:11 control groups and treatment groups and 26:13 we make a change to the treatment group 26:15 and we measure against the control to 26:17 understand the actual causal impact of 26:19 that change. Now, as you go up the 26:21 pyramid, this becomes much longer and 26:24 harder to do, and so the actual time to 26:27 feedback slows down. As you move down 26:30 the pyramid, it becomes a lot faster, 26:32 but you lose reliability. So, the 26:34 highest reliability is up the top here. 26:37 You ultimately want to be making very 26:38 large decisions within the business 26:41 based on lift studies or financial 26:43 metrics. But if you want to make quick 26:46 directional decisions dayto-day, the 26:48 platform level metrics can be helpful as 26:50 long as we understand the nuances of the 26:51 attribution. So I want to start listing 26:54 off the metrics in this section to give 26:56 you an idea of what you can track, why I 26:58 wouldn't track some of them and what 27:00 they become helpful for. So the most 27:02 common one you will always hear is me, 27:05 your marketing efficiency ratio. Now 27:07 people calculate this in two different 27:08 ways. They either take total ad spend 27:11 and they divide by total revenue. And so 27:13 this gives you a percentage. Or they do 27:16 the opposite and they do revenue divided 27:17 by ad spend which gives you an integer 27:19 like a 5x. And so 20% me and 5x me this 27:23 is the same number. We're just flipping 27:25 the division over itself. Now the reason 27:27 why these two different versions exist 27:28 is this is more of a marketer metric and 27:30 this is more of a CFO metric. The CFO 27:33 wants to understand what percentage of 27:34 revenue is getting allocated to 27:36 marketing spend so they can control it 27:37 over time. Hence they want to see a 27:39 percentage. A marketer wants to see ROI. 27:41 If we spend a dollar, how many dollars 27:42 are we getting out? And hence marketers 27:44 will typically look at it in this 27:45 direction. Now, one common mistake in 27:48 how MEI is miscalculated is that people 27:50 use total revenue rather than using net. 27:53 And so, you always do want to delineate 27:55 down into net revenue, which minuses out 27:58 returns. Obviously, really important in 27:59 fashion uh because if you look at total 28:02 revenue today, it will compress tomorrow 28:04 once the returns get processed. And so, 28:06 you always want to be going off in net 28:08 value. Now, ME is a fantastic metric for 28:13 CFOs because it allows you to have 28:15 visibility into percentage allocation on 28:17 the total P&L towards marketing and 28:19 control it. As long as you can control 28:21 this number and you can control gross 28:22 margin and operating expenses, you can 28:25 ensure consistent deliverability of a 28:28 percentage profit target. Now, the issue 28:31 with this metric is it's fundamentally a 28:33 bad metric for marketers. So the CMO 28:37 should not be held accountable to this 28:40 metric. The reason being is that paid 28:42 media or at least most of your 28:44 advertising dollars are not going 28:45 towards driving returning customer 28:47 revenue. Your advertising dollars are 28:49 going towards driving new customer 28:51 revenue because that is ultimately how 28:53 the business grows. And this is a really 28:54 important thing to understand which is 28:55 that if you stop acquiring new 28:57 customers, the business's revenue will 29:00 look like this. And the reason it will 29:01 look like this is because this is cohort 29:03 growth over time. And so when people 29:05 repeat in your cohort analysis, people 29:08 will repeat at a high percentage and 29:09 then it will decline. And so maybe 8% of 29:12 people come back after month one of 29:13 buying the product, then 6% 4% 3% 2% 1% 29:16 and then it asotopes down to quite a low 29:18 number. If you stop filling up the 29:20 bucket, your total revenue will just 29:22 follow that curve. And so you need to 29:23 continuously increase new customer 29:25 acquisition or be putting new customers 29:27 in to be able to maintain revenue or 29:30 increase it. And so we really want to 29:31 make sure that the media spend that 29:33 we're putting towards marketing is 29:35 driving new customer acquisition. 29:37 Therefore, we want to KPI it against new 29:39 customer acquisition. So instead, we 29:41 have a me, which is acquisition 29:44 marketing efficiency ratio. This is the 29:46 exact same formula, but instead we're 29:48 taking new customer rev and dividing by 29:51 ad spend. And once again, this can be a 29:53 percentage or an integer. It doesn't 29:55 matter. And so this might be a 4x. Let 29:57 me run you through an example of why 29:58 this is so important. And I see this in 29:59 literally every single audit I do, which 30:01 is why KPIing on the right metric 30:03 matters. So let's say in January you 30:05 were at a 4x me. Then in Feb, it jumps 30:08 to a 5x. Now off the back of this, due 30:11 to it being a bit of a lagging signal, 30:12 the way that you're looking at it, you 30:13 go, "Oh, actually Feb's really 30:15 efficient. Let's spend more money in 30:18 March." And so you go up and spend more 30:20 money in March, and it compresses your 30:22 ME back down, but you're fine because 30:23 this is a target. Now, this all looks 30:25 good. However, it doesn't take into 30:27 consideration what caused this number to 30:29 go up. Was it an improvement in 30:31 efficiency in new or was it just more 30:33 returning customers? So, when you go and 30:35 delineate down to acquisition me, what 30:37 you might find out is that this was a 30:39 two, this was then a 1.8, and then this 30:41 was a 1.6. 30:43 And so, what happened here was in Feb, 30:46 you actually had an increase in 30:47 returning customer revenue that propped 30:50 up this number. New customer revenue 30:52 went down. But because you're using this 30:54 number to make decisions, not this one, 30:57 you go, "Okay, well, let's spend even 30:59 more." And so you go and spend even 31:01 more. It further erodess new customer 31:02 acquisition. This becomes incredibly 31:04 unprofitable for the business, but you 31:06 don't notice it because once again, 31:07 you're indexing against me. And so 31:09 you'll start to make decisions in terms 31:10 of paid media allocation and the 31:12 efficiency of paid against a metric me 31:16 that isn't actually congruent to the 31:18 efficiency of paid. Then a common 31:20 question I get asked is, well, what is a 31:22 good acquisition? M what should we be 31:23 aiming for? And the answer to this is 31:25 the same as the fact that all revenue is 31:28 not equal. It's the same as acquisition 31:30 me is not equal. And so in some 31:32 businesses with a 60% gross margin, you 31:35 can obviously have a way worse 31:37 acquisition me and still be just as 31:39 profitable as someone else who only has 31:41 a 40% gross margin. And so the target 31:44 here is going to be respective towards 31:45 the margin profile as well as the 31:48 retention of the business. If you have 31:49 incredible retention and you have like 31:51 50% lift by the six-month mark, then 31:55 obviously you can have a much more 31:56 aggressive acquisition strategy, make 31:58 less money on first purchase because you 32:00 know these people are going to come back 32:01 and improve the profitability of the 32:03 business. So trying to compare AME 32:05 against different businesses uh becomes 32:08 quite difficult because they all have 32:09 different margin profiles and different 32:10 retention. The next metric that you have 32:12 here, the finance grade metric is going 32:14 to be new customer rev. Just tracking 32:16 this independently. Most people don't do 32:18 this. And then new customer orders as 32:22 well. And so this is going to show us 32:23 how volume is changing over time. And 32:25 then we can look at efficiency 32:27 separately. So this will tell us 32:28 efficiency. This will tell us volume. 32:29 Generally looking at CAC 2, it's similar 32:32 to acquisition me except CAC is going to 32:35 show us what our cost to acquire is. The 32:38 reason why you would measure both of 32:39 these is that acquisition me is actually 32:41 a better measure of profitability 32:42 because it encapsulates average order 32:43 value. CAC doesn't. But if our CAC 32:46 fluctuates quite a bit, it's always good 32:48 to be able to use that as a leading 32:49 indicator to understand why is the 32:52 product portfolio changing. What is 32:53 actually occurring that's causing CAC to 32:55 change irrespective of acquisition me? 32:57 And for those that don't know, CAC is 32:59 your total ad spend divided by the 33:01 amount of new customers acquired in that 33:03 particular time period. Then as some 33:05 final finance grade metrics here, and we 33:07 could go all day on putting together a 33:08 bunch of different finance grade 33:10 metrics. There's literally tens of them. 33:12 But the idea here is not to add, it is 33:14 to subtract. You want the least amount 33:16 of metrics possible to be able to have 33:18 an indication on whether you're on or 33:19 off track and whether to make uh changes 33:21 off the back of it. And so the next one, 33:24 the final one is going to be profit 33:26 contribution. Profit contribution is 33:29 revenue minus cost of delivery minus 33:32 marketing expenses. And this will give 33:34 you a dollar value in contribution 33:36 profit. You can also do this as a 33:37 percentage. you would just min us the 33:39 percentages here. Now, even better than 33:41 this really is to delineate down into 33:44 new customer profit contribution. And 33:46 this is quite a self-explanatory. Rather 33:48 than rev, you just do new customer rev 33:51 minus the cost of delivery associated 33:52 with this revenue minus the marketing 33:54 associated as well. Now, this is going 33:56 to tell us if we're profitable on 33:57 acquisition or not. Really, you should 33:59 have a dashboard that allows you to 34:01 track this number for the CFO. This 34:04 number should be the headline metric at 34:06 the top. These should be then the two 34:08 headline metrics next to this. You then 34:11 have CAC in the dashboard too. And then 34:13 you have your profit contribution 34:14 numbers. If you were tracking these 34:17 numbers here, this is really all you 34:19 need for a dashboard on where the 34:22 business is right now, how profitable it 34:24 is, whether the decisions we made in the 34:25 last week were good or bad, and whether 34:27 we should course correct. They have a 34:28 super high degree of accuracy because 34:31 they are based on the actual financial 34:33 metrics in the business. Now once you 34:34 become a very very big business let's 34:36 say high eight figures into nine figures 34:39 then there becomes a lot more nuance in 34:41 these metrics for example you cannot 34:44 claim AM 34:46 on meta spend let's say we're managing 34:49 meta and there's a bunch of meta-pend if 34:51 the business is also spending a million 34:53 dollars a month on influences because 34:54 the influences are going to drive rev 34:56 and so we can't look at this 4x and go 34:58 oh meta's doing amazing this month well 35:00 we spent a million dollars on influences 35:01 that you're not actually encapsulating 35:02 into this formula And so as more 35:04 channels get introduced and as more 35:06 complexity goes into the business, this 35:08 is where even finance level metrics 35:10 start to fall off in their reliability. 35:12 And that is where we need to go all the 35:13 way up to the top here and start to look 35:15 at lift studies to understand okay how 35:18 much is meta actually driving though as 35:21 a proportion of this new customer rev. 35:23 Now, at smaller revenue numbers, at 35:25 smaller amounts of channels, you can 35:26 just make the assumption that all the 35:28 spend drives, all the revenue, and you 35:30 will have crystal clear decision- making 35:31 and be able to scale very well, and 35:34 you'll remove all the confusion. But as 35:35 you start to get very big, that's where 35:36 the top of the pyramid actually exists 35:39 for. We then go from better metrics to 35:41 the best metric. Now, the reason why I 35:43 call this the best metric isn't 35:44 necessarily because it should be at the 35:46 top of your dashboard. It isn't 35:47 necessarily because this is the only 35:48 thing you should track. It's because it 35:50 is comparable across time and it is 35:53 comparable across businesses. And that 35:56 is what makes it a really good metric 35:57 for being able to identify particularly 35:59 across our entire client portfolio. 36:01 Where's everyone sitting right now? 36:03 Who's doing badly? Who's doing well? 36:05 Who's doing incredible based on just one 36:07 metric? If we used acquisition me, it's 36:09 impossible to index all of our clients 36:11 and understand who's doing well and 36:13 who's not. Because for some businesses, 36:14 a 4E is good. For others, it's terrible 36:17 because of the gross margin profile, 36:18 because of the spend on other platforms, 36:20 like all of these different nuances. But 36:22 LTGP to CAC can be compared across 36:25 anyone. I can ask one of our CPG brands, 36:28 "What is your one-year LTGP to CAC?" And 36:30 then I can go and ask one of our fashion 36:32 brands, "What is your one-year LTGP to 36:34 CAC?" And they're comparable. I can go, 36:35 "Oh, that fashion brand's better." They 36:36 are better on acquiring customers right 36:38 now and they're a more efficient 36:39 business. They probably, as long as 36:40 they're controlling operating expenses, 36:42 have a higher net profit margin. And so 36:44 this becomes a very good metric for 36:46 comparing against yourself historically 36:47 and comparing against others. Now why is 36:50 it so good? Well, it's because anytime 36:52 you have a ratio, it is generally a 36:54 superior metric because you have a 36:57 secondary pairing metric that allows you 36:59 to contextualize the primary. If you 37:01 just look at CAC, well, it has no 37:03 context. If you just look at LTGP, well, 37:05 it has no context. You bring them 37:07 together, the context is provided. Now, 37:09 the important component here is well, 37:10 what actual time value do you look over? 37:13 If you're just looking at yourself, you 37:15 can do whatever time you want. Okay? So, 37:17 we could do 30-day LTP to CAC and that's 37:19 fine. If you start comparing against 37:21 competitors, particularly in other 37:23 industries, like if you compare CPG to 37:25 fashion becomes very different because 37:27 there's very different uh retention 37:29 profiles. And so a CPG brand might 37:31 operate on a very low first purchase GP 37:34 to CAC compared to a fashion brand, but 37:36 the CPG brand might make way more money 37:39 because they have higher repeat rates 37:40 over the following year. So you just 37:42 want to extend this time period as you 37:44 start to compare to other people. But if 37:46 you're just comparing to yourself, 37:47 30-day, 90day, 180day, absolutely fine. 37:50 Now what are the benchmarks here? 37:52 Benchmarks and let's call this 37:53 benchmarks on 180day LTGPAC. If you are 37:56 below a one, this is very bad. This is 37:59 the equivalent of a six-month CAC 38:01 payback period if you're below a one, 38:03 which is not good. Um, unless you have a 38:06 lot of funding, you don't I would never 38:08 want to run a business or probably even 38:10 work with a business with a CAC payback 38:11 period greater than 6 months. Super 38:13 high-risk position to be in. You're 38:15 having to float so much capital to even 38:17 operate this thing. I don't like it at 38:19 all. 2 to three. Well, let's go 1:2. 1:2 38:22 is an okay position to be in. This isn't 38:24 bad. You're making money. you're 38:26 profitable on acquisition, but you're 38:28 not that efficient. And so, either your 38:30 cost to acquire a customer is too high 38:31 or you're not making enough in GP. Two 38:33 to three, this is good. This is where 38:36 you want to be. This is where you want 38:37 to scale. Any brand that's in this range 38:39 for us, we're looking at how we can 38:40 increase spend. 3 plus, honestly, this 38:44 is bad. And it's bad because you're 38:46 leaving money on the table. If you're at 38:48 over a 3 plus LTGP to CAC on a 180day 38:50 basis, you could be scaling way harder, 38:53 capitalizing way more on the current 38:55 acquisition channels that you have in 38:56 place, but you aren't. Now, if this is 38:58 due to a cash flow problem, then that's 39:00 completely understandable and fix the 39:02 cash flow. Don't take out large loans to 39:04 be able to capitalize here. But 39:05 ultimately, this is an unbelievable 39:07 position to be in. Most people aren't in 39:08 this position and therefore you should 39:10 put the pedal down and actually 39:11 capitalize on it. Now, just one 39:13 practical reality for everything we've 39:15 been going through is it's all reliant 39:16 on real-time gross margin numbers. The 39:18 LTGP to CAC means we need visibility in 39:21 gross margin. The profit contribution 39:23 means we need visibility into gross 39:24 margin. And so, what you need is that if 39:27 you're on Shopify, 39:29 make sure that you have cost of goods 39:32 tagged up on all of the products because 39:34 that is going to give you real-time 39:36 gross margin visibility as well as how 39:38 gross margin compresses during discount 39:40 periods. This can then flow through into 39:43 any external software into any Google 39:45 sheet reporting etc. which is going to 39:47 give you an accurate visible view into 39:50 gross margin. Without this you will have 39:52 to make a percentage assumption which 39:54 will be wrong and will substantially 39:56 mislead you. Let me explain why. Let's 39:59 say that you have a $100 product and it 40:03 has a $50 cost of delivery on it. This 40:06 means that you have $50 in gross margin, 40:10 aka 50% gross margin. And because of 40:14 this, this is what you actually tell 40:15 your agency or your internal team. You 40:17 go, "Hey, we've done the math. We've 40:19 looked at our reporting with the 40:21 accountant over the course of the last 3 40:22 months. And our average gross margin is 40:25 50%." Cool. We then go and bake that 40:27 into all of our models. We take your 40:29 revenue, we times by 50%. That's your 40:31 profit contribution. That's your LTP to 40:33 CAC. Everything is using this number. 40:35 However, you go into Black Friday and 40:37 let's say you have a 20% compression in 40:39 pricing. So, you do a 20% off storewide. 40:42 Now, your price goes down to $80. Now, 40:44 yes, average order value might go up. 40:46 So, overall in terms of basket size, you 40:48 get more, but the units is what is going 40:51 to determine gross margin. So, the unit 40:53 price goes down to 80. Your cost of 40:55 delivery remains the same at 50. 40:57 Therefore, your gross margin compresses 40:59 to $30. This means your new gross margin 41:02 is 30 divided by 80 which is 37.5%. 41:08 So your gross margin compressed during 41:11 this sales period by a lot. This is 41:14 12.5%. 41:16 If we do a percentage of the decrease, 41:18 this is like a 30% decrease from 50 to 41:20 37. So, it's really important that you 41:22 have cogs tagged up or else all of the 41:24 metrics that you have within your 41:26 dashboards that you're measuring against 41:27 will be wrong because it won't 41:28 encapsulate margin compression as you 41:30 discount. So, now moving into 41:32 incrementality and experiment design. 41:34 Fundamentally, the problem with 41:36 attribution is that it's correlation. 41:38 And so, what we're doing is someone 41:39 clicks on an ad, then they go and 41:41 purchase, and we're saying that this 41:42 click caused this purchase. But we don't 41:45 have proven causation. There's no 41:47 experimental design that tells us that 41:48 this click actually caused the purchase. 41:50 We are just assuming, okay, someone 41:51 clicked, then someone bought, probably 41:53 had something to do with each other. 41:55 Therefore, we should try to do more of 41:56 this and see if revenue goes up. Now, 41:57 that's absolutely fine and it's pretty 42:00 much causal. If you were a business that 42:02 no one knows about doing 0 in revenue 42:04 and the only marketing effort you had 42:06 was Facebook ads, then yeah, well, if 42:08 someone bought, there's probably no way 42:10 they bought anywhere else than clicking 42:11 on a Facebook ad. So, it's very tightly 42:13 correlated. Although it's still 42:15 correlation, it's good enough to be able 42:17 to assume that it's causal. However, 42:19 where the issue starts to arise is when 42:21 you have multiple different places in 42:23 which you capture revenue and multiple 42:25 different channels in which you 42:27 advertise on. And so rather than 42:29 advertising on just one platform, when 42:31 you introduce a second platform, maybe a 42:34 third platform. So let's call this Meta. 42:36 Let's call this Google. Let's call this 42:39 Tik Tok. And then we can capture revenue 42:41 over here on the.com website. We can 42:45 capture revenue here in the retail 42:47 store. we can capture revenue here on 42:49 Amazon or one of our wholesalers. Then 42:51 correlation becomes very confounded and 42:53 this is obviously the issue with 42:54 attribution, right? Is that we start to 42:56 spend more on Meta and how do we 42:59 actually measure this? How do we measure 43:01 the revenue capture? Because it's going 43:02 to drive revenue here, here, and here. 43:05 And then this is likely going to drive 43:06 more people over here, which is going to 43:08 lead into here. So it starts to become 43:09 very convoluted. And so the gold 43:11 standard here is to run something that's 43:13 called a geo lift experiment. What this 43:17 is is that you take a country and you 43:20 split by states. Now, I'm not going to 43:22 draw a country because my drawing is 43:24 terrible. So, just imagine that this is 43:25 a country and we're splitting up by 43:27 states. What we want to do is we want to 43:29 look for which states closely match the 43:32 other states in terms of multiple 43:35 variables. Now, generally, you will do a 43:38 clustering algorithm using historical ad 43:40 spend. And so, we want to take states 43:42 that have moved together historically in 43:44 the ad spend distribution to them. We 43:46 want new customer revenue to have been 43:48 tightly correlated historically. And 43:51 then ideally, we also want to throw in 43:53 new customer orders because revenue can 43:56 diverge from orders based on average 43:57 order value fluctuations. So we use 43:59 these three variables. We look at all of 44:01 the data in all of these states over 44:04 time and we look at which of these 44:06 states is tightly correlated with 44:08 another one. Now the case usually is 44:10 that there is no state that's tightly 44:12 correlated with another state and so 44:14 instead we start making combinations and 44:16 the combinations have an even better 44:18 effect which is that we can derisk out 44:20 of one particular state and we can use 44:22 multiple and so we might find and let's 44:24 just use the US as an example here that 44:27 we have Arizona let's have Texas and 44:30 then over here let's have New York and 44:32 then California we might find that when 44:34 you add Arizona and Texas's daily ad 44:37 spend together it looks like this. And 44:40 then when you add New York and 44:41 California's together, it looks like 44:43 this. 44:45 So they move in the same way. Then we do 44:48 the same thing for new customer revenue 44:49 and it looks like this. Then we do the 44:50 same thing for new customer orders and 44:52 it looks like this. Amazing. Now we know 44:54 we have two states that are tightly 44:56 correlated and predictive of the other 44:59 state. So at any point in time, we can 45:00 go back and we can look at Arizona and 45:02 Texas. We can go what was the revenue 45:04 here? Cool. We don't even have to look 45:06 at California and New York. we can 45:08 predict it with 99% accuracy. So we're 45:10 looking to build a predictive model that 45:12 tells us if we look at this and we don't 45:14 even look over there, we know what's 45:15 happening. Then that gives us the 45:17 ability to set this experiment up and go 45:20 okay well what we are then going to do 45:22 this is what's called our intervention 45:24 and during this day we're going to go 45:27 and introduce Tik Tok or let's say we're 45:30 going to take our meta cold targeting 45:32 campaign and we're going to double or 45:34 triple the budget. Okay, so we go and 45:37 let's say meta we 2x budgets. 45:41 Then 45:43 we still have our control states that 45:47 aren't impacted. We have our treatment 45:50 states and we want to see what happens. 45:51 And what we might see is that it is 45:53 still tightly correlated 45:56 and then it grows. And so then we can go 45:58 okay well what should it have done using 46:01 the counterfactual. So the 46:02 counterfactual is the control that's 46:04 predicting what the treatment would have 46:06 done and we would have predicted that it 46:09 would have done this. And so now using 46:11 the prediction which is predicted using 46:14 the counterfactual and then using the 46:16 actual output we calculate the 46:19 difference here. So how much additional 46:21 revenue was captured and let's say it's 46:24 $100,000. So there was an additional 46:26 $100,000 above the baseline expectation. 46:30 Let's say on Meta we spent an extra 40K 46:32 in ad spend. Well, then we know that 46:34 this $40,000 in ad spend drove this 46:37 $100,000 lift. Now, it's a 2.5x on the 46:40 dot, which is a 2.5 incremental rorowaz. 46:44 And this is ultimately the crux of a geo 46:47 lift experiment. Now, the reason why 46:49 this becomes so incredibly helpful is 46:51 because you can have multiple different 46:53 channels in here, too. So, we don't just 46:55 need to look at revenue when we're 46:57 looking at this graph. We could look at 46:58 retail revenue. So what is retail 47:00 revenue in this states versus this and 47:02 we can see if there was an impact from 47:03 metas-pend. We can then run this 47:05 experiment and then when it finishes we 47:06 choose different states and we run 47:08 another experiment. So we can continue 47:09 to measure the impact on revenue 47:12 realization. This also becomes critical 47:14 when you go really high up the funnel. 47:16 So if you were going to go and let's say 47:17 do TV or out of home campaigns, this is 47:20 a great way to measure it. You do out of 47:22 home in just these states and you see is 47:24 there a divergence from the control and 47:26 if there is okay well now we can start 47:28 to have a measurable outcome on our 47:31 spend input on a really top ofunnel 47:33 activity. Now there is a caveat here for 47:35 anyone in Australia which is that you 47:37 actually can't do state-by-state design 47:39 and incrementality testing. You can but 47:41 it's it's really not great. You need 47:42 massive spend levels and you can't run 47:44 another one soon after because you muddy 47:46 up the clustering data that's used to do 47:49 state selection. And so if you're going 47:51 to do geolo lift in Australia, you need 47:53 to do it at a commuting zone level, 47:55 which simply means you need to split the 47:57 country up into even further sub regions 47:59 so that you can have enough region 48:01 selection to be able to actually run a 48:03 lift experiment effectively. Now this is 48:05 a geolyft experiment that you do outside 48:07 the platform. You measure it against 48:09 revenue in the business. There's also 48:11 lift experiments in the platform. And so 48:14 you can run a lift experiment in Meta, 48:17 in Google, in Tik Tok. Usually they just 48:19 need relatively large budgets. In Meta 48:21 you can just run them all the time. 48:22 Fortunately in Tik Tok they always give 48:25 you a minimum spend. So you need to 48:26 spend at least 200 grand to be able to 48:28 actually run the test. Same thing with 48:30 Google. Now the difference between a 48:31 lift experiment in the platform and a 48:33 geolyft experiment is that because the 48:35 platforms have each individual user ID 48:39 rather than needing to split at a 48:41 country level to segment the control and 48:43 treatment they can just split at an 48:45 individual user level. So I can go into 48:48 bucket A and then someone else can go 48:51 into bucket B. And then this person will 48:54 see the ads. I won't see the ads. Then 48:57 meta, Tik Tok, Google, they measure, do 48:59 I still buy? They look at the pixel. 49:01 They see does Nathan end up going to the 49:02 website purchasing or he doesn't. But 49:05 these people do. Cool. We can do an 49:06 incrementality calculation. Now the 49:08 disadvantages of Lyft experiments in the 49:10 platform is that number one it relies on 49:14 all the data science and measurement of 49:16 the platform and so we are trusting that 49:19 Meta or Tik Tok or whoever it is is 49:21 giving us a real accurate read out of 49:24 the back of it. There's no way for us to 49:26 test their math. We don't get any 49:27 visibility into the actual raw data set. 49:29 So we can't actually verify whether it 49:31 is what they say it is. Now I'd hope 49:33 that it's legit but we can't verify it. 49:36 And it's important to know that because 49:37 in geolo lift experiments we can verify 49:39 everything because it's all our data and 49:40 we're running the experimental design. 49:42 The second disadvantage of a lift 49:43 experiment in platform is obviously the 49:45 limitations of attribution. And so this 49:48 is relying us on us being able to 49:50 attribute these people purchasing to the 49:52 website and the other people not 49:53 purchasing to the website. And so Lyft 49:56 experiments have disadvantages, pros and 49:58 cons. Pros, they're fast, they're quick 50:00 on meta, you can run them all the time. 50:01 Cons are relies on attribution. We don't 50:03 get any visibility into the data. So it 50:05 could potentially not be reliable. But 50:07 geoloyft experiments too also have cons 50:09 which is that they take a lot of time to 50:11 run. They typically do require a lot of 50:13 spend to be able to reach statistical 50:14 relevancy in the spend increase 50:15 required. Um and it relies on us having 50:18 to control variables at least within the 50:20 uh state selection through the test 50:22 period. And so these aren't easy to run 50:25 and these aren't just some magical 50:26 solution and neither are these. They 50:28 both have pros, they both have cons. You 50:29 should have both within your experiment 50:32 design strategy. One last comment here 50:35 in terms of another experimental design 50:37 that uh probably only the nerds watching 50:40 this video will pick this up and 50:41 actually do anything with it, but that's 50:42 using causal inference. Causal inference 50:44 is an open-source package open source by 50:46 Google about I'm going to say 11 or 12 50:49 years ago. And this is just a predictive 50:51 model for using a counterfactual to 50:53 predict what would have happened at the 50:55 point of intervention. So given the 50:57 example before, imagine you didn't split 50:59 the states. And imagine what you did was 51:01 at some point in time, let's call this 51:03 new customer rev right here, new 51:05 introduced YouTube advertising. And what 51:08 you want to do now retrospectively is 51:10 you want to go, okay, we introduced 51:12 YouTube here. What was the impact on 51:14 revenue? We think it made a difference, 51:15 right? Revenue was slightly lower and 51:17 now it's slightly higher after YouTube 51:19 came in. But was that because of other 51:21 factors? Would that have happened 51:23 anyway? Was this seasonality? How do we 51:25 know that truly YouTube caused this 51:27 increase in revenue? And now without the 51:29 geoloyft experiment you never can. Okay, 51:31 we can never prove causation here. But 51:34 we can try to use a predictive model to 51:37 be able to understand what it would have 51:39 done in terms of revenue using 51:41 third-party confounding variables. So 51:45 your daily revenue, what causes revenue? 51:48 Well, number one might be ad spend. 51:50 Number two, there might be seasonality 51:53 on particular keywords. And so we can 51:56 use seasonality keyword data. Number 51:58 three might be spend towards influences. 52:01 Number four, inflation data might have 52:04 some kind of correlation. Number five, 52:06 and obviously you can go and list any 52:07 kind of variable that you think is 52:09 predictive of your revenue. Then you 52:12 load it into the open- source package 52:15 called causal impact. I'm sorry, I put 52:17 causal inference up here. This is meant 52:18 to be causal impact. And then it will 52:21 give you a predictive outcome of what 52:23 this revenue would be using all of these 52:25 metrics during this time period. And so 52:28 ad spend will be doing something over 52:30 time and it will draw the relationship. 52:31 And then the seasonality of keywords 52:33 will be doing something over time and it 52:35 will draw a relationship and it will 52:36 draw a relationship between all of these 52:38 other variables to be able to predict 52:40 revenue. And so it might come out and go 52:43 using all of these variables and how 52:45 they've moved historically and then how 52:47 they're moving now, your revenue would 52:49 have done this. And now because of that, 52:52 we can go, oh, so YouTube wasn't 52:54 actually as effective as we thought it 52:56 was. And this is the reality of most 52:58 experiments is a lot of people have no 53:00 idea how to set up experimental design 53:02 correctly and so they get false outcomes 53:04 all the time. Which is why this whole 53:06 video exists and why data integrity is 53:07 so important. People introduce YouTube 53:09 or Tik Tok or something and they'll do 53:11 it in like October and then they'll have 53:13 a great November and they'll go, "Yeah, 53:14 YouTube was probably helping." How? Like 53:17 how did you measure that? How do you 53:18 know that YouTube drove an uplift during 53:21 one of the best periods of the year? You 53:22 don't. There's no measurement system. 53:24 And then once you actually start to get 53:25 some kind of uh causal experimental 53:27 design here, you start to find out that 53:29 all of your assumptions are wrong. And 53:32 so the actual impact of YouTube would be 53:34 this area right here. So this is the 53:37 predicted counterfactual, the red dotted 53:39 line. And then this is the actual 53:40 revenue. And so just like before, you 53:42 might find that this additional revenue 53:45 over this time period was let's say like 53:47 $10,000. Ad spend was $10,000. So 53:50 YouTube drives a 1x ROI in the business. 53:53 Now, causal impact obviously isn't 53:55 perfect because this is using just a 53:57 predictive model using thirdparty 53:59 variables. And so, is it helpful 54:01 directionally and does it allow you to 54:03 do some retrospective analysis? For 54:05 sure. And we have it built into our own 54:06 internal software so that we can do 54:08 things like this retrospectively. We can 54:10 go back and we can say, okay, you 54:11 introduced Tik Tok back here. Let's do a 54:13 causal impact analysis. And what do we 54:15 think Tik Tok roughly drove? It gives us 54:17 some direction. It gives us some idea. 54:18 Okay, it seems Tik Tok did nothing. 54:20 Okay, it seems Tik Tok did something. 54:22 Um, it's a way to apply Beijian 54:24 statistics into marketing, which is 54:27 ultimately what we're always trying to 54:28 do with measurement is we're trying to 54:29 apply real data science, statistical 54:32 modeling into any results that we're 54:34 trying to draw. Then on tracking, I want 54:36 to make three quick comments and then 54:37 move on. The reality is that we don't go 54:39 too deep on tracking because brands that 54:41 we work with are typically doing at 54:42 least 10 $20 million a year in revenue. 54:44 And so if tracking is not set up 54:46 properly at $10 to $20 million in 54:47 revenue, you have some big problems. I 54:49 also don't want to go too instructional 54:50 here on here's exactly how you do this 54:52 because you can just YouTube for a 54:53 dedicated video and find out exactly how 54:55 to fix these things. Number one is 54:57 Cappy. You want to make sure that Copy 54:58 is set up on Meta. This is your 55:00 conversions API. How this works is that 55:02 rather than purchase events directly 55:04 firing from the pixel and sending back 55:07 to Meta, the purchase event fires on 55:09 your own local server and then it 55:11 reroutes it back to Meta. The reason why 55:13 that's important is that it removes the 55:15 redundancy on the browser which has 55:16 become super unreliable due to iOS ad 55:19 blockers and privacy regulations. And it 55:21 also enables much better event matching 55:23 because it uses firstparty data. So it 55:26 passes back the user who bought their 55:28 email address, their phone number, their 55:30 customer ID. It obviously hashes all the 55:32 data. So it's all privatized, but that 55:35 way it can actually improve match rates. 55:37 It also allows for offline conversion 55:39 data. If someone buys in store and they 55:41 use an email or a phone number, we could 55:42 link it back to an actual ad being 55:44 served on the platform. And so you want 55:45 to make sure this this is actually 55:47 working, it's going to improve your 55:48 tracking, which is going to improve the 55:49 amount of data flowing into the 55:50 platform, which is going to improve 55:51 optimization on Shopify. You do not want 55:54 to use any marketing or attribution 55:57 reports in Shopify. They are generally 56:00 quite unreliable. This is ultimately 56:02 like the issue with attribution. Uh but 56:04 Shopify in itself, I haven't seen much 56:06 helpful data at all looking at the 56:08 marketing reports in there and the 56:10 attribution models that they're using. 56:11 And then lastly, G4. G4 in my opinion is 56:15 a redundancy layer. And so if everything 56:18 was to go bad and our platform stopped 56:21 tracking or something happened, we could 56:23 lean back on J4 for some quick 56:25 directional insight. We don't use J4 56:28 that much at all for day-to-day 56:30 decision-m because there's nothing in J4 56:32 that we just can't get out of Shopify or 56:34 the platforms. And so at least as it 56:36 stands in May 2026, I don't think G4 56:39 should really be a dashboard that you're 56:41 opening to make decisions within the 56:43 business on. I think there are much more 56:45 important metrics that should be tracked 56:47 to be able to index against the 56:49 performance of the decisions that we're 56:50 making. However, it is a good redundancy 56:52 layer to just have there. So I would 56:54 make sure everything is set up properly 56:55 there. it's firing, you've got your 56:57 segmentation of campaigns, etc., so that 56:59 if you ever do need to use it, if you 57:00 ever do need to fall back, all the data 57:02 has been collecting properly. Now, 57:03 lastly, I want to run through the nine 57:04 mistakes that we see in audits all of 57:06 the time. And we'll fire through this 57:07 relatively quickly. Number one is one 57:09 day view is running across all campaigns 57:12 and is being used in reporting. And so 57:14 there's massive overattribution. 30 to 57:16 50% of rorowaz is just coming from those 57:18 views. And so once we remove it, we get 57:19 a much more realistic view of what is 57:21 actually occurring in the business. 57:23 Number two is the Google halo effect. 57:25 Google's return on ad spend always looks 57:27 way better than it actually is because 57:28 it is taking credit for what is 57:30 occurring on other platforms. Is that to 57:31 say that you cannot acquire new 57:32 customers through Google? No. Is that to 57:34 say you can't go and spend $100 $200,000 57:36 a month on Google profitably that is 57:38 genuinely net incremental? No. Of course 57:40 you can. We have multiple clients that 57:41 we are. But you have to take into 57:43 consideration that Google will 57:44 overattribute. So when you see a 20 57:46 rorowaz on your pax campaign, even if it 57:48 has brand excluded, you just need to be 57:50 skeptical cuz you could just go and test 57:52 put an extra $1,000 into that campaign. 57:54 see if you get 20 grand back in topline, 57:55 you probably won't. And so the 57:57 incrementality of Google as a platform 57:59 is usually overstated. And so you just 58:01 need to be careful of the halo effect to 58:02 be able to see and understand the real 58:04 impact here. Number three is so much 58:06 brand waste due to an over 58:08 prioritization of rorowaz. And so 58:10 because rorowaz is the north star to get 58:11 rorowaz up on campaigns, we just leave 58:13 brand in. You wouldn't believe how many 58:14 large 8 n figure retailers I see this 58:16 on. And it just causes so much waste 58:18 within the account. Number four is over 58:21 segmentation. there's just way too many 58:23 campaigns comparative to either the 58:24 complexity of the business or the ad 58:26 spend level that they're at. Now, this 58:27 compounds into a measurement issue 58:29 because you have all of this data 58:30 fragmented and so decision-m becomes 58:32 very convoluted. Number five is having a 58:36 bunch of tertiary channels like 58:38 Snapchat, Pinterest, Reddit, all of 58:41 these other third party channels which 58:43 have such a low spend allocation and 58:45 have so much over attribution because 58:47 when spend is low, it will always just 58:49 prioritize the warm audience that's 58:50 hitting the website through the pixel 58:52 and so you end up with incredibly high 58:53 return on ad spend numbers on channels 58:55 that just aren't driving any new 58:56 incremental growth to the business. 58:58 Number six is existing customer 59:00 overspend which is just inflating all 59:02 the numbers and they're increasing 59:03 budgets here because rorowaz looks good. 59:05 Number seven is top ofunnel objectives 59:07 running on meta like traffic campaigns 59:09 or reach campaigns etc. Uh rorowaz is 59:12 terrible. There's no rorowaz which is 59:15 funny cuz you're going why would they 59:16 run this if there's no rorowaz but 59:17 they're running this because there's an 59:19 assumption of performance but there's no 59:21 measurement system in place to actually 59:22 measure it. almost I I actually don't 59:24 know anyone I've ever spoken to that has 59:27 run traffic campaigns or is actively 59:29 running traffic campaigns in their Meta 59:31 account that has a way to measure it. 59:33 Like how are we actually measuring the 59:34 impact? Oh well, we started running them 59:36 like a year ago and our agency said 59:38 revenue went up and so we just we drew 59:40 like correlation that traffic campaigns 59:42 equals better revenue. Okay, what was 59:44 the test design like? Oh, we just like 59:45 put a traffic campaign in at 3% of 59:47 budget uh during Black Friday and Black 59:50 Friday was pretty good. Okay, well that 59:51 3% of budget is probably gonna have no 59:53 impact on the business. Number one, no 59:55 measurable impact, particularly in just 59:56 a before and after like causal impact 59:58 trade. But number two, now we're just 1:00:00 wasting 3% of budget on something that 1:00:01 we can't measure. And so if you're going 1:00:03 to run something like a traffic campaign 1:00:04 or a reach campaign, that's fine. But 1:00:06 have a measurement system in place to 1:00:08 actually be able to validate it. 1:00:09 Otherwise, there's no point in running 1:00:10 stuff that you don't know whether it's 1:00:11 working or not. You're just making 1:00:12 assumptions. Number eight is overspend 1:00:14 on DPAs. This is a function of rorowaz 1:00:16 looking really good. And so people go 1:00:18 and spend more and more and more, but 1:00:19 it's not actually incremental because it 1:00:21 sits a bottom of funnel. Number nine is 1:00:23 no actual testing structure in place, 1:00:24 which really plays into number seven 1:00:26 here, which is that if you are going to 1:00:27 run something, you need to have a 1:00:30 testing or an experimental design in 1:00:32 place to be able to validate whether it 1:00:33 actually works or not. Because if you 1:00:34 can't validate if something's working, 1:00:36 it's not even worth doing. Like what is 1:00:38 the point if we don't know if it 1:00:39 actually is driving any kind of outcome 1:00:40 to the business? So what actual action 1:00:42 items should you take right now 1:00:44 dependent on your revenue level? 1 to 5 1:00:46 million. You should switch over to 7-day 1:00:48 click reporting on Meta. It's going to 1:00:50 much better correlate to acquisition me 1:00:52 in the business. You should tag up 1:00:53 Shopify cost of goods on all your 1:00:55 products so you can get real-time 1:00:56 visibility into gross margin and how 1:00:58 that's going to impact the next metrics 1:01:00 which is have a dashboard set up that 1:01:02 gives you a daily read on acquisition me 1:01:04 profit contribution, new customer profit 1:01:06 contribution, new customer revenue, new 1:01:09 customer orders, and blended CAC. And 1:01:11 then lastly, make sure that you're 1:01:12 excluding existing customers for 1:01:14 topfunnel campaigns and that you're 1:01:15 minimizing brand search spend at 5 to 20 1:01:18 mil. You want all of this, but you want 1:01:19 to start layering in lift studies within 1:01:21 the actual platforms, top offunnel 1:01:23 experimentation that's being measured 1:01:25 against those lift studies and using 1:01:26 incremental attribution for a second 1:01:28 validation layer beyond 7-day click. 1:01:31 Then once you start to go to $20 million 1:01:32 plus, you want to be running geolo lift 1:01:34 tests outside of the platform. You'll 1:01:36 have complexity now in the channel mix 1:01:38 as well as the revenue capture mix here. 1:01:40 And so you want to be validating, you 1:01:42 want a dashboard that ideally pulls in 1:01:44 those omni channel metrics from any 1:01:45 other revenue capture place. And you can 1:01:48 actually go and add one or two 1:01:50 additional metrics on top of this 1:01:51 baseline dashboard once you get to this 1:01:53 scale. And then you want incrementality 1:01:55 factors in here, which is actually the 1:01:57 metrics that I would recommend adding. 1:01:59 So when you run a geoloyft experiment 1:02:01 and you get an incremental return 1:02:03 output, you use this as a factor that's 1:02:06 applied to the inplatform attributed row 1:02:08 numbers and then that will give you a 1:02:10 more realistic view of how incremental 1:02:12 the rorowaz numbers are across all your 1:02:14 campaigns in all your platforms. The 1:02:15 only way to do this though is you need 1:02:16 to run consistent geolyft tests a lot in 1:02:18 a very high frequency so that you can 1:02:21 get incrementality factors across all 1:02:22 the channels and all the different 1:02:23 campaign types that you can then layer 1:02:25 into your reporting. Most brands that 1:02:26 fail to scale paid media don't usually 1:02:28 actually have a paid media problem. They 1:02:30 have a measurement problem that impacts 1:02:32 their decision-making loops. They're 1:02:34 optimizing to numbers that don't even 1:02:36 match what is actually occurring within 1:02:37 the business. And the numbers that they 1:02:39 are reading to be able to understand 1:02:40 what is doing well and what isn't is 1:02:42 actually misconstrued or misunderstood 1:02:45 as to where it sits in the funnel and 1:02:46 how the attribution model is actually 1:02:48 working. The brands that are winning in 1:02:50 2026 and 2027 are the ones that 1:02:52 understand all of this and are moving 1:02:54 towards financial based KPIs, 1:02:56 incrementality tests at the larger 1:02:57 levels of revenue or just cleaning up 1:02:59 reporting in platform if that's what's 1:03:01 required. If you're a performance 1:03:03 marketer and you've gotten this far, 1:03:04 please reach out to us at 1:03:05 hiringbluense.com.au. 1:03:07 We are always hiring for for A+ 1:03:09 performers within performance marketing. 1:03:11 And if you're a brand that's gotten this 1:03:13 far, feel free to reach out to us for a 1:03:14 free audit as long as you're doing at 1:03:15 least $5 million a year in revenue. Uh 1:03:18 we will put together an audit for you 1:03:19 that will be structured through 1:03:21 technical account structure, creative 1:03:22 and data integrity and assurance. So 1:03:24 everything here will be able to 1:03:25 translate into the actual data that you 1:03:27 have. We'll be able to show you use 1:03:28 cases of where you might have been 1:03:29 optimizing for me, but Ame would have 1:03:31 been better. And uh we'll give you the 1:03:33 dashboard and reporting that's going to 1:03:35 facilitate this whole thing. And if 1:03:36 you're not a PM or you're not a brand, 1:03:38 subscribe. Most e-commerce brands are 1:03:39 looking at the wrong number. They look 1:03:41 at rorowaz, they look at revenue, they 1:03:42 look at me, but very few of them 1:03:44 actually understand how cash moves 1:03:46 inside an e-commerce business, what 1:03:48 financial models to use, and how all of 1:03:50 this ties directly into paid media. And 1:03:52 that gap between what the ad account 1:03:54 says and what the bank account says is 1:03:56 where most brands get into trouble. 1:03:58 Today, I'm going to walk you through 1:03:59 every financial concept that you need to 1:04:01 understand. This isn't theory. This 1:04:03 isn't MBA stuff. This is core 1:04:05 application that we use every single day 1:04:08 with the eight and nine figure brands 1:04:09 that we work with. Finance knowledge 1:04:11 ends up being the bottleneck for a lot 1:04:13 of agencies and a lot of founders 1:04:16 because the ad account isn't the 1:04:17 business. Platform metrics end up being 1:04:19 a proxy for performance. Most teams end 1:04:22 up misdiagnosing their problem because 1:04:24 they think they have an ads problem or a 1:04:26 media buying problem, but a lot of the 1:04:27 time it's a measurement issue or it's a 1:04:29 margin issue or it's a cash flow 1:04:30 problem. Good paid media strategy should 1:04:33 always reconcile directly to the P&L. If 1:04:35 the ad account looks good, but the 1:04:37 business isn't improving, something is 1:04:39 fundamentally broken. Better data should 1:04:41 lead to better decisions and a lot of 1:04:43 reporting done internally in large 9-f 1:04:45 figureure businesses to seues as well as 1:04:47 small six to seven figure businesses 1:04:49 between an agency partnership. A lot of 1:04:51 it is vanity. So, we're going to strip 1:04:52 it away in this video. Firstly, we're 1:04:54 going to cover the P&L. We're going to 1:04:55 be going end to end on every component 1:04:57 of the P&L and how it translates into 1:05:00 direct to consumer e-commerce and 1:05:02 changes the decisions that you make on a 1:05:03 day-to-day basis. Then we're going to go 1:05:05 into unit economics. We're going to 1:05:06 break down everything in unit economics 1:05:08 that is important for day-to-day 1:05:10 decision-making as it ties into paid 1:05:12 media and the business as a whole. We're 1:05:13 then going to be touching on metrics. 1:05:15 This is the metrics that actually 1:05:16 matter. I'm not going to be giving you 1:05:18 20 different metrics that aren't 1:05:19 actually going to move the needle or 1:05:20 change your decisions. You want metrics 1:05:22 that fundamentally change your behavior 1:05:24 after you read them. We then have cash 1:05:26 flow verse profit. This is something 1:05:28 that most smaller founders don't 1:05:30 understand well enough. I guarantee 1:05:32 every agency that's either watching this 1:05:34 or an eight or nine figure marketing 1:05:36 manager that's working with an agency. 1:05:38 The agency doesn't understand how to 1:05:40 properly integrate cash flow versus 1:05:42 profit decision-making into the ad 1:05:44 strategy. And so we're going to be 1:05:45 breaking down exactly how you need to be 1:05:46 doing this, where the complexity of cash 1:05:48 flow gets introduced into an e-commerce 1:05:50 business versus any other type of client 1:05:52 that you're working with across paid 1:05:54 ads. And then lastly, we're going to go 1:05:55 and put it all together. We're going to 1:05:56 loop the P&L into unit economics, into 1:05:58 metrics, into cash flow versus profit so 1:06:00 that you can make better decisions and 1:06:02 you can know everything that you need 1:06:04 within finance for ecom. Starting off 1:06:06 with the P&L, we need to start at the 1:06:08 very top of the P&L, which is revenue. 1:06:10 Now, there's two really important 1:06:11 components to understand here with 1:06:13 revenue. Number one, there is a 1:06:14 difference between gross revenue and net 1:06:16 revenue. And the differences in 1:06:17 decision-making are enormous between 1:06:19 them. So, we're going to break that 1:06:20 down. Number two is a little bit more 1:06:21 simple. A lot of people know this, but 1:06:23 people don't think through it because 1:06:24 they don't run a 10, 20, $30 million 1:06:26 business, which is that revenue is not 1:06:28 profit. And so, you can have a $10 1:06:29 million business that's running at 5% 1:06:33 net profit and you'll be making 500k 1:06:37 per year in profit or income. Now, 1:06:40 technically, this won't actually flow 1:06:42 through to income in an e-commerce 1:06:43 business cuz net profit does not equal 1:06:45 cash flow. Or you could have a $5 1:06:47 million business that's at a 20% net 1:06:51 margin, which is healthy in ecom. And 1:06:53 that means that you're making $1 million 1:06:55 in profit. I would much rather every 1:06:58 single day of the week own this business 1:07:00 over this one. And the reason being is 1:07:02 that a $10 million business is a much 1:07:04 more stressful asset to hold than a $5 1:07:06 million business. And this is doing 1:07:07 double the profit. I think in agency 1:07:09 land and in revenue land, a lot of 1:07:11 people overprioritize just arbitrary 1:07:13 revenue growth with the idea that margin 1:07:15 will expand. For a lot of early stage 1:07:17 founders, margin actually never ends up 1:07:18 expanding and they just get themselves 1:07:20 into a bigger and bigger and bigger 1:07:21 business with the exact same amount of 1:07:23 profit that they were making 3 to four 1:07:24 years ago. And let me quickly expand on 1:07:26 that statement. What do I mean by people 1:07:28 scale to expand margin, but it doesn't 1:07:31 actually work? Well, the logic when you 1:07:32 scale an e-commerce brand is that you 1:07:34 have three buckets of expenses. You have 1:07:37 cost of delivery. You then have 1:07:39 marketing. And we're going to break all 1:07:40 of this down and I'll show you exactly 1:07:41 what's in each of them. And then you 1:07:43 have operating expenses. Now, operating 1:07:45 expenses should stay relatively fixed in 1:07:48 an e-commerce brand because you don't 1:07:49 need to expand people that heavily. So, 1:07:52 as you scale up, what should happen is 1:07:54 operating expenses should remain stable. 1:07:57 So, the blue line here will be opex. 1:07:59 Marketing will go up as revenue expands. 1:08:02 Cost of delivery will go up as well. And 1:08:04 then here's revenue at the top here. So 1:08:07 when we then calculate out profit, 1:08:09 what's actually happening to profit? 1:08:11 It's going up over time. It's actually 1:08:13 expanding as a percentage on the P&L 1:08:16 because operating expenses are remaining 1:08:18 flat. Now the issue is is that for 1:08:20 pretty much every small e-commerce 1:08:22 founder that I've worked with, and for 1:08:24 context, in the first four years of the 1:08:25 agency, we worked with nearly 300 7 1:08:28 figureure e-commerce businesses. So I 1:08:29 have a lot of experience working across 1:08:31 that size of business. And what almost 1:08:33 always happens is that seven figure 1:08:35 founders are just not good capital 1:08:36 allocators. And so operating expenses 1:08:38 actually ends up going up faster than me 1:08:41 and cost of delivery and therefore 1:08:43 profit remains flat. It remains stable 1:08:45 as a dollar figure which is obviously 1:08:47 not a position that you want to be in. 1:08:48 Now let's get back to revenue and 1:08:49 breaking down the P&L. When it comes to 1:08:51 revenue, you have net revenue and you 1:08:53 also have gross revenue. Now the 1:08:56 difference here is that net revenue 1:08:58 accounts for returns, refunds, 1:09:00 chargebacks, sometimes a discount 1:09:02 allowance as well, whereas gross is 1:09:04 simply their cash collected on topline. 1:09:06 Now you actually see this within 1:09:08 Shopify, but in other platforms you 1:09:10 won't. It's also important to note that 1:09:12 the Shopify gross to net breakdown 1:09:15 typically doesn't flow through into 1:09:17 accounting softwares like Zero, like 1:09:20 QuickBooks. And so you will only get a 1:09:22 net or a cash read within a P&L 1:09:25 accounting software. You won't get a 1:09:27 gross read. And this becomes really 1:09:29 critical for using the P&L to actually 1:09:31 make decisions. So when we look at gross 1:09:35 to net, this is really the data that you 1:09:37 want to be looking at. And the reason 1:09:38 being is that if you just have net 1:09:40 revenue at the bottom here and you're 1:09:41 looking at this within your accounting 1:09:42 software, let's say you have 100K one 1:09:45 month and then you have 110K the next 1:09:48 month. When you look at these two 1:09:49 revenue numbers, there's not really much 1:09:51 that you can substantiate out of it. You 1:09:53 can go, okay, revenue increased. We 1:09:55 don't really know why. We need now need 1:09:57 to look elsewhere in the P&L. Let's go 1:09:58 and look at marketing. Did marketing 1:09:59 expenses increase? Let's go and look at 1:10:01 operating expenses. Did we add another 1:10:03 staff member that's driving some kind of 1:10:04 revenue in the business? We have to 1:10:06 start looking elsewhere. But the answer 1:10:07 could actually be above this net revenue 1:10:10 number. If we go and look above net 1:10:12 revenue, what we might actually see is 1:10:14 that gross revenue across these two 1:10:16 months is the same. Maybe it's 120K here 1:10:19 and 120K here. But when we then go down 1:10:22 to refunds, refunds in this month were 1:10:24 at 20K, but refunds in this month were 1:10:27 only at 10. Then we go down to 1:10:29 discounts. There was zero change in 1:10:31 discounts and maybe shipping charges. 1:10:33 All shipping is free on this store. And 1:10:34 so then when we look at the difference 1:10:36 in revenue for this month, the 1:10:37 difference in revenue is actually coming 1:10:39 from refunds, which is a line item that 1:10:41 doesn't get encapsulated in most 1:10:42 accounting softwares if you're not 1:10:44 actually pulling through the gross to 1:10:46 net revenue dynamic into the software. 1:10:48 So really, really important that anytime 1:10:49 you're looking at revenue, we're not 1:10:51 just looking at net, but we're looking 1:10:52 at gross through to net to understand 1:10:54 these different levers cuz I could give 1:10:55 a hundred different examples of this. 1:10:57 Okay, we could change the discount line 1:10:58 item. So actually refunds were stable 1:11:01 across both months but the core 1:11:03 difference was that there was actually 1:11:04 10k in discounts in the prior month and 1:11:06 so because we had some kind of discount 1:11:08 running or discount code that didn't get 1:11:10 encapsulated in the following month and 1:11:12 therefore that's why we saw revenue 1:11:13 expansion. These are fundamentally 1:11:15 levers that exist within the business to 1:11:19 increase profitability, increase net 1:11:20 revenue. If we can decrease refunds, we 1:11:23 make more money. If we can decrease 1:11:25 discounts, we make more money. If we can 1:11:27 increase the amount of shipping 1:11:28 collected at checkout, we also make more 1:11:30 money. And so you want to be looking at 1:11:32 all three of these levers because often 1:11:34 there actually is profit to be unlocked 1:11:36 in better optimizing these three 1:11:39 numbers. If you can push refunds down a 1:11:40 little bit, if you can push discounts 1:11:42 down a little bit, if you can collect a 1:11:43 little bit more shipping at checkout, 1:11:45 which most people can do, most people 1:11:46 underolct at uh checkout, you can 1:11:49 improve the P&L. Next layer of the P&L 1:11:51 is cost of goods sold. This is where 1:11:53 people make a massive mistake in direct 1:11:55 to consumer ecom, which is that they 1:11:56 don't encapsulate all the actual 1:11:58 expenses associated with cost of goods 1:11:59 sold. Cost of goods sold isn't just the 1:12:02 cost that you paid to the manufacturer 1:12:04 for the item, but it's the total landed 1:12:06 cost of getting that product from the 1:12:08 manufacturer to yourself, then into the 1:12:10 customer's hands. And so we have the 1:12:12 product cost. We then have importing 1:12:14 taxes and duties. We then have the 1:12:16 landed cost. So this encapsulates the 1:12:18 freight to get it to your warehouse. And 1:12:19 then sometimes included in here, 1:12:21 sometimes categorized into shipping and 1:12:23 fulfillment instead is the cost of 1:12:26 shipping to the actual customer. So this 1:12:27 would be shipping charges. And this will 1:12:29 also encapsulate uh import taxes and 1:12:32 duties if you're shipping 1:12:33 internationally. Tariffs in as well as 1:12:35 an example. So sometimes you'll go and 1:12:36 encapsulate that here and just count it 1:12:38 in cogs. Sometimes you'll separate the 1:12:39 definition to shipping and fulfillment. 1:12:41 But overall both of these sit under a 1:12:44 category called cost of delivery. And so 1:12:47 whenever COOD is used or cost of 1:12:49 delivery, it is encapsulating all of the 1:12:52 cost of goods expenses and then all of 1:12:54 the shipping and fulfillment expenses. 1:12:55 And this is where people really mess up 1:12:57 product margin versus gross margin. This 1:12:59 definition is super misunderstood. When 1:13:02 you say these numbers to people, you'll 1:13:04 always get different answers. So let me 1:13:06 run through an actual worked example. 1:13:07 Let's say that you're selling a $100 1:13:10 t-shirt. What we will typically get from 1:13:12 a brand is they'll say, "Yep, on this 1:13:13 t-shirt or on this category, we have 70% 1:13:16 gross margin." But what they're 1:13:18 typically doing here is they're going to 1:13:20 someone internal within the wider team 1:13:22 if it's a 8 n figure retailer or if it's 1:13:24 a smaller business, they're just going 1:13:25 and looking at their PO and they're 1:13:27 going, "What did we actually pay for 1:13:28 this t-shirt to the manufacturer?" And 1:13:30 we paid $30 per unit. And therefore, 1:13:33 they're going, "Well, 100us 30 is 70. 70 1:13:36 divided by 100 equals 70%." Now, this is 1:13:39 product margin. This is not gross 1:13:41 margin. And why this is so critical is 1:13:43 because this GM number will be used in 1:13:45 setting KPI. If we know what the gross 1:13:47 margin is, we therefore know what 1:13:49 percentage allocation we should be 1:13:51 putting towards marketing to be able to 1:13:52 ensure that we sell through this 1:13:53 product. Now, with 70% gross margin, we 1:13:56 would honestly probably be fine spending 1:13:57 up to 20, even 25% on marketing to drive 1:14:00 sales here. However, if this is actually 1:14:02 a much lower number and we don't know 1:14:04 about it cuz we're getting given product 1:14:06 margin rather than gross, well, all of a 1:14:08 sudden we're driving this product at an 1:14:10 unprofitable marketing efficiency. So, 1:14:13 we then actually find out that it's $11 1:14:15 on average to ship this to a customer. 1:14:17 It's then $4 in pick and pack fees at 1:14:20 the warehouse. There's a 3% transaction 1:14:22 fee here, which is actually something I 1:14:24 missed in the cost of delivery expenses 1:14:26 before, which is transaction fees. They 1:14:28 always need to be accounted for in cost 1:14:30 of delivery because it is a cost to 1:14:32 fulfill the product. You have to pay 1:14:33 that transaction fee. A lot of people 1:14:35 put transaction fees in operating 1:14:36 expenses which is a mistake because the 1:14:38 expense is variable with revenue. And so 1:14:40 we have $3 on transaction fees. We 1:14:43 actually also have tax implications 1:14:45 here. So of this $100, we were actually 1:14:47 including tax which was $9, which is a 1:14:49 flowthrough expense. So we need to 1:14:51 remove that. We then have the actual 1:14:53 cost of landing this product to our 1:14:55 warehouse. So yes, it was $30 a unit, 1:14:57 but the actual PO cost a few thousand to 1:15:00 get to us and ship via air freight. And 1:15:02 so if we then cut that down into a unit 1:15:04 by unit cost, that's about an additional 1:15:06 $4 per unit. And then the last thing 1:15:08 here that we still haven't taken into 1:15:09 the picture is an assumed refund or 1:15:12 discount allowance. And this is 1:15:13 absolutely critical in the fashion niche 1:15:16 because in fashion refund rates can 1:15:18 range from 10% in some cases I've seen 1:15:20 up to 80%. And so gross margin 1:15:22 substantially changes when 80% of the 1:15:24 product is getting returned. And so we 1:15:26 also need to factor in a refund or 1:15:29 chargeback allowance. And then let's 1:15:31 also throw in a discount allowance too 1:15:33 because people might be using uh coupon 1:15:35 codes here at some particular rate. And 1:15:37 so let's have $15 of allowance here 1:15:40 because this allows for about $5 in 1:15:42 discounts to be used on average as well 1:15:44 as about a 10% refund rate which is 1:15:46 pretty standard within fashion. Now we 1:15:48 go and add all these expenses up and 1:15:50 we've got $46 in additional expenses to 1:15:54 actually deliver on this product. That 1:15:57 comes out to $76 1:15:59 in cost of delivery. So if we 1:16:02 recalculate our gross margin here, which 1:16:04 you do by taking price, you minus the 1:16:07 costs and then you divide by the price, 1:16:10 this equals 24%. 1:16:12 Now, this is an incredibly extreme 1:16:14 example where all of the additional 1:16:16 variable costs here have significantly 1:16:18 added out outweighed the unit cost. But 1:16:19 you can see how there can be such a 1:16:21 drastic difference between the assumed 1:16:24 gross margin, which is product margin, 1:16:25 and the actual gross margin within 1:16:28 e-commerce. And it's because in 1:16:29 e-commerce, you have all of these 1:16:30 associated additional variable cost that 1:16:32 aren't there necessarily in brick and 1:16:34 mortar retail. So, as we move down the 1:16:36 P&L design here, we've gone through 1:16:38 revenue, we've gone through cost of 1:16:39 delivery, gross margin. Now, we're at 1:16:42 marketing. Marketing is relatively 1:16:43 straightforward, which is anything 1:16:45 associated with marketing and 1:16:47 advertising the product falls into this 1:16:49 bucket right here. So, this is anything 1:16:51 on paid ads. This is any influencer 1:16:54 payments, this is any events, this is 1:16:57 any out of home. The big question here 1:16:59 becomes does content production go into 1:17:02 this bucket? And this is based on 1:17:04 whether it is a variable expense or 1:17:06 whether it is fixed within the business 1:17:08 to meet the content demand. And so what 1:17:10 typically happens through an e-commerce 1:17:12 direct to consumer business is that 1:17:14 content velocity or volume needs to flex 1:17:17 throughout the year in conjunction with 1:17:20 revenue expectations. And so a typical 1:17:22 e-commerce brand will look flat through 1:17:24 Q1. We'll have a spike at June for end 1:17:27 of financial year sales. Will then build 1:17:28 back up, have a big November, December, 1:17:30 and then fall back off. And so your 1:17:32 creative production or amount of content 1:17:34 entering into the ad account as well as 1:17:36 obviously investment in influences, 1:17:38 events, etc. needs to map to this. When 1:17:40 you try to map creative to this flexing 1:17:42 in demand, it becomes very difficult to 1:17:44 do with just an in-house team because if 1:17:46 you have four people, they're probably 1:17:47 underworked here. They're probably 1:17:49 substantially overworked here. And then 1:17:51 this is probably the only time a year in 1:17:52 which they have the right amount of 1:17:53 work. And so because of that, it's not 1:17:55 really an ideal model. So, what you do 1:17:56 instead is you have in-house here and 1:17:58 then all of this additional flex up in 1:18:01 creative volume requirements is 1:18:03 typically done by agencies or you could 1:18:05 pull in short-term hires that are just 1:18:08 there to fulfill a 3-month period. Now, 1:18:10 these agency expenses that flex, this is 1:18:12 a variable expense that should go into 1:18:15 your ME. These fixed costs down here, 1:18:17 this should be associated with OPEX. 1:18:19 This is a fixed expense that needs to be 1:18:21 held for indefinitely into the future 1:18:24 while content is king for being able to 1:18:26 drive revenue. Now, as you then move 1:18:28 into flexing this throughout the year, 1:18:29 that's when you can make a variable. 1:18:31 Now, some people would also argue that, 1:18:32 hey, let's just put all of this in the 1:18:33 me bucket because all of this is revenue 1:18:35 driving, which I also understand, but 1:18:38 that's going to be where there's a 1:18:39 little bit of custom design within how 1:18:42 you're approaching the P&L at an 1:18:44 individual brand level. Once we now go 1:18:46 past ME, which is the second big expense 1:18:48 bucket within an ecom brand, the biggest 1:18:50 expense is typically cost of delivery. 1:18:52 Second biggest expense is me. We now get 1:18:54 to another margin, which is contribution 1:18:57 margin. This is one of the best metrics 1:18:59 for indexing performance over time. And 1:19:01 I'll tell you why in a second. Now, 1:19:02 contribution margin is relatively 1:19:04 straightforward looking at this P&L 1:19:06 design. It is total revenue minus cost 1:19:08 of delivery minus all marketing 1:19:11 expenses. And this will give you your 1:19:12 contribution margin. Now, this can all 1:19:14 be calculated as both a percentage and a 1:19:17 dollar figure. So, revenue is 100%, you 1:19:20 have cost of delivery at 30%, me at 20%, 1:19:22 therefore your contribution margin is 1:19:24 50%. Now, why is this contribution 1:19:26 margin number so helpful? Well, it's 1:19:28 because if we're looking at it as a 1:19:29 percentage, we are one step away from 1:19:32 net profit. To get to net profit, all we 1:19:34 need to do over here is minus off OPEX. 1:19:36 So if we know that opex in the business 1:19:38 is kept relatively stable at let's say 1:19:41 10%. Well then that means that 50% minus 1:19:44 10% is 40% net. Now we will have a net 1:19:46 target as a business. You might want to 1:19:48 be holding 10% net margin or 20% or 30% 1:19:50 net margin. You can then back math that 1:19:53 into a contribution margin target. 1:19:55 Right? So let me give you an example. 1:19:57 Let's say you have a 20% net profit 1:19:59 target within the business. Great. If 1:20:01 you know your operating expenses will 1:20:03 always sit at 15%. We then just need to 1:20:06 work our way up the P&L. So over here 1:20:08 we're starting at the bottom and we're 1:20:10 working our way up and we can go okay 1:20:11 net plus opex means our contribution 1:20:15 margin needs to hit 35%. So we 1:20:17 effectively have a contribution margin 1:20:19 target that's really far up the P&L that 1:20:21 ensures that we hit a net profit target. 1:20:22 Same thing with dollar values too, 1:20:24 right? So we might have a net profit 1:20:26 target which is a million. We might have 1:20:28 a million dollar in operating expenses. 1:20:30 And so our contribution margin target is 1:20:33 these two added together, which is $2 1:20:36 million. And so we know for the year we 1:20:38 need to produce $2 million in 1:20:39 contribution margin to be able to get a 1:20:41 million in net profit. Now, one further 1:20:43 piece of clarification here is that 1:20:45 you'll actually hear contribution margin 1:20:47 being used, not in this definition. So 1:20:49 technically, this definition is what's 1:20:51 called contribution margin 3. 1:20:53 Contribution margin one is product 1:20:56 margin, which is what we talked about at 1:20:57 the start. Now why people call this 1:20:59 contribution margin one I have no idea. 1:21:01 People are just making it complicated. 1:21:02 Okay you can just call this product 1:21:04 margin but sometimes people call this 1:21:06 CM1. CM2 is gross margin. So we also 1:21:09 went through that at the start. This is 1:21:10 including all variable expenses 1:21:12 associated with cost of delivery. And 1:21:14 then CM3 is exactly what we just went 1:21:16 through here. This is typically called 1:21:18 contribution margin. Just really 1:21:19 important delineation because sometimes 1:21:21 people will say contribution margin and 1:21:23 they'll be referencing one of these 1:21:24 other definitions which is a weird thing 1:21:25 to do. I don't know why people do it, 1:21:27 but just make sure you understand those 1:21:29 definitions. The third largest expense 1:21:32 bucket is operating expenses. Now, how 1:21:34 to think through what actually sits in 1:21:36 operating expenses is very simple. It's 1:21:37 just anything that doesn't go in cost of 1:21:39 delivery or marketing. Now, the typical 1:21:41 three big expense buckets here within 1:21:44 direct to consumer is going to be people 1:21:46 at number one. It's going to be software 1:21:48 at number two, and then it's going to be 1:21:50 some kind of office or fulfillment 1:21:52 center here. Now it's important to 1:21:54 delineate that actual product 1:21:56 fulfillment like a 3PL or a warehouse 1:21:59 the objective is to fulfill actually 1:22:00 sits in cost of delivery. Same thing 1:22:03 technically with warehouse staff. So if 1:22:05 you have staff that are pickp packing 1:22:07 and shipping product, you actually 1:22:09 technically want that defined in cost of 1:22:11 delivery. What you do is you take their 1:22:13 salary and let's say take their salary 1:22:14 per day. So maybe it's like $400 per day 1:22:17 and then you divide by how many packages 1:22:19 they ship per day. Let's say they manage 1:22:21 100. Well, then you would take a $4 per 1:22:23 order expense and you would move that 1:22:25 over to your cost of delivery in your 1:22:27 actual accounting software. You can tag 1:22:29 particular staff and your bookkeeper can 1:22:31 do this and then they can reconcile into 1:22:33 a custom P&L report that will have 1:22:35 warehousing staff in your cost of 1:22:37 delivery. So, you'll have an accurate 1:22:39 gross margin number on a month-by-month 1:22:41 basis that encapsulates people that are 1:22:43 actually fulfilling the orders. And then 1:22:44 who you would want in operating expenses 1:22:46 for people is everyone that's not 1:22:48 associated directly with the actual 1:22:50 fulfillment of orders. So this would be 1:22:52 like your head of marketing, your head 1:22:54 of operations. List goes on. Now the 1:22:56 real key to operating expenses and where 1:22:58 most seven figureure e-commerce brands 1:23:00 actually get this very very wrong is 1:23:01 that they overinflate operating expenses 1:23:04 believing that this is the way to grow 1:23:05 the business because you commonly hear 1:23:06 reinvest into the business. That's how 1:23:08 you grow. But in e-commerce the way that 1:23:10 you reinvest into the business is 1:23:12 actually to invest in marketing. This is 1:23:14 the primary growth lever that exists 1:23:16 within the P&L for a business like this. 1:23:18 Going and simply hiring more people 1:23:20 generally isn't a revenue generating 1:23:22 exercise. Going and simply getting more 1:23:24 software generally isn't a revenue 1:23:26 driving exercise. Getting a bigger 1:23:28 office generally isn't a revenue driving 1:23:30 exercise. None of these expenses 1:23:32 generally drive much revenue. Now people 1:23:34 are by far the greatest leverage that 1:23:36 exists in any business. In fact, the 1:23:38 ceiling of revenue within a business is 1:23:39 generally the ceiling of the summation 1:23:41 of skills and talent within the people 1:23:43 that exist within that business. So, I'm 1:23:45 very much so of the opinion that there 1:23:48 is nothing more important on this entire 1:23:50 P&L than the people line item. However, 1:23:54 seven figure e-commerce brands generally 1:23:56 don't have the skill set to hire 1:23:58 incredible people yet. Number one, they 1:24:00 don't have a big enough business to 1:24:01 attract incredible talent. Number two, 1:24:03 they simply haven't hired, trained, and 1:24:04 gone through enough interviews to be 1:24:06 able to identify what those people look 1:24:07 like. And I say that from personal 1:24:09 experience. I've personally hired over 1:24:11 60 people in the last 5 years. And at 1:24:13 the start, I had no idea what to look 1:24:14 for. I was just going through interviews 1:24:16 trying to figure it out. And over time, 1:24:17 as you do more and more interviews, and 1:24:18 to date, we've probably done near 500 to 1:24:20 a,000 interviews. We have a pretty good 1:24:22 idea of what a really good high 1:24:23 performer looks like verse not. But when 1:24:25 you're getting started, you don't. And 1:24:27 so people overinflate their people 1:24:29 expense. They overinflate software 1:24:30 because they think this is going to 1:24:31 drive revenue. They end up overinflating 1:24:33 office and tertiary expenses and it ends 1:24:35 up destroying the P&L. So this is just a 1:24:37 really important one to watch because 1:24:39 honestly this is the in a larger 1:24:41 business this is the CFO. In a smaller 1:24:43 business this is the owner's 1:24:45 responsibility to be able to keep under 1:24:47 control. So we can now run through the 1:24:49 actual waterfall of the P&L here. So 1:24:51 starting at the top and I'm going to use 1:24:53 small numbers to keep this simple. we 1:24:55 have $100 in revenue and let's just 1:24:57 think about this as a daily revenue on a 1:24:59 tiny business so it makes sense. Then we 1:25:01 move down to cost of delivery which is 1:25:04 $40 and so this business ends up with a 1:25:07 $60 gross margin. Then the actual 1:25:10 marketing expense here is 20% of 1:25:12 revenue. They maintain a 20% me which 1:25:14 means that contribution margin is now 1:25:17 $40. Operating expenses these guys are 1:25:19 keeping it at 20% as well which is $20. 1:25:22 meaning all the way down the bottom 1:25:23 here, we get to $20 in what we can 1:25:27 define as net profit. Now, what you'll 1:25:29 notice here is that I'm calling it net 1:25:31 profit here, but over here it's called 1:25:32 Ibida. The reason for that is that 1:25:34 technically this $20 here isn't actually 1:25:37 net profit. Net profit sits after we 1:25:40 take out interest, tax, depreciation, 1:25:42 and amotization. Now, a lot of words if 1:25:44 you're not familiar with finance, it's 1:25:46 not really something that we need to get 1:25:47 too deep into in this video. The only 1:25:49 caveat that's really worth noting is 1:25:51 that typically to fund an e-commerce 1:25:53 business, most people rely financial 1:25:55 tooling like loans. And the reason for 1:25:57 that is it's very capital intensive. To 1:25:59 actually be able to grow quickly, you 1:26:01 need capital to make future inventory 1:26:02 purchasing whilst also continuing to 1:26:04 spend on media. And now with that, you 1:26:06 have interest repayments on those loans. 1:26:08 Question becomes where does interest 1:26:10 repayments go? A common mistake I've 1:26:11 seen on a lot of seven figure and 1:26:13 actually eight figure P&Ls is that 1:26:15 interest expenses will go into OPEX. 1:26:18 This is not true. Interest should not go 1:26:21 into opex. Now, the reason being is that 1:26:23 interest actually isn't an operating 1:26:26 expense. This is leverage to be able to 1:26:29 accelerate the growth of the business. 1:26:30 And therefore, if a buyer or a third 1:26:33 party wants to look at the true profit 1:26:35 generation of this business, they do not 1:26:37 want interest included in the IBIDA 1:26:39 number. Hence, you get interest before 1:26:42 earnings, tax, depreciation, and 1:26:43 amotization. Then, your interest 1:26:46 expenses go down here. And so let's say 1:26:48 that there was $10 in interest expenses 1:26:50 associated with loans. Then you would 1:26:52 have one last line item down the bottom 1:26:54 here which would be $10 in net profit. 1:26:56 So there is technically two definitions. 1:26:58 There's IBIDA which is the capability of 1:27:00 the business to generate earnings if it 1:27:02 didn't have loans outstanding which is 1:27:03 what a buyer wants to understand because 1:27:06 a buyer will just zero out the loans. 1:27:08 And then net profit is the actual profit 1:27:10 that the business ends up generating 1:27:12 after these repayments are made. And 1:27:14 this is the basic P&L structure. At a 1:27:16 high level, we can also, and this is a 1:27:18 very valuable exercise that I'd always 1:27:20 do, allocate percentages to each level 1:27:22 of the P&L to understand relative 1:27:25 percentages throughout the year because 1:27:27 we want to be looking at these on a 1:27:28 month-on-month basis to be able to 1:27:30 understand how these percentages are 1:27:31 changing. So, for example, our cost of 1:27:33 delivery is 40%, our gross margin is 60, 1:27:36 our me is 20, our contribution margin is 1:27:38 40%, this is 20%, IBIDA is 20%, and then 1:27:42 net profit is 10. And so then you can 1:27:44 look at these percentages over time and 1:27:45 go, are they expanding? Are they 1:27:47 contracting? What is our target 1:27:48 percentage allocations here? And this 1:27:50 becomes how you can use the P&L 1:27:52 effectively on a day-to-day or 1:27:54 month-to-month basis to be able to make 1:27:56 decisions. Okay. If me is creeping up, 1:27:58 well, we know our marketing efficiency 1:28:00 is declining and that's deteriorating 1:28:01 and compressing the bottom of the P&L. 1:28:03 So, we need to go and fix that. If our 1:28:05 gross margin is compressing, we need to 1:28:06 look into why that's the case. Is this a 1:28:08 discounting issue, a returns issue? like 1:28:10 what above the P&L is causing the gross 1:28:13 margin compression. So the way to look 1:28:14 at this is that if a target percentage 1:28:16 or a target number on this P&L is not 1:28:18 hitting target, the fault is everything 1:28:20 above it. So if we're not hitting our 1:28:22 IBIT target, it's an issue with either 1:28:24 or all operating expenses, me cost of 1:28:27 delivery. If we're not hitting our 1:28:28 contribution margin target, it's an 1:28:30 issue with me and cost of delivery. If 1:28:32 we're not hitting our gross margin 1:28:34 target, it's an issue with cost of 1:28:35 delivery or revenue. Because remember, 1:28:37 revenue isn't just net revenue, but it's 1:28:39 also gross. And so there's a few 1:28:40 different levers in here that we need to 1:28:41 look at as well. So this is how you 1:28:43 troubleshoot the P&L. If you're not 1:28:44 hitting numbers that you want, you want 1:28:45 to look upwards and look at the levers 1:28:47 that exist there that we need to pull on 1:28:48 and that we need to diagnose. So that is 1:28:50 the P&L explained. Moving into unit 1:28:53 economics. So unit economics, how unit 1:28:55 economics differs from the P&L is the 1:28:57 P&L is a zoomed out view of the entire 1:28:59 business. Unit economics is diving into 1:29:02 a specific unit. Now what this would 1:29:04 typically look like in retail is we 1:29:06 would be looking at a single unit. And 1:29:08 so if we're selling a t-shirt, we would 1:29:10 be looking at this t-shirt and breaking 1:29:12 down what is the price, what is the cost 1:29:13 of goods sold, what is the cost to 1:29:15 acquire a customer for this t-shirt, and 1:29:16 therefore what is the net profit on this 1:29:18 individual unit. Now in e-commerce, in 1:29:20 direct to consumer, this actually 1:29:21 changes a little bit. And we're actually 1:29:23 looking at baskets. And so when we talk 1:29:25 about unit economics, we're actually 1:29:27 typically talking about the basket 1:29:29 because an average cart doesn't just 1:29:32 have one unit in it. Most people will 1:29:34 buy 1.5 things or 1.7 things. That's 1:29:36 called your units per transaction. And 1:29:38 so because of that, we want to 1:29:39 encapsulate multiple different products 1:29:41 into any unit economic calculation. So 1:29:44 the single most important component of 1:29:45 unit economics is something that we 1:29:47 haven't spoken about yet, which is CAC. 1:29:49 This is your cost to acquire a customer. 1:29:52 It's one of the most important metrics 1:29:54 on acquisition and it's a metric that 1:29:55 most brands calculate completely wrong. 1:29:57 How you calculate this is you take total 1:29:59 advertising spend over any given time 1:30:01 period. So this could be on a daily 1:30:03 level, a weekly level, a monthly level. 1:30:05 And then we divide by the total amount 1:30:07 of new customers that we acquired in 1:30:09 that period. This is not the same as CPA 1:30:12 in the platform because CPA is running 1:30:15 off attributed numbers and will end up 1:30:17 showing you a better number than is 1:30:19 actually reflective within the business. 1:30:20 And then people will also try to 1:30:22 delineate CAC down to a channel by 1:30:23 channel level, which is also impossible 1:30:25 because you're relying on attribution. 1:30:26 and attribution has a plethora of issues 1:30:29 which is why we have this whole finance 1:30:31 video together because attribution is 1:30:32 just unreliable. Now the first question 1:30:35 that I typically get when I pull a 1:30:37 trailing CAC calculation for most 1:30:39 businesses is okay well what is a good 1:30:41 CAC? You have access to over 68 N figure 1:30:44 brands. I've personally consulted on 1:30:46 well over a thousand brands to date. I 1:30:48 have a pretty good reference what a good 1:30:49 CAC looks like. The answer to what a 1:30:51 good CAC is is that it's fundamentally 1:30:53 actually the wrong question because CAC 1:30:55 means nothing without contextualizing it 1:30:57 to a pairing metric. And that pairing 1:30:59 metric is gross profit on first purchase 1:31:02 because you could have a $100 cost to 1:31:04 acquire a customer. But if you're making 1:31:06 $10,000 1:31:08 on the order, this is an unbelievable 1:31:11 deal. You're paying $100 and you're 1:31:12 making $10,000 on the order. That's 1:31:14 insane. And so this could be like super 1:31:16 high ticket furniture as an example. Or 1:31:18 vice versa, you could be super low 1:31:19 ticket selling lollies online, which is 1:31:22 probably not a very good niche, and 1:31:23 you're making $20 in GP. This is a 1:31:25 terrible exchange. You're actually 1:31:26 losing $80 on each new customer acquired 1:31:29 here. And so, you need to understand 1:31:31 what is your gross profit on first 1:31:33 purchase, and it's really important on 1:31:34 first purchase. Cuz what people will do 1:31:36 here is they'll take their average order 1:31:38 value, which on Shopify, let's say it's 1:31:40 $100, and then they'll take their gross 1:31:42 profit percentage, which is typically 1:31:43 wrong, calculated wrong. We spent a lot 1:31:45 of time going through that and they'll 1:31:47 go, "Okay, our gross profit on average 1:31:49 is 70%. Therefore, we have a $70 GP, so 1:31:52 we're happy with $35 CAC." Well, is this 1:31:56 on new customers? Because returning 1:31:58 customers always have a higher average 1:32:00 order value, which pulls your average 1:32:02 order value number up. And so, if you 1:32:03 actually delineate down into new 1:32:05 customers, you might find that this is 1:32:06 actually 90 and that your GP is actually 1:32:09 even lower because maybe you have more 1:32:11 discount orientated front-end offers. 1:32:13 And so your GP compresses all the way 1:32:15 down to maybe 50. And so now your CAC 1:32:17 targets are completely wrong in 1:32:19 conjunction with your true GP. And so 1:32:20 this is really just the important of 1:32:22 data integrity within the business. 1:32:24 These numbers need to be calculated 1:32:25 correctly. You need to trust the 1:32:27 calculations or else you're just going 1:32:28 to scale off fundamentally flawed 1:32:30 metrics. So as a part of unit economics, 1:32:33 let's go into pricing strategy because 1:32:36 pricing becomes an enormous lever for 1:32:38 profitability of the business 1:32:40 particularly when products are 1:32:42 underpriced. So, as a new e-commerce 1:32:44 brand, what people will typically do is 1:32:46 they will use keystone pricing. Now, I 1:32:49 think anyone that's ever run a business 1:32:51 or started an ecom brand has done this, 1:32:53 which is that you take your cost of 1:32:55 product, so you go to the manufacturer, 1:32:56 product cost $10, and you just double 1:32:58 it. This is what's called keystone 1:33:00 pricing. So, as an example, a $6 cost of 1:33:03 goods just gets 2xed and becomes a $12 1:33:05 price. So many people price this way. I 1:33:07 reckon 40% of the startup market just 1:33:09 prices in this way. This is probably the 1:33:11 worst way you could ever price. Now, 1:33:13 it's because this cost of goods doesn't 1:33:14 encapsulate all the variable expenses 1:33:16 associated with fulfillment. And so, you 1:33:17 end up with way more compressed margins 1:33:19 than just 50%. I think the reason why 1:33:21 people use Keystone pricing is number 1:33:23 one, it's simplicity, but number two, 1:33:25 the fact that people don't want to 1:33:26 overpric. People think that margin is 1:33:28 bad. We can't have too much margin or 1:33:30 else we're just ripping people off. But 1:33:32 because of that, they don't take into 1:33:33 consideration all of the expenses 1:33:35 associated to actually fulfill within a 1:33:37 business. And so because of that, people 1:33:39 underpric and they can never actually 1:33:40 scale. Whereas the actual cost to grow a 1:33:43 business in all of the variable expenses 1:33:44 associated with fulfillment as well as 1:33:46 all the marketing expenses associated 1:33:48 and operating expenses taking up 15%. It 1:33:50 ends up being a lot. And so you end up 1:33:52 needing to price way higher than just a 1:33:54 doubling using keystone pricing. So the 1:33:56 next pricing model that people jump to 1:33:59 is IMU pricing, which is initial markup 1:34:01 pricing. So this is effectively the same 1:34:03 thing as Keystone, but rather than doing 1:34:04 a 2x, maybe we do a 3x or we do a 4x or 1:34:08 a 5x. So we're effectively taking the 1:34:09 cost of goods once again, and we're just 1:34:11 applying a multiple to this number. Now, 1:34:14 that might seem better, right? You're 1:34:15 like, well, Nathan said that the big 1:34:16 issue with Keystone is you're not 1:34:18 increasing by enough. So if we just 1:34:19 increase to 345x, then surely that fixes 1:34:21 the issue. It doesn't fix the issue 1:34:22 because this is just an arbitrary 1:34:24 multiple on a product cost, which has 1:34:27 nothing to do with the price elasticity 1:34:28 of demand within the market. It has 1:34:30 nothing to do with the competitors and 1:34:31 it has nothing to do with the actual 1:34:33 cost structure associated with 1:34:35 delivering that particular product. And 1:34:36 so yes, we could just 4x it now and the 1:34:38 price is $24. Amazing. But this might be 1:34:40 substantially overpriced compared to 1:34:42 competitors. This might still not 1:34:44 encapsulate enough margin to be able to 1:34:46 actually fulfill on this business model. 1:34:48 And so all of these things have to be 1:34:49 taken into consideration. The real core 1:34:51 takeaways here is that you want to use 1:34:54 waterfall pricing from bottom to top to 1:34:57 be able to identify the price that you 1:34:59 need to actually price the product to 1:35:01 make money. And then from there, you 1:35:03 make the decision of do we even sell 1:35:05 this product? So here's all the expenses 1:35:06 associated with actually delivering on 1:35:08 the product. We need to start at the 1:35:09 bottom. So what is our actual 1:35:11 contribution profit target here? Okay, 1:35:12 is it 20%, 30%, 40% based on our 1:35:15 existing P&L understanding and how much 1:35:17 operating expenses we have. So, you 1:35:19 could really start all the way down the 1:35:20 bottom here, right? And take this a step 1:35:21 further and say that our net profit 1:35:23 target is 10%. We know that our 1:35:26 operating expenses 10%. Therefore, our 1:35:29 contribution target is going to be 20%. 1:35:32 Okay, great. Now, we need to start 1:35:33 moving up. How much do we think it's 1:35:35 going to cost to acquire a customer on 1:35:36 this product? Now, based on our 1:35:37 understanding, based on the ad account, 1:35:39 based on other products, we think our 1:35:40 CAC is going to sit at $30. Now, what's 1:35:42 the return allowance? This is also 1:35:43 called a shrink allowance, which is that 1:35:45 we're allowing for shrink in discounts, 1:35:47 returns, etc. Well, we probably want a 1:35:50 3% return allowance, probably a 3% 1:35:52 discount allowance. Shipping and 1:35:53 fulfillment on this product, we've gone, 1:35:55 we've reached out to Opost or the 1:35:56 shipping courier, and we know that this 1:35:58 is going to be $11. Then we have the 1:36:00 cost of goods, which we know is $6 over 1:36:02 here. And now we can work our way all 1:36:04 the way up to pricing. So, we need to be 1:36:05 priced at 30 + 11 + 6 plus 3% 3% of this 1:36:10 total price. And then we need to come 1:36:11 out to a 20% contribution margin. And so 1:36:13 we can run all of this math, which I'm 1:36:15 not going to do for the sake of this 1:36:16 video, but let's say this takes us up to 1:36:17 a $100 price because 100 - 6 - 11 takes 1:36:20 us down to about 85. Then we minus 1:36:22 another 6% which takes us down to 79. 1:36:25 Take us off another 30, that takes us to 1:36:28 49. And so down the bottom here, we have 1:36:31 a $49 contribution profit, which is 1:36:33 actually a really high contribution 1:36:35 margin, way above our actual target. So 1:36:37 if we just keep reworking these numbers, 1:36:38 I believe this should come out to 1:36:39 probably like a $70 price. So to hit 1:36:41 this contribution margin target with 1:36:43 this CAC with these unit economics, we 1:36:45 come out to a $70 price at the top. This 1:36:47 is how you price correctly. And you then 1:36:50 go and you do competitor research. And 1:36:52 guess what? If this $70 price is 1:36:54 ridiculous, if everyone else is pricing 1:36:55 at $30, $40, well, either number one, we 1:36:57 need to figure out, can we position into 1:36:59 a blue ocean where no one's actually 1:37:01 selling a premium version of this 1:37:02 product? Can we position this as 1:37:04 premium? Can we sell at this price? And 1:37:06 if we can't, we don't sell the product. 1:37:08 The unit economics don't work end to 1:37:09 end. We can't hit our profitability 1:37:11 targets by selling this product any less 1:37:13 than $70. No one will buy it at $70. 1:37:16 Let's cut the product entirely and not 1:37:17 sell it. So that is how you need to be 1:37:18 thinking through pricing strategy. Now 1:37:20 you need to understand the unit 1:37:22 economics of discounting. This is where 1:37:25 most brands destroy their margin without 1:37:27 even realizing it. Now I'll give you a 1:37:28 really quick worked example. So if we 1:37:30 have a $100 average order value or 1:37:33 product price, this means we're going to 1:37:35 come out to a $40 gross margin. Now what 1:37:38 we can do here to calculate our break 1:37:39 even return on ad spend. So what return 1:37:41 on ad spend do we need to be to break 1:37:43 even is we take our gross margin which 1:37:45 is 40%. So break even return on ad spend 1:37:48 equals 1 / 40%. And this equals 2.5. So 1:37:53 we need to achieve at least a 2.5 return 1:37:55 on ad spend to break even and therefore 1:37:58 make money. So this is the lowest we can 1:37:59 be. Great. This is a really important 1:38:01 number for not only your internal team 1:38:02 and your agency to know because if let's 1:38:04 say an ad set or a campaign is below 1:38:06 this you're losing money. It's really 1:38:07 important to understand what this number 1:38:08 actually sits at, but more important to 1:38:10 understand where this number sits at 1:38:12 during a discount period. So, if we take 1:38:14 this exact same example, you're on 30% 1:38:17 discount. So, you get slashed to a $70 1:38:19 price. $70 price, your cost of goods 1:38:22 stays the same. And so, we have our $60 1:38:24 cost of goods here. That then comes out 1:38:26 to a $10 gross margin. If we then 1:38:29 recalculate our break even return here, 1:38:31 our break even return becomes 1 / 10%, 1:38:35 which is 10. So, our break even return 1:38:37 on ad spend, the efficiency that we need 1:38:39 to operate is 5x what it was up here off 1:38:42 just a 30% off discount, which is crazy. 1:38:44 And it's because discounting 1:38:45 exponentially increases the break even 1:38:48 return or the efficiency that needs to 1:38:49 be hit within paid media. And so, 1:38:51 anytime you're discounting, it is really 1:38:53 important to run this exact calculation 1:38:55 to be able to understand what efficiency 1:38:57 level we need to be at. Now, all of this 1:38:58 math can also be done based on a CAC or 1:39:01 a CPA number. So rather than doing this 1:39:03 division to calculate break even return, 1:39:05 all you do is this gross margin number 1:39:07 is your break even CAC. So your break 1:39:08 even CAC just simply equals gross 1:39:10 margin. So here break even CAC is 40. 1:39:12 Here break even CAC is 10. Obviously a 1:39:14 $10 CAC is absolutely insane. You're 1:39:16 probably never going to hit that across 1:39:18 paid platforms. Therefore, this discount 1:39:20 will never work for acquiring new 1:39:21 customers. So this shouldn't go out 1:39:23 publicly. Now this isn't to say that you 1:39:24 can't discount. You can't do blank 1:39:26 discounts. Of course you can. You just 1:39:27 need to understand how it affects the 1:39:28 efficiency targets within paid and 1:39:31 whether you can actually hit those 1:39:32 efficiency targets. You always need to 1:39:33 just run that math which is that if 1:39:34 we're going to do a 20% off discount, 1:39:36 how does our efficiency on paid media 1:39:37 need to change to maintain the same 1:39:38 level of profitability or improve 1:39:40 profitability? And if it's insane, if 1:39:42 it's like we need to go from a 2x row to 1:39:44 a 12, well that is physically 1:39:45 impossible. The discount isn't going to 1:39:47 create that much of an uplift in demand 1:39:49 and conversion rates. Therefore, we need 1:39:50 to rethink the discount approach and 1:39:52 what we're actually doing with this 1:39:53 offer. And so, let me give you some 1:39:54 other options. Well, bundling is a very 1:39:57 big one. Everyone knows this. This is 1:39:59 pretty generic, which is that if you 1:40:00 bundle, you get economies of scale 1:40:03 because you might go from one unit in 1:40:04 the cart, too. But this doesn't increase 1:40:06 your shipping and fulfillment cost from 1:40:07 1 to two. Generally, you will get maybe 1:40:09 a 20% inflation in the shipping and 1:40:11 fulfillment cost of the product. And so, 1:40:13 therefore, you actually get better gross 1:40:14 margin when you bundle. And you can give 1:40:15 that margin away to the customer, and it 1:40:17 doesn't affect your GM percentage. 1:40:19 Bundling also increases average order 1:40:21 value substantially, which allows for 1:40:22 you to have a higher cost of acquiring a 1:40:24 customer on the platform and still make 1:40:26 more money. And so bundling is still a 1:40:29 really effective way to provide a 1:40:31 discount to the end consumer, provide an 1:40:33 offer that seems to have a value 1:40:35 discrepancy in the market that allows 1:40:36 someone to get over the line and buy, 1:40:38 but it is beneficial to you in regards 1:40:40 to the unit economics of that bundle. 1:40:42 Another example here is a gift with 1:40:45 purchase. And so this is where you just 1:40:47 need to become good at calculating unit 1:40:50 economics correctly on offers because 1:40:52 this is what allows you to validate 1:40:53 whether an offer will work. It allows 1:40:55 you to bake in the assumptions and then 1:40:56 you can go and actually run it in 1:40:58 public. So an example of this would be 1:41:00 if you buy two beach towels, you get a 1:41:02 free bag that actually holds the beach 1:41:03 towel. Now the beach towels might be 1:41:05 $100 each is what they're priced at and 1:41:07 they're pretty high margin. Let's say 1:41:08 70% or something. So you go and when 1:41:10 people buy two, it's $200, but they get 1:41:12 a free bag. And you can say the free bag 1:41:14 is valued at or sold on the website at 1:41:17 maybe $50, $60. You can price it really 1:41:19 high. It can be seen as a premium bag. 1:41:20 Uh but the reality is is that the cost 1:41:22 of goods on this bag for you is maybe 1:41:24 $4. And so you're going to forgo $4 of 1:41:26 margin to get an extra $100 in revenue. 1:41:29 Really good exchange. Okay, your gross 1:41:30 margin is pretty much going to stay the 1:41:32 exact same as a percentage, but you're 1:41:34 doubling average order value by giving 1:41:35 this free bag away. So gift with 1:41:37 purchase ends up working really well if 1:41:39 the offer is crafted well and if the 1:41:41 gift is relatively low cost. And then 1:41:44 the last is straight uh discounting 1:41:47 which I just told you compresses margin 1:41:49 a lot and you have to be really careful 1:41:50 about but the caveat here is that you 1:41:52 can do this on grade C inventory. We'll 1:41:55 talk about stages of inventory later in 1:41:57 this video and we'll break down actual 1:41:59 strategies to be able to move inventory 1:42:00 and focus on cash versus net profit and 1:42:02 the marketing strategies associated. But 1:42:04 just as a call out, if you are going to 1:42:05 do flat discounting, you want to do it 1:42:07 on inventory that isn't moving. And then 1:42:09 you can move that inventory back into 1:42:10 cash, even if it is at break even. It 1:42:13 doesn't really matter because the 1:42:14 inventory wasn't going to sell anyway. 1:42:15 Now, I can't talk about unit economics 1:42:17 and finance without talking about LTV 1:42:20 and repeat purchase rates. And the 1:42:22 reason this is so critical is because 1:42:24 acquisition is expensive. Fundamentally, 1:42:26 it is expensive to acquire customers. 1:42:28 Particularly if we start going into an 1:42:29 industry like CPG, consumer package 1:42:31 goods. you're not going to probably even 1:42:33 be profitable on first purchase. And the 1:42:35 reason being is that you just have 1:42:36 competitors that will outspend you, 1:42:38 outbid you at auction, on Meta, on 1:42:40 Google, and they can go and pay $200 to 1:42:42 acquire a customer because they have 1:42:43 this massive lifetime value that they 1:42:45 can then realize on second, third 1:42:46 purchase. And if you don't have that, 1:42:48 you will lose because your competitors 1:42:49 will just spend more than. And so 1:42:51 lifetime value becomes a critical 1:42:53 component in profitability of most 1:42:55 business models. I can give you a 1:42:57 completely different example of this 1:42:59 outside of e-commerce as a whole, which 1:43:00 is actually the agency model, which is 1:43:02 something that I'm very familiar with. 1:43:03 In the agency model, most agencies won't 1:43:06 run profitable on first purchase or 1:43:08 first invoice on month one. Most 1:43:09 agencies will run at a 3 to six month 1:43:11 CAC payback period. So, they will spend 1:43:13 X amount to be able to get a client, 1:43:15 whether that's associated with 1:43:16 marketing, sales costs, etc., etc., 1:43:18 networking, events, and then once they 1:43:20 have a client, they will only start 1:43:21 making money after 3 to 6 months due to 1:43:23 the CAC. Same thing in CPG, same thing 1:43:25 in fashion, retail, etc. Except in these 1:43:28 other industries, you need to carefully 1:43:30 and meticulously understand what your 1:43:31 LTV actually is so that you can operate 1:43:33 that model profitably. Because if you 1:43:35 are an agency, let's say, and you don't 1:43:37 understand all the associated cost to 1:43:39 get the customer and you miscalculate 1:43:41 CAC or you miscalculate LTV, you are 1:43:44 actually in a very different 1:43:45 profitability position and you can 1:43:46 really negatively impact yourself. So, 1:43:48 let's run through a quick example. Let's 1:43:50 say you're doing 200k a month in 1:43:52 revenue. Of that 200k a month, $140,000 1:43:56 is new customer revenue. And let's say 1:43:58 this is coming from 50k in ad spend. 1:44:00 That then means you have $60,000 in RC 1:44:04 revenue. And let's say that the cost 1:44:05 here is like 2.5K, which is associated 1:44:08 Clavio costs and maybe an agency 1:44:10 associated with driving repeat revenue. 1:44:12 Now, your new customer profit 1:44:14 contribution, which isn't a metric that 1:44:15 we've gone through yet, but it's just 1:44:16 contribution margin, but we're doing it 1:44:18 on new customer revenue only. This is 1:44:21 140 minus 50. So, we have $90,000, but 1:44:24 we obviously also need to minus cost of 1:44:26 delivery and gross margin. So, let's 1:44:28 assume a 50% gross margin. That means 1:44:30 that this is going to come to 70 - 50. 1:44:32 This is going to be $20,000 in profit 1:44:34 contribution that we're making each 1:44:36 month on new customers. Now, what about 1:44:39 RC revenue? Well, 30 - 27, this is 1:44:43 27,500 1:44:45 on return. And so this business, which 1:44:46 is a pretty typical business in terms of 1:44:48 new customer revenue to returning 1:44:50 customer revenue percentages at this 1:44:52 kind of size, this business is making 1:44:54 more on returning customers in profit 1:44:56 per month than they are on you. And this 1:44:58 is super typical because majority of 1:44:59 your profit ends up coming from 1:45:01 returning customers. Now, why is that 1:45:02 the case? Well, because it's very hard 1:45:04 to get a new customer. You have to pay 1:45:05 money to get them. Returning customers 1:45:07 come back due to the product, the 1:45:09 experience, and the brand affiliation. 1:45:11 So you actually don't need to spend much 1:45:13 at all on getting a returning customer 1:45:14 and they drive tons of profit 1:45:16 contribution in the business. Now once 1:45:18 you start getting into this position, 1:45:19 that's where you start needing to 1:45:20 understand, well, can we actually push 1:45:22 up new customers even harder and 1:45:23 subsidize this acquisition cost with all 1:45:25 the profit that we're making on 1:45:26 returning and continue to grow the 1:45:28 business that way. And that's where I 1:45:29 just want to provide a massive 1:45:31 hesitation to most people, which is that 1:45:33 most people calculate LTV wrong. Number 1:45:35 one, they will calculate LTV based on 1:45:38 just infinity. And so when we look at 1:45:40 lifetime value, uh, as you extend the 1:45:42 time period in which you're looking at 1:45:43 how much a customer is worth to you, it 1:45:45 increases forever. And so we can look at 1:45:47 how much a customer is worth to us after 1:45:49 6 months and it might be $100. And then 1:45:51 we can look at 12 months and it might be 1:45:52 $10. Then we can look at 18 months, it 1:45:54 might be $120. And as you just keep 1:45:55 extending that time horizon, LTV 1:45:58 increases forever. Now it does somewhat 1:46:00 asotope. So it will look something like 1:46:02 this. But still, if we measure from here 1:46:05 to here, there is still an increase. And 1:46:07 so what you want to be very careful of 1:46:09 is not just looking at an LTV 1:46:11 calculation based on total customer data 1:46:13 over forever and instead you want to 1:46:16 restrict it to a particular time period. 1:46:18 Now a good way to do this is to look at 1:46:20 90day or 180day LTV. The second 1:46:24 delineation you want to make is you want 1:46:26 to go to LTGP. You want to be looking at 1:46:28 gross profit, not value. Now, LTV in the 1:46:31 traditional value sense actually is 1:46:33 lifetime gross profit. When we measure 1:46:35 this against software because this is a 1:46:37 software metric and in software, 1:46:38 lifetime value is typically very close 1:46:40 to lifetime gross profit and therefore 1:46:41 it's the same thing. But in e-commerce, 1:46:43 we want to make this clear delineation 1:46:45 because some people and a lot of 1:46:46 softwares actually will give you LTV as 1:46:48 a revenue number. And so we want to 1:46:49 understand what the gross profit are of 1:46:51 these customers at 90 180 days. And then 1:46:54 we can use that to be able to understand 1:46:55 what we can actually pay to acquire a 1:46:57 customer. Now, let's take unit economics 1:46:59 and apply it at a product level. So, on 1:47:01 product level economics, let's say you 1:47:03 have a hoodie that you sell and you sell 1:47:05 it at $90 retail with a $21. So, you 1:47:09 have a hoodie at $90, cost of goods 21. 1:47:11 So, you have gross margin of $69. Our 1:47:13 cost to acquire a customer through the 1:47:15 hoodie on our advantage plus scaling 1:47:17 campaign is $45 right now. So, that's 1:47:19 our CPA. We're just going to call it 1:47:20 CAC. So, our contribution margin here is 1:47:22 $24. Now, we have a t-shirt that's 1:47:24 priced way less. Cost of goods is about 1:47:26 the same. Actually, our gross margin is 1:47:29 only 27, but we have a CAC of 15 and we 1:47:31 have a contribution margin of $12. So, 1:47:34 right now, the hoodie is driving 2x the 1:47:36 contribution margin to the business per 1:47:39 sale. So, this firstly is an important 1:47:40 number for us to be across. Okay, 1:47:42 understanding what actual products 1:47:43 within the portfolio is driving us the 1:47:45 most contribution margin when we break 1:47:47 our unit economics down at a product 1:47:50 level. The second component that becomes 1:47:52 important here is understanding how we 1:47:54 should structure our campaigns 1:47:56 accordingly. Should we be splitting 1:47:58 these out? Because if we're using a 1:48:00 bidding strategy on Meta, Google, Tik 1:48:03 Tok, which is maximize conversions, and 1:48:05 this is the default bidding strategy on 1:48:07 every advertising platform. When you use 1:48:08 maximize conversions, what it's 1:48:10 optimizing for is the lowest CPA. So, 1:48:13 whatever product is driving the lowest 1:48:15 CPA, whatever ad is driving the lowest 1:48:17 CPA, that will get prioritized, that 1:48:19 will get the spend. Now, if we look at 1:48:21 this example here, the t-shirt actually 1:48:23 has the lower CPA. And so, Tik Tok, 1:48:26 Meta, Google will distribute all your 1:48:29 spend here, but this product is driving 1:48:31 a lower contribution margin per sale. We 1:48:33 would actually prefer Meta to put all of 1:48:35 our spend up here. Yes, it's going to 1:48:36 cost us a little bit more money, but 1:48:38 we're going to drive more contribution 1:48:39 margin per product sold. So this is the 1:48:42 better place to put cash right now. 1:48:44 Maximize conversions won't do that. Now 1:48:45 you can fix this in the platforms in a 1:48:47 number of ways. You can have 1:48:48 segmentation across product categories 1:48:50 based on product level economics which 1:48:51 is really important to do. You can 1:48:53 change the bidding strategy to maximize 1:48:55 for conversion value and therefore we 1:48:57 will actually prioritize this because it 1:48:58 has a better rorowaz. This actually has 1:49:00 a worse rorowaz too. So maximize 1:49:01 conversion value won't fix this issue 1:49:02 either. So segmentation is the way that 1:49:04 you actually need to fix this. Now the 1:49:06 other reason why this is an important 1:49:07 exercise is particularly in fashion what 1:49:10 you were trying to do is sell through 1:49:12 all of your product highest contribution 1:49:14 margin possible. What we need to start 1:49:16 thinking about in this instance is this 1:49:18 t-shirt will this sell anyway without us 1:49:21 even pushing it and paid cuz the cost to 1:49:23 acquire is so low here that it seems 1:49:25 like there is some kind of virality 1:49:27 component. there is some kind of product 1:49:29 market fit that's just getting this 1:49:30 t-shirt to sell kind of regardless of 1:49:33 our paid media spend because this is so 1:49:35 hyperefficient. So the question we need 1:49:36 to ask particularly if we're an omni 1:49:38 channel business with retail stores or 1:49:40 we're just a very large business in 1:49:41 general is would people have bought this 1:49:43 product anyway if we didn't spend? And 1:49:46 if that's the case well then guess what 1:49:48 let's not spend. Let's then take our 1:49:50 contribution margin up on this order by 1:49:52 $15 to $27. and let's actually 1:49:55 reallocate all the spend to the hoodie 1:49:57 which might not be selling naturally or 1:49:59 organically and we need the paid spend 1:50:01 to be able to drive sellrough rate. So 1:50:02 this is where product level economics 1:50:04 becomes incredibly important. It's 1:50:06 something you should be breaking down. 1:50:07 You can obviously build Google sheets 1:50:09 around this. Have visibility build it 1:50:10 into your decision-m. Now moving into 1:50:12 the metrics that matter. How we think 1:50:14 through this internally is through a 1:50:16 pyramid. We have platform level metrics 1:50:18 down the bottom. We then move into 1:50:20 finance grade metrics which we've been 1:50:22 covering a lot in this video. And then 1:50:23 up the top we move into incrementality 1:50:26 testing which probably isn't applicable 1:50:27 for 80% of people watching this but for 1:50:29 the 20% that's doing over $10 million a 1:50:32 month that has a presence in the US or 1:50:34 multiple different markets might be omni 1:50:36 channel this is going to be critical for 1:50:38 ensuring measurement within the business 1:50:39 and it's going to tie into understanding 1:50:41 how finance ties into actual platform 1:50:43 metrics. So if we start down the bottom 1:50:45 we have rorowaz CPA CTR multi-touch 1:50:48 attribution like triple whale etc. Now 1:50:50 why is rorowaz down the bottom of the 1:50:52 pyramid? Why is rorowaz not good? The 1:50:54 reason why return on ad spend isn't a 1:50:56 good metric is not because calculating 1:50:58 ROI is bad, calculating return on 1:51:00 investment is a great thing to do. The 1:51:02 reason why rorowaz is unreliable is 1:51:04 because it relies on attribution. And 1:51:07 attribution is fundamentally an 1:51:09 unknowable reality where you're trying 1:51:11 to connect correlation within platforms 1:51:13 with faulty data to be able to prove 1:51:15 causation. And what do I mean by all 1:51:17 that complex language? When someone gets 1:51:19 served an ad and then they click on the 1:51:20 ad and then they go to the website and 1:51:22 they purchase. You might think that this 1:51:24 is causation that this ad click caused 1:51:27 the purchase. But in a lot of cases 1:51:29 particularly in large omni channel 1:51:30 retailers that actually is not the case. 1:51:33 This is correlation and in fact all 1:51:35 attribution is just correlation. We are 1:51:38 saying that these two events are 1:51:39 correlated. The traditional way to 1:51:42 explain correlation versus causation is 1:51:44 that at the same time throughout the 1:51:46 year, swimming deaths go up and ice 1:51:47 cream sales go up. And you go, well, is 1:51:49 swimming deaths causing ice cream sales? 1:51:52 Is ice cream sales causing swimming 1:51:53 deaths? No, they're not causal. They're 1:51:56 correlated to summer. When it is summer, 1:51:58 more people die when they're swimming in 1:52:00 the ocean and more people buy ice cream. 1:52:01 And it's the same thing here. Ad clicks 1:52:03 is not necessarily causal to purchasing. 1:52:06 It is simply correlated. And so because 1:52:08 of this return on ad spend ends up 1:52:10 lacking validity and congruency to the 1:52:13 P&L. What you see a lot of the time is 1:52:16 that people will go and say rorowaz is 1:52:17 unbelievable. Wow but my business is 1:52:19 dying. What is going on? And it is 1:52:21 because rorowaz has a few dynamics to 1:52:23 it. Number one because of the way that 1:52:25 it attributes. It will always 1:52:27 overattribute to the bottom of funnel. 1:52:29 Because people that for example might 1:52:30 see a billboard up here and this is 1:52:32 obviously a super extreme example but 1:52:34 let's say someone sees a billboard and 1:52:35 they become aware of uh the brand. They 1:52:37 then see a TV ad which reinforces that 1:52:39 they actually really want to buy your 1:52:40 stuff because there was an influencer in 1:52:41 there who they connect with. Then 1:52:42 eventually they get an ad on Facebook. 1:52:44 They click on the ad and they buy. Now 1:52:45 did that ad on Facebook cause them to 1:52:47 buy? Probably not. Like yes it got it 1:52:49 over the edge at that specific moment 1:52:50 but they were going to buy regardless at 1:52:52 the next point of activation. It was the 1:52:54 billboard on the TV that actually warmed 1:52:56 them up at the top of funnel but these 1:52:57 got no attribution. Now, it's the same 1:52:59 thing across platforms. And so, a really 1:53:01 evident example of this actually right 1:53:02 now for us, and this might change 1:53:04 depending on when you're watching this 1:53:05 video is Tik Tok to Meta to Google. Now, 1:53:08 this isn't applicable for everyone. This 1:53:10 is a very unique circumstance, but it at 1:53:12 least gets the point across, which is 1:53:13 that we're finding at the moment for 1:53:14 some very large retail brands that Tik 1:53:17 Tok is actually underattributing 1:53:18 substantially because it's getting 1:53:20 impressions. It's getting in front of 1:53:21 people in new markets, but they don't 1:53:23 actually click off the platform much. or 1:53:24 if they do click off the platform, they 1:53:26 click off, they look at the product 1:53:27 page, but they don't buy. Then once we 1:53:28 have them in the pixel, Meta goes and 1:53:30 follows up and retargets them. And we 1:53:32 end up capturing a lot of increased 1:53:34 demand on Meta when we increase our Tik 1:53:36 Tok spend. And then finally, people that 1:53:38 still haven't bought after Tik Tok and 1:53:40 Meta, they end up going to Google. They 1:53:41 search for the brand name, they click 1:53:43 and buy. And so when we increase Tik Tok 1:53:45 spend, our meta rorowaz goes up and our 1:53:48 Google rorowaz goes up. Tik Tok doesn't. 1:53:50 If we're using rorowaz as the indicator 1:53:52 for performance and budget allocation 1:53:54 across the business particularly and 1:53:56 this is where this becomes a very large 1:53:57 issue is when this is reporting up to a 1:53:59 seuite a seuite will see return on ad 1:54:02 spend numbers across these different 1:54:04 channels and go okay tik tok we should 1:54:06 cut it's not driving returns meta is 1:54:09 okay let's decrease budget and team 1:54:11 let's put more budget into Google that 1:54:12 would be a fundamentally terrible idea 1:54:14 because what we're effectively saying 1:54:15 there is let's cut all the top of funnel 1:54:16 generation and move to just bottom of 1:54:18 funnel bottom of funnel will stop 1:54:19 working unless there is top of funnel 1:54:21 generation. And so rorowaz ends up 1:54:22 misleading people in terms of decision-m 1:54:24 unless there's a lot of additional 1:54:25 nuance in understanding what is actually 1:54:27 driving impact in the business. And so 1:54:29 we need to go to better measurement 1:54:30 systems to be able to actually 1:54:31 encapsulate this, understand it, and 1:54:33 then communicate it to a seauite or if 1:54:35 you're a small business just to you as 1:54:36 the founder. Now we could go on and on 1:54:38 about the other limitations of 1:54:40 attribution and return on ad spend here, 1:54:41 but this is fundamentally the crux of 1:54:43 why this metric misleads you in terms of 1:54:45 decision-making. So then we move one 1:54:47 stage up in the pyramid and we get to 1:54:49 financial grade KPIs. Now these are 1:54:52 metrics like acquisition me. You can 1:54:54 throw me in here as well. We've got 1:54:56 profit contribution which we've ran 1:54:58 through. We have CAC which we've ran 1:55:00 through. So these are all metrics that 1:55:01 aren't relying on the inplatform 1:55:03 attributed numbers to be able to 1:55:04 calculate them, but instead they're 1:55:06 using the actual financial metrics that 1:55:08 exist within the business. So the actual 1:55:10 amount of new customers that you're 1:55:11 acquiring, the actual new customer 1:55:12 revenue, and then we're dividing by the 1:55:14 actual ad spend. So this is all metrics 1:55:16 that aren't relying on any kind of 1:55:18 correlation calculation but instead are 1:55:20 directly associated to financial metric. 1:55:23 And so because of that the reliability 1:55:24 is much better. As we move up the 1:55:26 pyramid reliability improves however 1:55:29 speed decreases. And so what ends up 1:55:31 being the case with these metrics is 1:55:33 them going up and down. You need to look 1:55:35 at this on slightly longer time periods 1:55:37 to be able to make correct decisions 1:55:39 whereas you will get faster feedback 1:55:41 loops generally speaking on your 1:55:43 inplatform metrics. So why would we even 1:55:45 use this bottom half of the pyramid at 1:55:47 all? Why would we even look at rorowaz 1:55:49 CPA? They are helpful in directional 1:55:51 campaign feedback and ad feedback within 1:55:54 the siloed platform. So what we do not 1:55:56 want to do, what would be a big mistake 1:55:58 is comparing a rorowaz number on a 1:55:59 Google campaign to a rorowaz number on a 1:56:01 meta campaign because they have 1:56:03 different attribution models. They're 1:56:04 sitting at different points in the 1:56:05 funnel. They're driving different 1:56:07 incremental impact to the business. If 1:56:09 we took both of the campaigns and we 1:56:10 pushed budget up, one of them would 1:56:12 drive excess return compared to the 1:56:13 other irrespective of the rorowaz number 1:56:15 on the campaign. So these are helpful 1:56:17 directionally within the own platform. 1:56:19 So if we're looking at two meta 1:56:20 campaigns next to each other, we can 1:56:21 compare rorowaz, we can use that to make 1:56:23 decisions. But if we start comparing 1:56:24 platforms, if we start zooming out and 1:56:26 making larger business level decisions 1:56:27 based on these numbers, that's where we 1:56:28 can start to mislead ourselves and go 1:56:30 into using low reliability, low data 1:56:33 integrity level metric. And then all the 1:56:35 way up the top here, we have 1:56:36 incrementality testing. We won't dive 1:56:38 into this in too much detail in this 1:56:40 video, but the crux of the way that this 1:56:42 works is that you take an area, 1:56:44 generally the whole country. You split 1:56:45 down by states. In Australia, you have 1:56:47 to split down by commuting zones rather 1:56:49 than states because there's not enough 1:56:51 state selection. And from there, you 1:56:52 hold out a particular area. Now, 1:56:54 typically it might even be two states. 1:56:56 So, we can go and grab two different 1:56:57 states that are next to each other. And 1:56:58 then we hold them out. Now, a hold out 1:57:00 means that we increase spend everywhere 1:57:02 else except here. and we see what is the 1:57:04 differential or we can just increase 1:57:06 spend here in the control group or we 1:57:08 can cut spend entirely. So we have a lot 1:57:10 of different options in terms of test 1:57:11 design here but the idea is that when we 1:57:13 isolate a large control verse treatment 1:57:16 group we can start to measure the 1:57:18 incrementality of changing different 1:57:19 campaign types. And so we might go and 1:57:21 take that Tik Tok campaign I was talking 1:57:23 about before that doesn't seem to be 1:57:24 doing well and we might double budgets 1:57:26 in these states. And then over a 30-day 1:57:28 period, we measure the revenue 1:57:29 realization difference over that 30 days 1:57:31 and maybe new customer revenue doubles. 1:57:33 And so as a function of that, we can 1:57:34 take the lift in new customer revenue 1:57:36 against the control. We can do an 1:57:37 incremental return on ad spend 1:57:39 calculation and we end up getting a very 1:57:41 accurate read on the impact that these 1:57:43 campaigns are making. Now that is an 1:57:45 unbelievably simplified explanation of 1:57:47 how this test design actually works. 1:57:49 It's actually a lot more complicated. I 1:57:51 could make a 2-hour video just on the 1:57:53 data science and approach to building 1:57:54 incrementality tests, which I think we 1:57:56 actually might do at some point. That is 1:57:57 the crux of the idea as to how it works. 1:58:00 I do not recommend that you go away from 1:58:02 this video and think, "Oh, that was a 1:58:03 really interesting idea. Let me just 1:58:04 make a Google sheet and start doubling 1:58:06 budgets in states." If you do that, you 1:58:07 won't get good results. It'll pretty 1:58:09 much always say states aren't 1:58:10 incremental because the test design 1:58:11 isn't set up correctly. You'll end up 1:58:13 way overspending or under spending. You 1:58:14 won't get statistically relevant 1:58:16 results. you need at least a fundamental 1:58:17 understanding of data science to be able 1:58:19 to start to approach this portion of the 1:58:21 pyramid. Um, but that is ultimately the 1:58:23 most reliable option. It's just very 1:58:25 slow. You can't run a lot of them and 1:58:26 you're quite restricted. Now, I said 1:58:28 before that I was a little bit hesitant, 1:58:29 put me in here. And the reason why I was 1:58:31 a little bit hesitant is that me is 1:58:33 revenue divided by ad spend. Sometimes 1:58:36 people will do this the other way round 1:58:37 and they'll do ad spend divided by 1:58:39 revenue, which will give you a 1:58:40 percentage number. This instead will 1:58:42 give you a multiple number. And so if we 1:58:44 for example have $100 in revenue and $50 1:58:46 in ad spend, our me would be a two. So 1:58:48 every dollar we spend on ad spend, we 1:58:50 get in revenue. You could also look at 1:58:51 this as an ROI calculation. Now this 1:58:53 looks really good and this looks like a 1:58:54 good way to index the performance of 1:58:56 total spend across all of our media 1:58:58 channels against total revenue. But the 1:58:59 reason why it's fundamentally a terrible 1:59:01 metric for indexing the performance of 1:59:02 paid media over time is that it includes 1:59:05 a large bucket of revenue that has very 1:59:07 little to do with paid ad, which is 1:59:09 returning customer revenue. And so we 1:59:10 really want to strip returning customer 1:59:12 revenue out of this calculation because 1:59:14 we don't want marketing taking credit 1:59:16 for all these returning customers that 1:59:18 would have returned anyway. So that's 1:59:19 why we always want to delineate all of 1:59:21 our finance grade metrics into new 1:59:23 customerbased metric. And so rather than 1:59:25 looking at me, we want to look at a me 1:59:28 which is new customer revenue divided by 1:59:30 ad spend. And this might actually give 1:59:32 us in this example a 1.4 which might not 1:59:34 actually be profitable at all for us. 1:59:36 And so we might want to be rethinking 1:59:37 our whole acquisition strategy based on 1:59:39 this delineation to a better number. Now 1:59:41 there is a metric that's better than 1:59:42 everything that I just put down in the 1:59:44 finance grade section of that pyramid. 1:59:46 And it's because I wanted to spend some 1:59:47 time on it by itself and give you 1:59:48 benchmarks. Now this is LTGP to CAC. 1:59:52 Similar to what we talked about before 1:59:53 when we had CAC and I said that we 1:59:54 needed a pairing metric to be able to 1:59:56 contextualize it. This is 1:59:57 contextualizing those two numbers. So 1:59:59 we're taking the gross profit on a 2:00:00 customer and we're dividing by the cost 2:00:02 to acquire that customer. And this will 2:00:03 give us an integer. And so as an 2:00:05 example, if we make $100 on a customer, 2:00:07 it costs us $33 to actually get that 2:00:09 customer. This would be 100 divided by 2:00:11 33, which is 3.33 LTGP to CAC. Now, 2:00:15 where we need to be really careful with 2:00:16 this metric is that once again, just 2:00:18 having an unrestricted time on lifetime 2:00:21 value is a terrible idea cuz this can be 2:00:23 measured across 6 years and so we end up 2:00:25 way overspending, but we don't actually 2:00:26 realize this cash for like four years 2:00:28 into the future. And so you want to put 2:00:29 a time constraint here. Now generally 2:00:31 what we will do is we will measure this 2:00:32 across two time horizons. We'll measure 2:00:34 it across 30-day LTGP to CAC and we will 2:00:37 do 90day LTGP to CAC and then sometimes 2:00:39 we'll do 365 days as well depending on 2:00:42 how aggressive the acquisition strategy 2:00:43 is. Now as benchmarks here what we want 2:00:46 to be seeing on either of these metrics 2:00:49 depending on how aggressive the 2:00:50 acquisition is in the business is that 2:00:52 there is multiple different levels that 2:00:53 you can be at. So you can be at sub one, 2:00:55 you can be at 1 to two, you can be at 2 2:00:57 to three, you can be at 3 plus. If 2:00:59 you're under a one, this is generally 2:01:01 speaking a terrible position to be in 2:01:03 because it means that you are actually 2:01:04 losing money on acquisition. You are 2:01:06 losing profit because you're paying more 2:01:08 to acquire a customer than the gross 2:01:10 profit on first purchase that you're 2:01:12 making or within the first 90 days. You 2:01:13 have to have incredible retention after 2:01:16 this point to be able to support losses 2:01:18 on acquisition. Most brands that I audit 2:01:20 that are losing on acquisition do not 2:01:22 have good enough retention to support 2:01:24 it. So this typically is not an actual 2:01:26 acquisition strategy most of the time. 2:01:29 This is poor efficiency on acquisition. 2:01:32 So this is simply a fact that your ads 2:01:33 in your acquisition funnel isn't good 2:01:35 enough and this is a bad position to be 2:01:36 in. Some people 0.01% of people can get 2:01:39 away with this and they have the finance 2:01:41 capability and the modeling to be able 2:01:42 to actually uh operate at this scale and 2:01:44 they have really good LTV. Most people 2:01:46 can't do this. You do not want to be 2:01:47 below a one on 30-day or 90day. It is 2:01:49 worth noting that being below a one 2:01:51 isn't necessarily just a CAC issue. It 2:01:53 isn't necessarily just poor efficiency 2:01:55 on acquiring, but it could be poor uh 2:01:57 gross profit. So, if you only have like, 2:01:59 let's say, less than $70 in gross profit 2:02:00 on first order, that's probably a 2:02:02 problem. You just don't have enough 2:02:03 gross profit to be able to substantiate 2:02:05 acquisition at scale on a paid platform. 2:02:07 We then have one to two. This also isn't 2:02:10 a great position to be in for most 2:02:11 people. Now, if you're in CPG and you 2:02:14 have really good retention dynamics, and 2:02:16 I'll give you some benchmarks on that 2:02:17 later on, then you can operate in this 2:02:19 area. But for most people with mediocre 2:02:22 retention, that's okay. And if you think 2:02:23 your retention is good, it's usually 2:02:25 mediocre. You will know if you have 2:02:26 excellent retention because the numbers 2:02:28 become very evident and you can push 2:02:29 acquisition like crazy. So one to a two. 2:02:31 You also generally don't want to be 2:02:33 here. This is generally not good. Dash 2:02:35 average. If you're really wanting 2:02:36 aggressive growth, if you're financing 2:02:38 hard, if you have investors and you need 2:02:40 to just uh throttle up revenue, sure you 2:02:42 can scale on this, but it's not ideal. 2 2:02:44 to three, this is optimal. This is where 2:02:45 you want to be. This is a great zone. 2:02:47 you should be scaling up budgets. Now, 2:02:49 to give you a little bit of context 2:02:50 here, what does a two elig look like? 2:02:53 Well, if you have $100 in gross profit 2:02:55 on first purchase, that's simply a CAC 2:02:57 of 50. And so, take whatever your gross 2:02:59 profit number is on first order, divide 2:03:01 by two, and that's what your CAC would 2:03:03 need to be to hit this optimal range, to 2:03:05 hit the bottom end of the optimal range. 2:03:06 Now, greater than a three, this is 2:03:08 actually also a big mistake. Being less 2:03:10 than a one is is a mistake. Being 2:03:11 greater than a three is a mistake 2:03:12 because this means you're just leaving 2:03:14 money on the table. you could scale very 2:03:16 aggressively here, be unbelievably 2:03:18 profitable, and you're probably 2:03:19 underleveraging paid media or whatever 2:03:22 marketing channels you're using to drive 2:03:23 this efficiency. Now, let's add two 2:03:25 quick notes here before we move on. Note 2:03:26 number one regarding attribution and 2:03:28 incrementality and everything that we 2:03:30 discussed on the pyramid. Uh when it 2:03:32 comes to attribution, what you want to 2:03:34 make sure of is that you're not using 2:03:36 view through conversions in the 2:03:37 platforms. And so when you're looking at 2:03:39 meta specifically, you want to be making 2:03:41 sure that you're using 7-day click as 2:03:43 your optimization or your reporting. 2:03:45 This is going to give you much tighter 2:03:47 congruency to acquisition me. In fact, 2:03:49 in most businesses, when you look at 2:03:51 their acquisition me and then you look 2:03:53 at their 7-day click rorowaz on meta, it 2:03:55 is super correlated and that's because 2:03:57 it doesn't include all these viewrough 2:03:59 conversions that is overattributing in 2:04:01 the platform. For those that don't know 2:04:02 what a view through conversion is, it's 2:04:03 when a user sees your ad on Facebook, 2:04:05 doesn't click, but then buys within 24 2:04:07 hours, Meta can claim the conversion. 2:04:09 And a lot of those people are going to 2:04:10 buy anyway. They're existing customers. 2:04:12 List goes on. And so, you want to be not 2:04:13 including them in your reporting. You 2:04:15 also don't want to be including existing 2:04:16 customers. As I said before, existing 2:04:18 customers aren't that incremental on the 2:04:20 platform, and so you don't want to be 2:04:22 overspending here. You want to go to 2:04:23 breakdown audience segments and look at 2:04:25 where your spend's going. Often, people 2:04:26 are putting way too much spend to 2:04:27 existing customers, and their frequency 2:04:29 is way too high. So you want to pull out 2:04:30 the frequency column and you want to 2:04:32 make sure that over the last 7 days it's 2:04:34 under an eight. Any higher than an eight 2:04:36 and you're definitely overspending 2:04:37 because you're serving to existing 2:04:38 customers more than eight times a month 2:04:40 which is not incremental. So there's a 2:04:42 big existing customer trap on the 2:04:43 platforms. The platforms always want to 2:04:45 spend here because they know they can 2:04:46 overattribute and they always want to 2:04:48 expand their attribution windows because 2:04:49 they know they can overattribute and if 2:04:51 they can attribute more revenue you will 2:04:52 spend more. Now, as a bit of a formula 2:04:54 here, cuz I get this question a lot, 2:04:56 even from 700, $800 million brands. I 2:04:59 get this question when when we come in 2:05:00 and talk to them and have calls with 2:05:01 them, which is what percentage should we 2:05:03 be allocating to existing customers? 2:05:05 You're saying that existing customers 2:05:07 aren't that incremental. Well, so then 2:05:08 what percent should we be allocating of 2:05:09 our budget? Should it be 20%, 30%, 40%? 2:05:11 Well, because 80% of our revenue as a 2:05:14 large 9 figure brand is coming from 2:05:15 existing customers cuz we have almost 2:05:16 full market saturation in Australia. So, 2:05:19 what are we doing? The question isn't 2:05:20 percentage. Thinking about percentage 2:05:22 allocation of ad spend to existing 2:05:23 customers is just a bad way to look at 2:05:26 it. What we instead want to look at is 2:05:28 total amount of existing customers. So 2:05:30 how many existing customers do we have? 2:05:32 How many times do we want to show them 2:05:34 an ad? And then from there we can 2:05:36 calculate how much we need to spend 2:05:37 because as long as we know the CPM we 2:05:40 can get the total spend per month. So 2:05:41 I'll give you the formula and I'll give 2:05:43 you an example which is let's say you 2:05:45 have 300,000 existing customers and you 2:05:48 want to serve how many ads to them? 2:05:49 Let's say you want to serve three times 2:05:51 a month. Then you just times by your 2:05:52 CPM. So go into the platform, do an 2:05:55 audience segment breakdown, look at 2:05:56 what's your CPM on existing customers, 2:05:58 and let's say it is $2.50. Now, that's 2:06:01 super low. It's probably going to be a 2:06:02 lot higher than that, but let's just say 2:06:04 that for the sake of this example. That 2:06:05 means that you're going to have to spend 2:06:07 $2,250 2:06:09 per month on existing customers to hit 2:06:11 them three times. Now, you can go and 2:06:12 change these variables and that will 2:06:14 change the outcome and tell you how much 2:06:15 to spend. Reality is you need to spend 2:06:17 way less than you actually think. Most 2:06:18 people think, "Oh, we're a massive 2:06:20 brand. We have multiple millions, if not 2:06:22 tens of millions of customers. We need 2:06:23 to spend hundreds of thousands a month 2:06:25 targeting them." You typically don't. 2:06:26 You could hit them with a 3 to six 2:06:27 frequency and you could just spend 10, 2:06:29 maybe on the upper side, $30, $40,000 a 2:06:32 month, and you're completely fine. I see 2:06:33 tiny businesses spending $40,000 a month 2:06:35 on existing customers. And so, this is 2:06:37 the math. This is what you want to be 2:06:38 doing. Let's move on to the final topic 2:06:40 here on metrics that matter, which is 2:06:42 the profit frontier. as a founder or a 2:06:44 head of digital or a head of marketing, 2:06:45 the question you need to be able to 2:06:47 answer is, if we were to spend an extra 2:06:50 $10,000 next month, where would we put 2:06:52 that budget? Where would we get the best 2:06:54 incremental impact of that media spend? 2:06:56 Most people don't know that answer 2:06:58 reliably enough. They don't know 2:07:00 actually where they should be allocating 2:07:02 their spend and therefore they get into 2:07:03 a position where they continue to drive 2:07:05 spend up across platforms, but they 2:07:07 don't see incremental returns. And so 2:07:09 how you need to be thinking through this 2:07:10 problem is what's called the next best 2:07:13 dollar. And so we want to have all of 2:07:15 our platforms, let's say Meta, Google, 2:07:18 Tik Tok, add in a bunch of other 2:07:19 channels if you're spending on them. 2:07:21 Ideally, you shouldn't be spending on a 2:07:22 bunch of tertiary channels, but let's 2:07:23 say you are. And then if we just take 2:07:25 spend here, we want to take Meta up to 2:07:27 the level of spend in which we stop 2:07:30 getting returns that we couldn't get on 2:07:31 the other platform. So let's say this is 2:07:33 the current allocation of spend. Okay? 2:07:35 Maybe we're like 100,000 here, 50,000 2:07:37 here, 70,000 here. What we want to know 2:07:39 is if we went and put an extra 10,000 2:07:41 into Tik Tok off the top here, what 2:07:43 revenue return would we get of this 2:07:45 $10,000? Same thing for Google, same 2:07:47 thing for Meta. And this is what we're 2:07:49 constantly trying to solve for because 2:07:50 ultimately to grow the business, we need 2:07:52 to continue allocating more marketing 2:07:54 dollars across the current marketing 2:07:55 channels. And we need to do it in the 2:07:57 most efficient way possible. The only 2:07:58 way to really understand this is to 2:08:00 unfortunately run incrementality tests. 2:08:02 We would want to for example go and put 2:08:03 10 into meta but do it in a controlled 2:08:05 incrementality test where we can get a 2:08:07 read on what this is. Same thing for 2:08:09 Google, same thing for Tik Tok. Now we 2:08:10 can also just do this intuitively over 2:08:12 time. You could use MM as well, some 2:08:14 marketing mix models to be able to 2:08:16 identify where you should be putting 2:08:17 media spend at a lower revenue threshold 2:08:19 like seven figure brands. This should be 2:08:20 relatively intuitive if you're a good 2:08:22 performance marketer. But this is 2:08:23 ultimately the problem that you should 2:08:24 be solving for. And you should also be 2:08:26 thinking about this inverse. And so, 2:08:28 could we pull 10,000 out of Google, 2:08:30 reallocate it to Meta, and get a better 2:08:32 ROI? Maybe this 10,000 here is only 2:08:35 driving us a 2x, but we could go and put 2:08:37 another 10,000 into Meta, and it would 2:08:39 give us a 4x on new customer revenue. 2:08:42 And so, we should be rebalancing budgets 2:08:44 and reallocating and scaling across here 2:08:46 based on incremental returns and the 2:08:48 profit frontier. And so, the profit 2:08:50 frontier is that you want to go all the 2:08:52 way up to the point in which you're 2:08:53 making no more profit past this nominal 2:08:55 dollar in spend. And so you want to find 2:08:57 the point in which your acquisition me 2:09:00 becomes the break even point. And once 2:09:02 you hit that break even point 2:09:03 incrementally, that's where you stop 2:09:04 spending. And you're doing that across 2:09:06 all channels at all times. And only 2:09:07 until you've reached the profit frontier 2:09:09 on all of these three primary channels 2:09:11 do you go and move on to adding in 2:09:12 additional channels. So cash flow verse 2:09:14 profit. Profit doesn't equal cash. And 2:09:17 that's ultimately why paid media within 2:09:19 e-commerce is way more complex than in a 2:09:21 service-based business or than in a SAS 2:09:24 business or particularly in the info 2:09:26 space. And it's because when you scale 2:09:28 up, you need to commit cash to future 2:09:31 inventory purchasing which substantially 2:09:33 restricts the actual dividends that 2:09:35 could be yielded within the business. 2:09:36 And so if you see an e-commerce business 2:09:38 doing 100k in profit, reality is 2:09:40 founders probably taking no money. even 2:09:42 a million in profit, $2 million in 2:09:44 profit, respective to the total revenue 2:09:46 and growth rate, there may be actually 2:09:48 no profit available at the end of the 2:09:49 day cuz it all gets reinvested into 2:09:51 future inventory buying. So let's say 2:09:53 that the P&L here shows $2 million in 2:09:56 profit per year, but the balance sheet 2:09:58 has $3 million on it. So the balance 2:10:00 sheet will show you the assets of the 2:10:02 business. The assets in e-commerce is 2:10:04 typically unsold inventory. Now there 2:10:07 could be buildings on here if they own 2:10:08 the office, if they own the warehouse, 2:10:10 etc. But typically for most businesses, 2:10:12 we're just talking about inventory 2:10:13 sitting on the balance sheet as well as 2:10:15 cash sitting on the balance sheet. So 2:10:17 the reality of this business right here 2:10:19 is that they are not in a good position. 2:10:22 The brand actually has negative cash 2:10:24 flow depending on how we're looking this 2:10:26 across time and how the cash is actually 2:10:27 moving. But if they made $2 million in 2:10:29 profit, but $3 million in inventory 2:10:31 didn't sell, well, they actually made no 2:10:33 money. They would actually be cash flow 2:10:35 negative. they would be down a million 2:10:37 dollars in cash because yes, they made 2 2:10:38 mil, but they bought 3 million in 2:10:40 inventory and it never moved. Not good. 2:10:41 So, this is where we get into the 2:10:42 inventory death spiral or what can also 2:10:45 be called as skew rationalization. So, 2:10:48 all of your inventory has a grade 2:10:51 associated to it. And most people are 2:10:53 probably aware of this, but the 2:10:54 marketers and the performance marketers 2:10:56 watching this aren't. And so, this will 2:10:58 become incredibly helpful in making paid 2:11:00 media decisions and tying this into how 2:11:02 we should be approaching the ad account. 2:11:03 Grade A is fastm moving inventory. This 2:11:07 is inventory that is selling quickly 2:11:09 that we honestly don't need to worry 2:11:11 about. We need to just be thinking about 2:11:12 how do we not go out of stock. Grade B 2:11:14 inventory is medium sell to rate. This 2:11:17 is slowm moving sell to rate and then 2:11:20 this is not moving at all. Now Shopify 2:11:23 actually auto ranks your inventory 2:11:24 anyway as long as the inventory is 2:11:26 within Shopify. You can actually see 2:11:28 this yourself. I think it's under the 2:11:29 products tab. Now the key here really is 2:11:31 that people will do product launches 2:11:33 because product launches are one of the 2:11:35 best levers, one of the best ways to 2:11:36 scale an ecom brand. Ultimately you have 2:11:38 brands like Grunes, like AG1, like IM8, 2:11:42 all of these brands have gotten 2:11:44 enormous, hundreds of millions of 2:11:45 dollars, close to a billion dollars with 2:11:47 just one product. But that is misleading 2:11:50 you in terms of how most e-commerce 2:11:51 brands need to grow because those brands 2:11:53 have done so well because they're in the 2:11:55 supplement space. And in the supplement 2:11:56 space, you have a very unique advantage, 2:11:58 which is that you can reposition your 2:11:59 product into technically like different 2:12:01 products. So, if you're selling a 2:12:02 multivitamin, you can say that that 2:12:04 multivitamin helps with gut health, but 2:12:06 it also helps with hair health, but it 2:12:07 also helps with all these other 2:12:08 problems. And so, because of that, you 2:12:09 have a really large total addressable 2:12:11 market that you can reposition the 2:12:12 product into. With most brands, you 2:12:14 cannot reposition your product enough to 2:12:16 gain a large total addressable market. 2:12:18 So, you need to do future product 2:12:19 launches. Now, when you do that, what 2:12:21 ends up happening is they don't work. A 2:12:23 lot of product launches fail. And so 2:12:25 they either land into grade B, grade C, 2:12:28 or you get an initial pop from organic 2:12:30 and existing customers and you can sell 2:12:32 through maybe the first 30 40% of 2:12:33 inventory and then it doesn't move at 2:12:35 all. And so then you get stuck in a 2:12:37 situation which is that every time 2:12:38 you're trying to grow through product 2:12:39 expansion, you were just adding more 2:12:41 cash onto the balance sheet that is 2:12:43 restricting the cash position of the 2:12:45 business and the ability for you to 2:12:46 scale. This is where skew 2:12:47 rationalization becomes incredibly 2:12:49 important, which is that as new products 2:12:50 are introduced into the product suite, 2:12:53 you need to be thinking about how can we 2:12:54 rationalize our skew count down always 2:12:57 to be able to decrease the amount of 2:12:58 available products, to be able to 2:12:59 decrease the amount of inventory on 2:13:00 hand. There also needs to be some 2:13:03 congruency here between the marketing 2:13:05 team and the finance team. And this is 2:13:06 where an enormous mistake is made, not 2:13:08 only on agency side, but internally a 2:13:10 lot of the time, which is that let's 2:13:12 play through an example of this, which 2:13:13 is that you go and you launch a new 2:13:15 product. it doesn't do very well. Okay, 2:13:17 it works on the organic list. You get a 2:13:19 bit of a pop out of it. Maybe you do 2:13:20 100k, but you ordered a million dollars 2:13:22 worth in sell value. And then because of 2:13:24 that, it ends up over here in grade D or 2:13:26 grade C. It's barely moving. You look at 2:13:28 the sell through rate and you go, we 2:13:29 currently have 400 days of inventory on 2:13:32 hand, which means it's going to take 400 2:13:34 days for us to sell through all this 2:13:35 product, which isn't good. Okay, we want 2:13:36 to move this back to cash as soon as 2:13:38 possible. Now, what will normally happen 2:13:39 is the CFO or the owner or the manager 2:13:41 or whatever it is, will look at this and 2:13:43 go, "Ah, that's annoying. 2:13:45 and then try to solution it themselves. 2:13:47 But a lot of the time, we can actually 2:13:48 just solve this through paid media. 2:13:50 Okay, we can take this product that 2:13:52 didn't do too well and we can start 2:13:54 pushing spend behind it on paid media at 2:13:56 break even or maybe even a slight loss, 2:13:58 dedicate isolate it into its own 2:14:00 adfunnel. So, it doesn't even have to 2:14:01 sit on the website. It can sit 2:14:03 separately. We can push it through 2:14:04 whitelisting pages. What we've also done 2:14:05 for a client is we've actually taken 2:14:07 their grade D inventory and we've pushed 2:14:08 it into a new country that we knew would 2:14:10 perform well. we could discount heavily, 2:14:12 wouldn't erode the brand equity within 2:14:14 the primary market, and then we can just 2:14:16 move all this stock using paid and turn 2:14:17 it back into cash. Now, people don't 2:14:19 have this conversation enough with the 2:14:21 paid media team because the paid media 2:14:23 team, the agency, is always KPI on 2:14:24 profit, and this can be really 2:14:26 detrimental. If we're trying to just 2:14:27 maximize the profit position of the 2:14:28 business, what ends up happening is all 2:14:30 of our spend will actually go up here 2:14:31 and we'll end up really messing the 2:14:33 business up because they'll have all 2:14:34 this grade BCD inventory that doesn't 2:14:36 actually get prioritized in paid media 2:14:37 at all cuz it doesn't have a good ROI, a 2:14:39 good row, a good efficiency. We need to 2:14:41 push grade BCD inventory in paid to be 2:14:43 able to turn it back into cash. This is 2:14:45 where you do need an understanding of 2:14:46 the inventory position of the business 2:14:48 so that we can make better decisions on 2:14:50 paid for the overall health of the 2:14:52 business. All right, we can't talk about 2:14:54 finance without talking about cash 2:14:55 conversion cycles. The actual formula 2:14:57 for cash conversion cycle is DIIO plus 2:14:59 DSO minus DPO. So days in inventory, 2:15:02 days sales outstanding, days of payable 2:15:04 outstanding. Now this is a 20-minute 2:15:06 video on itself. In fact, we've actually 2:15:07 put out a lot of content on cash 2:15:09 conversion cycles, how to calculate it, 2:15:10 how to measure it, what's good, what's 2:15:12 bad, etc. So instead of going too deep 2:15:14 here, I instead just want to explain the 2:15:16 top level concept so that if you're in 2:15:18 marketing, you can understand how this 2:15:20 actually relates into the business and 2:15:21 where you might need to be understanding 2:15:23 of these components. If you actually own 2:15:24 the business or you're a CFO, I 2:15:26 recommend watching our other videos that 2:15:27 go into a little bit more detail on this 2:15:29 topic. But the fundamental idea here is 2:15:30 that if you sell 10 units of product and 2:15:33 you make $50 in gross profit here, and 2:15:36 once again, I'm just making numbers up, 2:15:37 but this will make sense. What ends up 2:15:38 happening here is because you only have 2:15:40 $50 left in your hand, you can only then 2:15:42 go and buy 11 units on your next 2:15:45 purchase order. And then you'll make $55 2:15:47 in gross profit. And then this will 2:15:49 allow you to maybe buy 12 units on the 2:15:51 next order. And then this goes on and on 2:15:53 and on. And so the limiter in this 2:15:54 business is not the actual selling of 2:15:57 the product. It isn't marketing. It 2:15:58 isn't ROI on Facebook ads. It's the fact 2:16:01 that they're just cash limited in their 2:16:03 ability to buy future inventory because 2:16:05 the amount of gross profit that is 2:16:06 generated off this purchasing only 2:16:08 allows a slight increase in future 2:16:09 purchasing. Now, what you need to also 2:16:11 add in here is let's say that there is a 2:16:13 90-day lead time from putting the 2:16:15 inventory in to actually arriving at 2:16:17 your warehouse to when you can start 2:16:18 selling. And let's assume that you have 2:16:19 very poor terms with the manufacturer 2:16:21 cuz these POS are so small. So, you 2:16:23 actually have to put the money down 90 2:16:25 days prior to the actual inventory 2:16:27 coming. then you're in a terrible 2:16:28 position because you have to buy these 2:16:30 11 units after you sell these units to 2:16:33 get the cash. And so you actually then 2:16:34 have to go out of stock for 90 days here 2:16:36 cuz once you have this cash, then you 2:16:38 can buy the units, then you can actually 2:16:39 go and sell again. And so what you would 2:16:41 want to do here in this example is you 2:16:43 would want to negotiate on better 2:16:45 supplier terms so you aren't out of 2:16:46 stock for 90 days. And so you might 2:16:48 actually negotiate that you only need to 2:16:50 pay 20% up front here. And so after you 2:16:52 sell two units, you have enough money to 2:16:53 put in your next order. And then as 2:16:55 you're selling through these other 2:16:56 units, you're going through this 90-day 2:16:57 period. Then once these actually get 2:16:59 shipped, you pay the rest. And so maybe 2:17:01 the shipping time is 20 days. You're 2:17:02 actually only out of stock for 20 days 2:17:03 rather than 90. Now, another thing that 2:17:05 you could do is you could reduce the 2:17:07 lead time from the manufacturer. So 2:17:10 rather than it taking 90 days, you could 2:17:11 try get a different manufacturer that 2:17:13 can reduce this down to maybe 30. And 2:17:15 then that fixes the whole problem in 2:17:16 itself. Or the last one is you use uh 2:17:19 some kind of inventory financing. And so 2:17:22 you actually take a loan out to pay for 2:17:25 this inventory so that you can sell 2:17:26 through, get this profit. This profit 2:17:28 pays down the loan as you're going. And 2:17:29 then this inventory comes. You then 2:17:31 start selling this inventory. You open 2:17:33 up another loan and this loan pays for 2:17:34 the next purchase. And so you can get 2:17:36 ahead by opening up some kind of 2:17:37 financing. Now the issue is that 2:17:39 financing is very expensive. When 2:17:41 companies are loaning, you'll pretty 2:17:42 much always have to put yourself up as 2:17:43 collateral as well. And it's super 2:17:44 risky. anytime you're introducing a lot 2:17:46 of debt into the business, you're 2:17:47 banking on the fact that you could sell 2:17:49 these future units at an efficiency 2:17:50 that's going to allow you to pay down 2:17:52 the loan. And so, if you are agency side 2:17:54 and you're working with like seven 2:17:55 figure businesses, they're almost always 2:17:57 using some kind of financing to be able 2:17:59 to actually grow. Um, and you need to 2:18:00 take into consideration that rapid 2:18:02 growth for these types of businesses 2:18:04 usually actually isn't possible. The 2:18:05 limiter isn't on how quick we can 2:18:07 increase ad spend or how efficient the 2:18:09 ads are. The limiter is the cash 2:18:10 conversion cycle. And so I have actually 2:18:12 seen single-handedly probably five times 2:18:14 now seven figure businesses that have 2:18:16 gone to an agency. The agency has 2:18:18 absolutely crushed it. They've ramp 2:18:19 spend from $20,000 a month to $200,000 a 2:18:22 month. They haven't actually had any 2:18:23 visibility into any of this with the 2:18:26 client. The client in the background to 2:18:28 fuel this growth has gone and taken 2:18:30 millions of dollars out in inventory 2:18:32 financing to be able to fuel the growth 2:18:34 to be able to buy all the future 2:18:35 inventory required to hit that kind of 2:18:37 tripling per year in growth rate. And 2:18:39 then as they've done that, their 2:18:40 interest repayments get so large and the 2:18:43 ads start becoming inefficient to where 2:18:44 net margins get squeezed down to zero. 2:18:46 And now, yes, we've taken a business 2:18:48 from doing $2 million a year to $7 2:18:50 million a year. Like, amazing. Good job, 2:18:52 agency. But the business has gone from 2:18:54 cash flow positive, founders actually 2:18:56 taking an income to the business is near 2:18:58 zero on net profit after all the 2:19:00 interest repayments and the compression 2:19:02 in marketing efficiency as they've 2:19:04 achieved that additional scale. And so 2:19:06 this is why this is so important for at 2:19:07 least people to be across so that they 2:19:09 can understand that you just like can't 2:19:10 triple an ecom business in a year. You 2:19:12 just can't because the cash requirements 2:19:13 on inventory are just so enormous. And 2:19:15 so you either need a lot of 2:19:17 self-funding. So the person starting the 2:19:19 business needs to have multi-millions 2:19:21 liquid. They need funding from someone 2:19:23 who's providing the multi-millions or 2:19:24 they need to go into some kind of 2:19:25 financing. Otherwise, you just can't 2:19:27 scale inventory orders fast enough to 2:19:29 keep up with the scale that you want to 2:19:31 achieve within the business. And a lot 2:19:33 of the time where this gets a little bit 2:19:34 dangerous is that uh people will pull 2:19:36 marketing efficiency down to achieve 2:19:38 larger scale but to do that it yes it 2:19:42 accelerates the growth rate but it also 2:19:45 increases the amount of debt that the 2:19:46 brand needs to take on which actually 2:19:48 compresses margins down to zero. And so 2:19:50 you would actually be better off growing 2:19:51 slightly slower at a slightly lower me. 2:19:55 So a better efficiency, less marketing 2:19:56 spend, having less interest repayments 2:19:58 and making way more money rather than 2:20:00 just trying to arbitrarily accelerate 2:20:02 your growth rate but increasing interest 2:20:04 payments and making marketing efficiency 2:20:06 worse. So that's just where you need to 2:20:07 be very very careful in your financial 2:20:08 modeling. Now this is where ultimately 2:20:10 there starts to become a CFO verse CMO 2:20:14 disconnect in a lot of businesses in 2:20:16 this core decision which is let's run 2:20:17 through an example. February arrives and 2:20:20 inventory from November is still unsold. 2:20:23 Maybe there was some new products in 2:20:24 here that didn't do so well and they've 2:20:25 fallen into grade C and D inventory. 2:20:27 Now, most CFOs will actually look at 2:20:30 this and go, "Okay, we need to cut media 2:20:33 spend because we actually don't have 2:20:35 enough cash right now to be able to fuel 2:20:37 the current marketing spend in Feb 2:20:39 through to April." And so, let's go and 2:20:41 take our marketing expenses and cut them 2:20:43 by 40%. And now this is quite logical 2:20:46 when you're just opening up and looking 2:20:47 at the P&L because when you look at the 2:20:49 P&L uh cost of delivery might be 30%, 2:20:51 marketing might be 25% and then opex is 2:20:54 15 and so profit in this business is at 2:20:57 30%. Now when revenue suddenly decreases 2:21:00 in Feb and we actually don't have a lot 2:21:02 of cash available me might go up and 2:21:04 spike to 30% which starts compressing 2:21:07 profitability down and profitability 2:21:09 dips to 25%. Now naturally the reaction 2:21:12 from the CFO should be this has gotten 2:21:14 out of control. ME has accelerated up. 2:21:17 We need to cut marketing spend. 2:21:18 Marketing team cut your budgets. Now 2:21:20 there's two issues here. Number one is 2:21:22 that in this business it might very much 2:21:24 so be the fact that the marketing is 2:21:26 driving the revenue. And so this 2:21:28 compression in marketing efficiency is 2:21:30 an efficiency issue on the current media 2:21:32 spend. We shouldn't be cutting media 2:21:34 spend because if we cut media spend, 2:21:36 revenue will dip further. And so you can 2:21:38 get into these uh cyclical situations 2:21:40 where marketing spend drops, revenue 2:21:42 drops, profitability decreases. To fix 2:21:44 profitability, marketing spend needs to 2:21:45 drop even more. And you keep dropping 2:21:47 marketing spend to try to fix the issue, 2:21:48 but it doesn't fix it. The other issue 2:21:50 is that to fix this grade CN inventory 2:21:52 issue for November, what we actually 2:21:54 need to do is quite counterintuitive. We 2:21:55 need to increase marketing spend even 2:21:57 further because we need to sell out of 2:21:59 the CND inventory and it's not going to 2:22:01 be profitable to do so. And so what we 2:22:03 actually need to do here is we need to 2:22:05 take me that's now inflated to 30% and 2:22:08 we need to say hey you know we actually 2:22:09 need to move this in the short term to 2:22:11 35%. And this additional 5% will be 2:22:14 purely for driving all the CND inventory 2:22:17 so that we can turn this back into cash 2:22:19 which is going to give us a more a 2:22:20 better position from a cash perspective 2:22:22 which is going to allow us to get back 2:22:24 into the old revenue position. So let's 2:22:26 then put it all together. First place to 2:22:28 start is in forecasting. Forecasting is 2:22:30 critical for anything financial related 2:22:32 because it allows us to set our KPIs and 2:22:35 our expectations that we're then 2:22:36 measuring off on an ongoing basis to be 2:22:38 able to call square. Anytime you are 2:22:39 forecasting, the number one rule of 2:22:42 forecasting in e-commerce direct to 2:22:43 consumer is that you always need to 2:22:45 forecast new customer revenue separate 2:22:48 to returning customer revenue. And the 2:22:50 reason being is that both of these 2:22:51 revenue buckets have different 2:22:52 underlying levers that impact the 2:22:54 realization of the revenue. So for new 2:22:56 customers, this is primarily going to be 2:22:58 marketing. Now, this might be just 2:23:00 advertising spend for some brands. This 2:23:01 might be advertising spend plus events 2:23:03 and influencers. This might advertising 2:23:05 plus influencers plus other tertiary 2:23:07 channels as well. When it comes to 2:23:08 returning customer revenue, yes, 2:23:10 advertising spend is going to drive a 2:23:12 little bit of returning customer 2:23:13 revenue. Yes, an influencer activation 2:23:14 might drive a little bit of returning 2:23:16 customer revenue, but primarily 2:23:17 returning customers is going to come 2:23:19 from uh product launches. It's going to 2:23:21 come from marketing events like discount 2:23:23 periods and it's going to come from 2:23:25 other activations that are going to prop 2:23:27 up and give returning customers a reason 2:23:30 to come back. There is also obviously 2:23:32 direct communication channels in here as 2:23:34 well like email and SMS. These are 2:23:37 different inputs compared to over here. 2:23:40 Therefore, when we are forecasting, what 2:23:42 are we doing? We're actually just 2:23:43 forecasting inputs and therefore getting 2:23:46 an output which is the forecast or the 2:23:48 budget. And so we need to look backwards 2:23:50 into the inputs to be able to then 2:23:51 determine how we're actually getting the 2:23:53 final number. For returning customers, 2:23:55 you want to look at all of your 2:23:56 returning customer cohort. So every 2:23:58 month back to let's go Jan 2021 all the 2:24:01 way to Feb 2021. And this goes on and on 2:24:04 and on all the way to today. There was a 2:24:05 certain amount of new customers that you 2:24:07 acquired. Maybe back here it was 100, 2:24:08 then it was 110, then it was 150. All of 2:24:11 these customers repeat at a certain rate 2:24:13 over time. And you can see this in your 2:24:15 cohort analysis. And so people might 2:24:16 come back at 7% in the first month, then 2:24:19 6%, then 4%, then two, then one, and 2:24:22 then ultimately it asotopes down to 2:24:24 usually a pretty small number. What you 2:24:26 can then do is you can take all of these 2:24:28 old cohorts and the amount of customers 2:24:29 that you acquired then, and you can go, 2:24:31 okay, well, they're about to enter into 2:24:33 month 32 of them being a customer. What 2:24:35 is the repeat rate on average of someone 2:24:37 after month 32? And maybe it is 0.1%. So 2:24:40 then you take your 100, you times by 2:24:42 0.1%. You then times by whatever your 2:24:45 average order value is on returning 2:24:47 customers, which let's say it's $100. 2:24:49 And so we would expect this cohort to 2:24:51 give us $1 in returning customer revenue 2:24:54 next month. And then we do that for the 2:24:55 next cohort and the next cohort and the 2:24:56 next cohort and every single cohort that 2:24:59 we've had in the past. And this will 2:25:00 then give us a realistic extrapolation 2:25:02 of what we should expect returning 2:25:04 customers to contribute to next month. 2:25:06 This number ends up being usually within 2:25:08 about 10% accuracy. To get it within 1 2:25:10 to 2%, you do two things. Number one, 2:25:12 you bake on seasonality. And so you look 2:25:15 at uh average seasonality in the last 3 2:25:17 years across the calendar year and you 2:25:19 just apply a factor based on returning 2:25:21 customer revenue seasonality. Number 2:25:23 one. Number two is you then go and 2:25:24 superimpose marketing events that are 2:25:26 going to substantially change these 2:25:28 numbers. So obviously if you have a 2:25:29 major marketing event that's going to 2:25:31 occur this year that didn't occur last 2:25:32 year, there's going to be revenue driven 2:25:34 from that. you want to figure out what 2:25:35 is your uh expected revenue realization 2:25:37 and then you add that into the forecast. 2:25:39 Then on new customer revenue now in paid 2:25:42 ads this is relatively straightforward 2:25:43 when you have larger media makes or if 2:25:45 you have other channels like influencers 2:25:47 etc. This is where you have to build 2:25:48 your own modeling around this and this 2:25:50 is obviously where modeling and finance 2:25:52 and data science becomes pretty critical 2:25:54 once you once you achieve scale as an 8 2:25:56 to9 figure brand. To keep it simple for 2:25:58 paid ads, what you want to do is you 2:25:59 want to take your acquisition me on the 2:26:01 y- axis. You want to take spend on the 2:26:04 x- axis and you want to plot the last 2:26:06 let's for the sake of this video say 1 2:26:08 year of daily data. What you'll see is 2:26:10 that every day there is a certain 2:26:11 acquisition me that is associated to 2:26:14 that day. And what should end up 2:26:15 happening is you should have something 2:26:17 like this. Now if you go and put a 2:26:19 logarithmic regression or a linear 2:26:20 regression whatever has the best fit for 2:26:22 the model you will get an average of the 2:26:24 expected efficiency at a certain spend 2:26:26 level. Now you will have outliers like 2:26:28 these outliers over here. These are 2:26:29 typically sales periods. So we want to 2:26:31 actually remove these from the data set. 2:26:33 We also know that if we're only going to 2:26:35 be spending a minimum of $1,000 a month, 2:26:37 we could also just remove anything under 2:26:39 $1,000 a month out of the model too. And 2:26:41 that's going to improve the accuracy of 2:26:43 modeling on these higherend spends. Then 2:26:44 what you're also going to have is all of 2:26:45 these over here are going to be November 2:26:47 and Black Friday periods that are 2:26:49 significantly overinflating the actual 2:26:51 efficiency that you would expect during 2:26:53 BAU. So you want to go and remove all 2:26:55 these two. And then what you'll get is 2:26:56 this line through all of the days last 2:26:58 year at certain spend at certain 2:27:00 efficiency. And then you know okay if we 2:27:01 go and spend $8,000 a day here we know 2:27:04 roughly what efficiency we should 2:27:06 expect. And this is how you model out 2:27:08 spend and new customer revenue 2:27:09 expectations. You would actually back 2:27:11 propagate from your new customer goal. 2:27:13 So, if you wanted to get like, let's 2:27:14 say, $100,000 in new customer revenue, 2:27:17 uh, and let's say you want to do it at a 2:27:18 4 a.m., you would just go to a four on 2:27:21 the graph, which might be here. You 2:27:22 would go across and you go, "Okay, it's 2:27:24 right here. Can we spend enough to get a 2:27:26 4 AM and this much revenue?" And the 2:27:28 answer might be no. Okay. Well, we 2:27:29 fundamentally have an issue here. We 2:27:30 need to rethink what we're going to do 2:27:32 differently this year to generate 2:27:34 outsized returns compared to the average 2:27:36 of last year. Are we going to do some 2:27:37 kind of marketing event? Is there going 2:27:38 to be a new product launch? Is there 2:27:39 going to be a new channel? Are we 2:27:40 tripling creative production? like what 2:27:42 actual input or lever is going to 2:27:44 generate this outcome. And this is 2:27:45 ultimately the exercise of forecasting. 2:27:46 The exercise of forecasting is to look 2:27:48 at the realistic expectation against the 2:27:50 target. There's going to be a delta. 2:27:51 There's always going to be a delta. The 2:27:52 board wants you to hit 50 million. You 2:27:54 do this and you go, the math says we can 2:27:56 only hit 40 million. And then you look 2:27:58 at, well, what are the inputs required 2:27:59 to achieve that $10 million difference? 2:28:01 We talked about the profit frontier. We 2:28:03 talked about how funnels overattribute 2:28:05 at the bottom, underattribute at the 2:28:06 top. But I want to reinforce this idea 2:28:08 cuz it's a really common mistake in 2:28:10 budget allocation, which is that you 2:28:11 will always see the best rorowaz at the 2:28:13 bottom, you will always see the worst 2:28:15 rorowaz at the top. And so if you have a 2:28:18 meta campaign that's attributing at a 2:28:20 2x, and you have a Google campaign 2:28:22 that's attributing at a 6x, don't simply 2:28:24 go and put more budget here. This is 2:28:26 going to be overattributing because it 2:28:27 sits at the bottom of funnel. It might 2:28:28 have branded key terms. It might be 2:28:30 retargeting. Even if you have a bunch of 2:28:31 exclusions in place, probably bottom of 2:28:33 funnel, and you're actually going to see 2:28:34 better incremental impact putting 2:28:36 budgets here. Obviously, always do this 2:28:37 within a controlled test. Make sure that 2:28:39 you're on top of whether this is 2:28:40 actually true or not for your particular 2:28:42 business. But be very careful with 2:28:44 rorowaz reads in the platform because 2:28:45 it's going to usually make you 2:28:47 overallocate to bottom of funnel efforts 2:28:48 and then you'll wonder why you're not 2:28:50 growing when you're increasing budgets. 2:28:51 one thing that we haven't touched on at 2:28:53 all here. And this starts to get a 2:28:54 little bit outside of finance, but I 2:28:57 think it's a really important mention, 2:28:59 which is brand at 1 million to I would 2:29:02 actually argue probably 20 million. You 2:29:04 can just brute force revenue through 2:29:06 performance marketing. Okay, performance 2:29:07 marketing through ads can get you here 2:29:09 very easily. You put $1 in, you get $4 2:29:12 out, and then you continue to scale. But 2:29:14 eventually what usually ends up 2:29:15 happening is you hit diminishing 2:29:17 returns, which is that as you try to put 2:29:19 more spend into these platforms, as you 2:29:20 try to start pushing past $30,000 a day, 2:29:23 $40,000 a day in ad spend, you just 2:29:25 can't get any further. And the way to go 2:29:26 further is through some kind of brand 2:29:28 effect. And I say this from personal 2:29:30 experience myself. Obviously, in the 2:29:31 early days, we worked with three 500 7 2:29:34 figure brands either full-time or in 2:29:36 some kind of consulting capacity. And 2:29:37 these days, we currently work with over 2:29:39 60, eight, and nine figure brands and 2:29:41 close to 10ig brands as well. And so we 2:29:44 have seen both sides of the spectrum. 2:29:45 All these small brands that are very 2:29:46 reliant on performance marketing and all 2:29:48 of these large retail brands that aren't 2:29:49 reliant on performance marketing at all. 2:29:51 And in fact, there's actually a huge 2:29:52 opportunity in performance marketing cuz 2:29:54 they don't do it very well. And this is 2:29:55 due to the brand that they have within 2:29:57 the platform. You can open up and I do 2:29:59 this all the time. You can open up one 2:30:00 fashion ad account that's doing maybe $8 2:30:02 million a year and you can look at all 2:30:04 the core metrics in the Facebook ad 2:30:05 account, the click-through rates, the 2:30:06 CPMs, the CPCs, the conversion rate. And 2:30:08 then I can go and open up another 2:30:10 fashion ad account of a business doing 2:30:11 $300 million a year. And the crazy thing 2:30:14 is that the ad account over here has 2:30:16 better metrics on everything. They have 2:30:18 better CPMs. They have better 2:30:19 clickthrough rates. They have better 2:30:20 CPCs. They have better rows. Everything 2:30:22 is better. And you go, how is that even 2:30:23 possible? They're doing like 20x the 2:30:26 volume. They're doing 20x the ad spend. 2:30:28 That just doesn't make sense because as 2:30:29 you scale paid media, you hit 2:30:31 diminishing returns. So why are they not 2:30:33 seeing all their numbers degrade? And 2:30:34 it's because of this overarching brand 2:30:36 effect that they have in the market. 2:30:37 They have so much market saturation. 2:30:39 They have so many associations that have 2:30:40 been built through external marketing 2:30:42 efforts that sit outside of the ad 2:30:44 account that inside the ad account it 2:30:45 looks really good, but it's because of 2:30:46 everything that they're doing outside of 2:30:48 the ad account that makes it look good. 2:30:49 And so I can go in any day on a brand 2:30:51 that every single person knows and run 2:30:52 ads and I'll have incredible 2:30:54 clickthrough rates cuz everyone knows 2:30:55 who they are. But if I go in on a brand 2:30:56 that no one knows who they are and I'm 2:30:57 trying to push, obviously the 2:30:58 performance marketing has to be a lot 2:31:00 better and that's why you hit 2:31:00 diminishing returns. And so also just 2:31:02 when you're thinking about marketing 2:31:03 expense allocation, I would always be 2:31:06 having some kind of budget towards 2:31:08 branding efforts. And by branding 2:31:10 efforts, it's making associations within 2:31:12 the market that is going to put you in 2:31:14 front of the customer where they are. 2:31:15 And so if your customers are commonly in 2:31:17 a particular area or at a particular 2:31:19 event or looking at particular things, 2:31:21 that's where you want to show up to be 2:31:23 able to build associations. I'll give 2:31:24 you two personal examples of this, which 2:31:26 is that I've recently bought running 2:31:28 gear from two different brands. I bought 2:31:30 from 2xU which is an Australian brand. 2:31:32 It's actually a client of ours and then 2:31:34 another brand which is 247 represent. 2:31:38 And the reason I bought from this brand 2:31:40 was because all of the running 2:31:43 influencers that I follow, all the 2:31:44 people that I watch YouTube videos of 2:31:46 every week, all the people that I follow 2:31:47 on Instagram, they are all either 2:31:49 sponsored by 247 or they just wear it. 2:31:51 They make associations with it. And so 2:31:53 because of that, that natural 2:31:54 association that's been made within 2:31:56 market of where I end up showing up on 2:31:58 the content that I consume is the reason 2:31:59 why I bought it. Had nothing to do with 2:32:01 quality, had nothing to do with anything 2:32:02 except for the fact that I follow all 2:32:04 these running influencers, ended up 2:32:05 following the founder, following his 2:32:07 story, watching podcasts of him, and 2:32:08 that's ultimately what pushed me to the 2:32:10 purchase. I would argue that all of 2:32:12 those uh influencer deals, all of the 2:32:14 branding exercises, them showing up to 2:32:16 run clubs, etc., that probably doesn't 2:32:18 have direct profitable ROI. They're 2:32:20 probably not getting the coupon code 2:32:21 that the influencer has at checkout, 2:32:23 which I don't think they even do, but 2:32:24 let's say they did do it. Probably not a 2:32:26 profitable exchange, but it's the 2:32:27 overarching branding effect of making 2:32:29 those associations that ends up pushing 2:32:31 tons of people to purchase. On 2XU, it's 2:32:33 the branding of premium. Now, 2XU's 2:32:35 products are incredibly premium. I think 2:32:37 they're probably one of the highest 2:32:39 quality products in Australia in this 2:32:40 market. But honestly, I don't think that 2:32:42 even matters for me in terms of my 2:32:43 purchasing decision. I didn't purchase 2:32:45 because I knew the product was quality 2:32:47 cuz I bought online. I hadn't seen it. I 2:32:49 bought because of the perception of 2:32:51 quality and so it is the brand 2:32:52 perception that they have built that 2:32:54 this is the highest quality uh 2:32:55 activewear clothing in Australia that is 2:32:58 causing me to buy. Now once again was 2:32:59 this through some kind of performance 2:33:01 marketing ad? No. Was this through a 2:33:02 Google ad? No. It was through the 2:33:04 overarching associations that they make. 2:33:07 It's through the messaging that they 2:33:08 have and it's through the way that they 2:33:09 show up in the creative as well 2:33:11 particularly in the campaign shoots that 2:33:13 makes the perception that it is super 2:33:15 high quality which ultimately drove me 2:33:17 towards that conversion. And so both of 2:33:18 these purchases likely wouldn't have 2:33:20 happened through any kind of performance 2:33:22 marketing effort. They actually occurred 2:33:23 through brand which is why that we can't 2:33:24 understate this and we need to have it 2:33:26 as a portion of the video because it is 2:33:27 unbelievably important particularly as 2:33:29 you continue to scale and it should be 2:33:31 thought through as a budget allocation 2:33:33 of an expense on the P&L. So wrapping 2:33:35 this up, if there are five things that 2:33:37 you should be walking away with as key 2:33:39 takeaways to take forward in your 2:33:41 business or working with a client, it is 2:33:44 number one, know your gross margin and 2:33:46 know how to calculate it correctly. You 2:33:49 need to understand variable costs. You 2:33:51 need to understand the difference 2:33:52 between product margin and gross margin. 2:33:53 You need to understand how that also 2:33:55 then flows through into contribution 2:33:56 margin. Number two is you need to 2:33:58 understand the definitions and you need 2:34:00 to have live dashboards that track 2:34:02 acquisition me profit contribution 2:34:04 ideally LTP to CAC not just being over 2:34:07 here tracking rorowaz on a day-to-day 2:34:09 basis and having weekly rorowaz reports 2:34:11 this is not productive at all for core 2:34:12 decision-m and moving the business 2:34:14 forward the third is that you want to be 2:34:16 separating all new versus returning 2:34:20 customer metrics you want to be tracking 2:34:23 new customer economics acquisition me 2:34:25 new customer profit contribution new 2:34:27 customer revenue, new customer cohort 2:34:28 size separate from returning because 2:34:31 ultimately the levers underlying them 2:34:33 are different. This also obviously 2:34:34 applies into what we were just talking 2:34:36 about around forecasting. Number four, 2:34:39 you want to understand the difference 2:34:41 between a cash verse a P&L play. It 2:34:45 might very much so be the case within 2:34:46 the business that the current limiter 2:34:48 isn't marketing spend or marketing 2:34:50 efficiency, but it's the cash conversion 2:34:51 cycle. And so there is no point in 2:34:53 arbitrarily pushing budgets up and 2:34:55 trying to scale if it's just going to 2:34:56 cause an increase in interest expenses 2:34:58 on the P&L and a compression in me. 2:35:00 There also needs to be constant 2:35:01 communication between either the 2:35:03 internal marketing team or you and the 2:35:04 agency as to the inventory position 2:35:07 within the business across the different 2:35:08 SKUs so that there can be strategies 2:35:10 employed to be able to actually decrease 2:35:13 profitability, decrease acquisition me, 2:35:15 decrease efficiency, but prioritize the 2:35:18 turnover of inventory into cash to make 2:35:21 the business overall healthier. And then 2:35:23 number five is that you want to be using 2:35:26 the P&L and all of these other financial 2:35:28 tools to be able to identify the 2:35:32 constraint in the business. And so when 2:35:34 you can understand how to read the P&L 2:35:36 and structure it out, you can understand 2:35:38 how to KPI at each level. And then when 2:35:40 you start falling below KPI, you can 2:35:42 look above that level in the P&L to 2:35:44 understand, okay, what has changed? What 2:35:47 levers are there? And then how can we 2:35:49 pull on those levers to rectify and 2:35:51 course correct back to where the target 2:35:53 actually is. If you made it this far, 2:35:55 thanks for watching for an hour and a 2:35:56 half. And if you are an e-commerce brand 2:35:59 doing over $5 million a year, there'll 2:36:00 be a link somewhere in the bio to reach 2:36:02 there'll be a link somewhere below in 2:36:04 the description to reach out and get a 2:36:05 free audit from ourselves where we'll 2:36:07 run you through all of this financial 2:36:08 modeling, but we'll actually apply it to 2:36:10 your business. And if you're a 2:36:11 performance marketer that's gotten this 2:36:12 far, please reach out. We're always 2:36:13 hiring for a play of performance 2:36:14 marketers. Click on the website, reach 2:36:16 out to us somehow. You could also email 2:36:17 hiring bluesdigital.com.au 2:36:20 and we'll look at your application. Most 2:36:21 brands think they have an ad problem, 2:36:23 but they actually have an offer problem. 2:36:24 And most of the brands that think they 2:36:26 have an offer problem actually have a 2:36:27 discounting problem. So, in the next 90 2:36:29 minutes, we're going to run you through 2:36:31 the why as to why offers matter so much 2:36:33 in e-commerce. What an offer actually is 2:36:36 if we break it down into its fundamental 2:36:38 components. We'll then go through the 2:36:39 economics of crafting an offer that 2:36:41 works for you. We'll go through the five 2:36:43 mechanics of an offer in e-commerce. 2:36:46 I'll then show you the right offer. How 2:36:47 do you make an offer that actually fits 2:36:48 your particular brand? Because it 2:36:50 changes niche to niche. Then we'll go 2:36:52 into offer discipline. Lastly, I'll give 2:36:54 you the seven offer failures that you 2:36:57 don't want to do. And then we'll tie it 2:36:58 all together so you know exactly how you 2:37:00 can improve your offer in your store 2:37:02 right now to see better performance 2:37:03 across the entire funnel. Now, the 2:37:05 reason why offers are so important is 2:37:08 because they are the single biggest 2:37:09 change that a DTOC operator can make to 2:37:12 have the largest inflection upwards in 2:37:14 performance with practically zero cost. 2:37:16 We're not saying go and make more 2:37:17 creative. We're not saying go and design 2:37:19 an entire new website. We're saying if 2:37:21 you just change the offer, the way that 2:37:23 the value is presented to the customer, 2:37:25 you can generate significantly higher 2:37:27 conversion rates, higher average order 2:37:28 value on the exact same cost to acquire 2:37:30 a customer. Most marketing calendars for 2:37:32 8 to 9 figure brands are just 52 weeks 2:37:35 of weekly promotions. And the result 2:37:37 ends up being that the customer is just 2:37:39 trained for discounts and they think 2:37:41 that they have offers sorted because 2:37:42 they're rotating in new offers every 2:37:44 week. But that's the complete wrong 2:37:45 definition of how you should be thinking 2:37:46 about this problem. A 25% discount on a 2:37:49 $100 average order value erodess gross 2:37:52 margin from $60 down to $35, which is 2:37:55 nearly a 50% reduction. Now, even if 2:37:58 average order value lifts by 20%, you're 2:38:00 still in a worse position. So, let me 2:38:02 show you what an offer actually is. 2:38:04 Because if you can't explain exactly how 2:38:06 an offer works, you definitely can't 2:38:08 design one. So, this is the hierarchy of 2:38:10 influences to how you can think about 2:38:12 all of these different variables as they 2:38:15 pertain to performance within the 2:38:16 business. So, when you zoom out and look 2:38:18 at a direct to consumer or retail 2:38:19 business, the least important thing, but 2:38:22 it builds the base that allows 2:38:23 everything else is the budget. 2:38:25 Ultimately, if you have more budget, you 2:38:27 will sell more product. If you have less 2:38:29 budget, you will sell less product. And 2:38:30 so, the budget that you actually put 2:38:32 into the platforms is effectively the 2:38:34 fuel on the fire that allows everything 2:38:35 else to go. Now, if everything else is 2:38:37 bad, it won't do much. Which is why the 2:38:40 next step is account structure. Now, you 2:38:41 can put all the budget that you want 2:38:43 into the funnel, but if your account 2:38:45 structure is set up in a way that's just 2:38:46 retargeting the same people over and 2:38:48 over again, or it's pushing the wrong 2:38:49 products, or it's got a myriad of issues 2:38:51 with it, then it will let down 2:38:53 everything above it. But once again, 2:38:55 these two things are at the base of 2:38:57 importance. These actually aren't the 2:38:59 important things that are going to drive 2:39:00 substantial delta in the business. Then 2:39:02 we move up and we go to creative. 2:39:03 Substantially more important than 2:39:05 account structure and budget. This is 2:39:06 how people actually hear about us. Then 2:39:08 we get to the offer. This is what 2:39:10 they're actually getting and why they 2:39:12 should purchase right now. Then the 2:39:14 brand is why people should buy from you 2:39:16 rather than a competitor. And then the 2:39:18 product is why people buy it all. And 2:39:20 this becomes incredibly important at 2:39:21 super bottom a funnel in repeat 2:39:23 purchasing. So this is really the 2:39:24 hierarchy of importance when it comes to 2:39:27 creating a purchase decision within the 2:39:29 consumer. The higher up the pyramid, the 2:39:31 bigger the lever, but the harder it is 2:39:33 to change. And because it's hard to 2:39:35 change everything up here, most teams 2:39:37 just spend all their time down here 2:39:39 changing account structure and budgets 2:39:40 around thinking it's going to do 2:39:41 something. The top three is actually 2:39:43 where massive delta becomes unlocked 2:39:45 within performance. An offer doesn't 2:39:48 just mean a discount. That's not what 2:39:49 we're talking about here. In fact, you 2:39:50 don't have to discount at all and you 2:39:52 can have a unique offer. And we'll give 2:39:53 you a bunch of examples of that later 2:39:55 on. A discount is a percentage off or a 2:39:58 coupon code or a store-wide markdown. 2:40:00 Whereas an offer is product positioning. 2:40:03 So, what problem does this product 2:40:04 actually solve for who and why right 2:40:07 now? Number two is the architecture of 2:40:10 the price. So, it's not just are we 2:40:12 doing a percentage off, but we have more 2:40:14 optionality here. Are we doing some kind 2:40:15 of threshold offer? Are we doing a units 2:40:17 per transaction offer? Do we have a 2:40:19 bundle structure where we're discounting 2:40:22 by 20% but we're forcing units per 2:40:24 transaction to be up at three to four so 2:40:26 we actually don't take any gross margin 2:40:28 compression. Number three is the value 2:40:30 mechanic. Ultimately when it comes to an 2:40:32 offer what we are doing is we are 2:40:33 creating a price discrepancy within the 2:40:35 market. And this, in my opinion, is the 2:40:37 best way to think through this problem 2:40:39 of creating value and generating an 2:40:42 offer that converts people at higher 2:40:44 rates, which is that from the consumer's 2:40:46 perspective, the value of the offer is 2:40:49 the perception of the gap between the 2:40:52 cost of goods sold for the business and 2:40:55 the value that they're actually getting. 2:40:57 That's why a blank 50% off or 60% off 2:40:59 discount works so well, because a 2:41:01 consumer sees that discount and they go, 2:41:03 "That's got to be a cost of goods sold. 2:41:05 this business can't even be making money 2:41:07 on selling that. Hence, that is a 2:41:09 fantastic deal because no one is making 2:41:11 money in this exchange. So, I am getting 2:41:13 all the value arbitrage. However, from 2:41:16 the business's perspective, a good offer 2:41:18 is the opposite. It is how do we get 2:41:21 people to spend as much money as 2:41:23 possible above cost of goods sold whilst 2:41:25 making them think that they're paying 2:41:27 cost of goods sold. And so, a consumer 2:41:29 wants to know that they're getting a 2:41:30 good deal. They want to think they're 2:41:32 getting all of this value and it's only 2:41:34 costing them this price. Wow, they must 2:41:35 not be making much money. But then the 2:41:37 advertiser or the business selling the 2:41:40 product is actually making a ton of 2:41:42 money. But it's the way that they have 2:41:43 reorientated value within the offer that 2:41:46 is making the consumer think it's a good 2:41:48 deal. And sometimes it is just a good 2:41:50 deal for both. Like sometimes the 2:41:52 company might own a software company and 2:41:54 therefore they can give free licensing 2:41:56 deals away to some software with the 2:41:58 product. And so for the consumer, it is 2:42:00 an incredible deal because maybe they 2:42:01 get to save on their tech stack when 2:42:02 they get a software. And from the 2:42:04 business's perspective, it's also a 2:42:06 great deal because they own the software 2:42:07 anyway and it costs them absolutely 2:42:09 nothing to deliver it to the customers 2:42:10 as an add-on. And so it's great for them 2:42:12 too. And so the value can obviously 2:42:15 coexist for both the advertiser and the 2:42:17 consumer at the same time. And that's 2:42:19 what creates an incredibly valuable 2:42:21 offer. So on the value mechanic, this is 2:42:23 how we're creating the perception of 2:42:25 value and effectively collapsing the 2:42:27 consumer towards the perceived cost of 2:42:29 goods price. So this could be free 2:42:31 shipping, free gift, buy one get one, 2:42:33 samples, second order tier, a 2:42:36 subscription unlock. There's a lot of 2:42:37 different mechanics here that allow us 2:42:39 to increase perceived value. The next 2:42:41 one is risk reversals. This exists in 2:42:43 every industry. It's not ecom specific. 2:42:45 This is where you're going to reverse 2:42:46 risk by having a returns or a guarantee 2:42:48 in place. Maybe there's a trial period. 2:42:50 Maybe it's a sample offer, so you're 2:42:52 just buying the sample and that's it. 2:42:53 And then there's urgency, which has 2:42:54 existed for as long as time in 2:42:56 marketing, which is that you want 2:42:57 scarcity. If it's a drop, if it's a 2:42:59 seasonality tie-in, if it's just a 2:43:01 window in which this product's available 2:43:03 before it goes out of stock, this is 2:43:05 your fifth lever to be thinking about. 2:43:06 So discounting, technically, sure, is an 2:43:10 offer, but it's only one of five options 2:43:14 in terms of how we actually create value 2:43:16 within the offer. It's only really 2:43:19 correlated with price architecture. 2:43:21 Everything else, how we position the 2:43:23 product, the value mechanic, the risk 2:43:25 reversal, the urgency, all of this 2:43:27 should also be used in conjunction with 2:43:29 the discount or the discount doesn't 2:43:30 even have to be applied at all so that 2:43:32 we can create the best offer possible. 2:43:34 Now, some of the best offers that exist 2:43:37 do not use flat discounting. Flat 2:43:39 discounting is really easy math to run. 2:43:41 It also substantially impacts uh your 2:43:43 perception in market of where your price 2:43:45 is anchored. And so if you end up flat 2:43:46 discounting too much, well guess what? 2:43:48 All the customers get trained that 2:43:49 you're on flat discount and then you 2:43:51 become a discount orientated brand and 2:43:52 now your margin permanently compresses. 2:43:54 You launch the business thinking you 2:43:55 were at 60% GP. Now suddenly you're at 2:43:57 45% because you always have to discount 2:43:59 because it's the only way that people 2:44:00 come back and buy again. And so before 2:44:01 you go into any kind of offer curation, 2:44:04 it's really important that you 2:44:05 understand the economics of discounting. 2:44:07 But let's run through a base example. 2:44:08 We've got a $100 average order value. 2:44:10 We've got a cost of goods of $40. For 2:44:13 the sake of this, let's just assume cost 2:44:14 of goods has all variable expenses, 2:44:16 shipping and fulfillment, transaction 2:44:17 fees, etc. Let's just assume it's all in 2:44:19 there. Gross margin or gross profit is 2:44:21 therefore $60. Our break even rorowaz is 2:44:26 1 divided by our gross margin. Now, our 2:44:29 gross margin here is 60%. Therefore, our 2:44:31 break even rorowaz is a 1.67. Now, our 2:44:35 target CPA at a six rorowaz is average 2:44:38 order value divided by 6. So 100 / 6 = 2:44:42 1667. 2:44:44 And therefore at this target our 2:44:47 contribution profit per order is 60 - 16 2:44:51 which is $43.33. 2:44:53 So this is the brand at full price. 2:44:55 Every single brand should know this 2:44:57 exact waterfall for every single product 2:44:59 in the business. Then what we need to do 2:45:01 is model out the offer. Now we'll start 2:45:04 with a very simple offer which is not 2:45:06 what I would run. I think it's pretty 2:45:08 bad, but 25% off sale. So, what's going 2:45:10 to happen here? Well, average order 2:45:12 value, or at least let's call it price 2:45:14 per unit for the moment, and you'll 2:45:16 understand why we're changing this up in 2:45:18 a second. Price per unit compresses to 2:45:20 $75. Now, here's what people sometimes 2:45:23 don't take into consideration when they 2:45:24 do this modeling, which is that when you 2:45:26 do a discount, units per transaction 2:45:28 increases. So, units per transaction is 2:45:31 the amount of units in the cart in the 2:45:33 order. Let's say that in this case on 2:45:36 average there was one unit per cart 2:45:38 which is unlikely pretty much everyone 2:45:40 has above an average of one because 2:45:42 someone is going to add two things to 2:45:43 cart and it's going to pull this up. For 2:45:45 the sake of simplicity let's say that 2:45:46 this is at a one. Now once you go on 2:45:48 sale people buy more because people want 2:45:50 to capitalize on the sale and so units 2:45:53 per transaction might actually go up 2:45:56 substantially. Maybe you never go on 2:45:58 sale and so it rises all the way to a 2:46:00 1.63. Now what that means is that 2:46:03 average order value which is price per 2:46:06 unit or average unit retail times by 2:46:08 units per transaction equals $122. So 2:46:12 actually where we thought discounting 2:46:13 would compress average order value, it 2:46:15 actually increased it. We actually get 2:46:17 more money on each customer. However, 2:46:19 our cost of goods sold is still 40% but 2:46:22 it's not 40% of this average order 2:46:24 value. It's 40% of the retail value 2:46:27 which is $162. 2:46:29 Right? This is $162 2:46:32 retail price prior to us discounting by 2:46:35 25%. So our cost of goods is this times 2:46:37 40% which is $65 which equals a gross 2:46:41 margin of $57. Our break even rorowaz 2:46:45 therefore increases to 2.22. And then if 2:46:47 we want to hit the same contribution 2:46:50 profit per order that we were outside of 2:46:52 the sale. The way that we calculate this 2:46:54 is we take the gross margin, we minus 2:46:56 off our target contribution margin, and 2:46:59 that gives us $13 2:47:01 as a target CAC or cost to acquire a 2:47:04 customer. If you want to convert this to 2:47:06 return on ad spend, you just take 2:47:08 average order value and you divide by 2:47:10 CAC, which is 122 divided by 13.67, 2:47:13 which is an 8.97. So coming up on a 9. 2:47:16 Now, the rorowaz previously was a six. 2:47:19 So, we need to increase return on ad 2:47:21 spend by 50%. Okay, we need to go from a 2:47:25 six efficiency to a 9 efficiency on a 2:47:29 25% sale off campaign. And this is why 2:47:32 understanding the economics of an offer 2:47:33 is so important because on the surface 2:47:36 when you just run the topline numbers, 2:47:37 it looks really good. Okay, if we run 2:47:39 this discount and units per transaction 2:47:40 go up, we're going to make way more 2:47:42 money on each customer. We're going to 2:47:43 make $122 rather than 100. This is 2:47:45 great. The sale is working in our favor. 2:47:47 We're doing more revenue per customer 2:47:49 because of the sale. But then once we 2:47:51 start to waterfall down through gross 2:47:53 margin, which is actually lower. So even 2:47:55 though we're doing more revenue here, 2:47:56 we're doing less gross margin. So gross 2:47:58 margin as a percentage is compressed 2:48:00 enormously. Then once we go to the break 2:48:01 even row, it lifts quite a lot. Then we 2:48:03 go to the CPA and it kind of gets 2:48:04 ridiculous. And then we go to the 2:48:06 rorowaz and we go, wait a second, for us 2:48:08 to hold the same level of profitability, 2:48:10 we need to be 50% more efficient. Is 2:48:13 that going to realistically happen? Now 2:48:15 for some people, yes. I know a ton of 2:48:17 clients that come to the top of my head 2:48:19 where we actually model out 70 to 80% 2:48:21 improvements in efficiency during sales 2:48:23 periods because they never go on sale 2:48:24 and they're a luxury brand. And so when 2:48:26 they do, the spike in sales is enormous. 2:48:29 I have other brands that I can think of 2:48:31 where we would model in a 5% improvement 2:48:34 in efficiency because everyone's just so 2:48:36 used to them discounting that it doesn't 2:48:38 matter to them and they have so much 2:48:39 market saturation and penetration into 2:48:41 the Australian market being known as a 2:48:43 discount brand that it doesn't matter if 2:48:45 they discount doesn't really lift sales 2:48:46 that much materially and so you really 2:48:48 need to understand how your efficiency 2:48:51 reacts to different levels of sales so 2:48:54 that you can model out what you believe 2:48:55 efficiency lift will be to understand 2:48:58 whether the offer is actually going to 2:48:59 work for you or not. Now, I'm going to 2:49:00 get ahead of myself a little bit and I'm 2:49:02 going to run you through a way better 2:49:03 offer and all of the economic mechanics 2:49:06 behind it so you can understand why 2:49:08 offer curation becomes so important. 2:49:10 Now, a lot of what I'm about to throw 2:49:12 onto the board, I haven't even spoken 2:49:13 about yet. It's further in the video in 2:49:15 terms of agenda, but I'm just going to 2:49:17 throw the whole kitchen sink at an offer 2:49:19 here so you can get an understanding of 2:49:21 how you can go from a 25% offer to 2:49:23 instead something crazy with so much 2:49:25 free stuff with so much layered in and 2:49:27 you actually have better gross margin. 2:49:28 So, let's actually contextualize this 2:49:30 business and let's call it a supplement 2:49:32 business. Here's the new offer. So, the 2:49:34 offer now is buy a 90-day supply, get 2:49:37 25% off. You get free access to the 2:49:40 brand's app where they have meal plans 2:49:42 for you, dedicated meal plans every 2:49:43 single day. On top of that, you get a 2:49:44 free mystery gift. Don't know what it 2:49:46 is, but you get a mystery gift. And you 2:49:47 also get this 25% off discount for life 2:49:51 as long as you stay on subscription. And 2:49:52 normally, the subscription discount is 2:49:54 10%. And so you're getting double the 2:49:56 normal subscription discount as long as 2:49:57 you stay on subscription and you don't 2:49:59 cancel. So what does that do to the 2:50:00 mechanics? Well, on average order value, 2:50:02 it's going to increase average order 2:50:04 value by about 2 1/2x. Reason being is 2:50:06 there's already a bit of take rate on 2:50:07 the 90-day offer up front. So this isn't 2:50:10 the 90-day offer. 30-day might be about 2:50:12 $80. 90-day is $21. 2:50:16 And so there's a little bit of take rate 2:50:18 here, which is pulling the average order 2:50:19 value up, but it isn't great. Now, 2:50:20 average order value jumps to about $190 2:50:24 because the take rate on this shoots 2:50:26 through the roof. Now, how is the actual 2:50:28 margin profile impacted? Well, cost of 2:50:30 goods is 40% and retail price here is 2:50:33 253. Spibble are getting a massive 2:50:35 discount. And so, cost of goods is 101 2:50:38 based on 40% cost of goods. However, 2:50:41 cost of goods isn't 40%. And the reason 2:50:43 being is that when we ship three of the 2:50:45 product, we actually get economies of 2:50:47 scale in shipping and fulfillment. So 2:50:49 where on shipping one unit it costs us 2:50:51 $12 to ship the package. When we ship 2:50:54 two units it only costs us 14. And when 2:50:56 we ship three units it only costs us 17. 2:50:59 Now we have the unit cost baked into the 2:51:01 price as it scales. And so we are 2:51:03 technically charging onto the customer 2:51:06 3x this amount baked into the price the 2:51:08 base price of the product. But we don't 2:51:10 have to pay this amount times three. We 2:51:11 actually get a quite a large saving 2:51:13 here. We've baked in $36 into our 2:51:15 pricing for shipping. it's only costing 2:51:17 us 17, which means we actually have $19 2:51:21 in savings we've just pulled together. 2:51:23 So, this cost of goods or cost of 2:51:25 delivery actually drops to 82, meaning 2:51:28 our gross margin is now $108. Now, if we 2:51:31 want the same contribution margin per 2:51:33 order, what do we do? We just minus this 2:51:35 off. To get this contribution profit per 2:51:37 order, we need a cost to acquire of 65. 2:51:40 190 / 65 equals 2.9. 2:51:44 So to make the same contribution margin 2:51:46 per order now we only need a 2.9 rorowaz 2:51:49 rather than a six. Our efficiency can 2:51:51 have and this is ultimately where 2:51:53 evergreen acquisition offers like this 2:51:56 crush. This isn't an offer that you 2:51:57 rotate in on Black Friday. This is an 2:52:00 offer you could run all year round 2:52:02 because it substantially compresses the 2:52:04 efficiency that you actually have to sit 2:52:05 at. Now obviously we probably don't want 2:52:06 to go down here. We just want to 2:52:08 increase contribution margin per order. 2:52:09 Rather than making $43 per order, why 2:52:11 don't we make 50? Why don't we make 60? 2:52:13 However, we did some other things in 2:52:14 this offer, too. We added a free mystery 2:52:16 gift. Now, we didn't actually include 2:52:18 that in COGS. The reason why we didn't 2:52:20 is it's inconsequential. You can put 2:52:22 free mystery gifts in that have a $2 to 2:52:25 $3 cost of goods. And you can get pretty 2:52:27 flexible here. You could just give 2:52:28 sample packs from other products. If 2:52:30 you're in supplements, um you could just 2:52:32 throw in something like a drink bottle 2:52:33 and as long as you make a very large PO, 2:52:36 the cost of goods on that drink bottle 2:52:37 throwing can be very low as long as it 2:52:39 fits within the package and doesn't 2:52:40 drive up your shipping and fulfillment 2:52:42 costs. There's a lot of stuff that you 2:52:43 can do on mystery gifts. You can do 2:52:44 mystery gifts that don't even have a 2:52:46 cost of goods associated with them, 2:52:48 right? And so you can do a partnership 2:52:50 or a collaboration with another brand in 2:52:53 which another brand actually gives you 2:52:55 the gift to give away and it's 2:52:56 co-arketing because the other brand is 2:52:58 like, if we can get some free samples in 2:53:00 your orders and you can frame it as a 2:53:02 free mystery gift, amazing because these 2:53:03 people might come and buy from us. And 2:53:05 so there is tons of stuff you can 2:53:06 actually do here to be able to have a $0 2:53:09 cost of goods or a1 to $3 on a mystery 2:53:12 gift. Then we have the lifetime 2:53:14 subscription discount. Now the objective 2:53:16 of this is to keep retention high and 2:53:20 keep them on quarterly billing. And so 2:53:22 what will happen is to get this 25% off 2:53:24 discount, they not only have to buy a 2:53:26 90-day supply, but they have to tick the 2:53:29 subscription box. If they don't tick 2:53:31 subscription, they don't get the offer. 2:53:32 Now because of that, it pulls people in 2:53:34 to the subscription. Number two, they 2:53:35 get a discount that they otherwise 2:53:36 wouldn't get if they subscribe at a 2:53:38 different point of the year. So, you can 2:53:39 put urgency around lifetime subscription 2:53:41 discount. And then what that does is 2:53:43 substantially increases LTV, 2:53:45 particularly on a 90-day basis because 2:53:47 on the 90-day is when all of the repeat 2:53:50 transactions occur and substantially 2:53:52 bumps up cohort lift. Not only can you 2:53:55 be more aggressive on acquisition here 2:53:56 to acquire more customers just on first 2:53:58 purchase contribution profit, but you 2:54:01 could probably be even more aggressive 2:54:03 because the actual retention on this 2:54:06 offer versus this offer is going to be 2:54:07 substantially different. You're probably 2:54:09 going to have double the amount of 2:54:10 repeat purchasing coming through from 2:54:12 this offer than you will over here. And 2:54:14 then you can continue to stack really 2:54:16 lowcost things into the offer here. What 2:54:18 you do have to weigh up is it can get to 2:54:20 the point where you're just stacking so 2:54:21 much stuff that it's it's just too much 2:54:22 for the consumer. Simple scales when it 2:54:24 comes to offers. Obviously, you want to 2:54:26 layer in as much meaningful value as you 2:54:29 can, but you also want to sub subtract 2:54:31 as much as possible to only give the 2:54:33 stuff that people care about. And so a 2:54:35 really valuable exercise after this 2:54:36 offer would be live for let's say 90 2:54:38 days, 120 days is to then survey 2:54:41 customers and get an idea of which of 2:54:43 these three additions in the offer did 2:54:45 they like the most that convince them to 2:54:48 buy. And you might find that no one 2:54:50 could care less about the mystery gift. 2:54:52 It means nothing. And therefore, let's 2:54:54 remove it and let's test adding 2:54:56 something else or let's just not have it 2:54:58 there at all because it just adds 2:54:59 additional complexity and let's see if 2:55:01 conversion rates are impacted. You might 2:55:02 find that the app really got a lot of 2:55:05 people to buy and this really pulled 2:55:06 them in. But then once they went onto 2:55:08 the app, they were like, "Oh, this is 2:55:09 useless. This is just a bunch of chatb 2:55:11 meal plans. I don't really care about 2:55:13 this at all." And it decreased brand 2:55:14 perception and impacted retention 2:55:17 revenue. And so even though this helped 2:55:18 a lot on the front end of the offer, it 2:55:20 actually substantially negatively impact 2:55:23 the back end of the offer. And so then 2:55:25 we make a change here. We either change 2:55:26 the product and the deliverable or we 2:55:28 pull it out of the offer entirely. And 2:55:29 so this is where you iterate on an offer 2:55:31 over time to improve the acquisition 2:55:33 economics. So you're making more profit 2:55:35 on first purchase, you're driving down 2:55:36 the cost to acquire a customer, but 2:55:38 you're also working to improve 2:55:41 retention, too. There's two curves that 2:55:42 you really need to understand here when 2:55:44 it comes to the economics of an offer, 2:55:45 which is that as you increase flat 2:55:47 discount rate across the offer, your 2:55:49 required return on ad spend to break 2:55:51 even increases linearly. And so your 2:55:53 break even return on ad spend at a 10% 2:55:56 might be 1.9 and then it goes up to 2 2:55:59 and then it goes up to 3 4 5 etc. And 2:56:01 then eventually obviously the break even 2:56:03 point becomes pretty much zero because 2:56:05 sorry return on ad spend reaches 2:56:07 infinity because there is a point in 2:56:08 which there's no margin left. This is 2:56:10 fine. I think most people have their 2:56:11 head around this. What they don't have 2:56:12 their head around is that target return 2:56:14 on ad spend increases exponentially. And 2:56:17 so if you have a target rorowaz goal 2:56:20 based on a contribution profit target. 2:56:23 So as we went through before in the 2:56:25 example, let's say that your goal is to 2:56:26 hit $40 in contribution profit. The 2:56:29 reality is is that margin compresses to 2:56:31 $40 much faster than it compresses to 2:56:33 $0. And so as a product of that, if you 2:56:36 were to graph this, and we have tooling 2:56:37 that allows us to do this. So reach out 2:56:39 to us if you're a brand and we'll just 2:56:40 send it through and you can model all 2:56:42 this yourself, which is that your return 2:56:43 on ad spend will actually look like 2:56:44 this. And so as you increase discounting 2:56:48 your required rorowaz to hold the same 2:56:50 contribution profit increases 2:56:52 exponentially which means that the 2:56:54 difference between like let's say a 15% 2:56:56 discount and a 25% discount can be 2:56:59 absolutely enormous in the rorowaz 2:57:02 requirement. This might require a 20 2:57:04 rorowaz. This might require an eight. 2:57:06 And it's like wow we're never going to 2:57:07 hit a 20. So any kind of discounting at 2:57:10 this range of the graph is just 2:57:12 inapplicable for us. We can't even do 2:57:13 it. This is another graph that's really 2:57:15 important to understand. And if you can 2:57:16 model this based on your own data, even 2:57:18 better. Now, we will do this modeling 2:57:20 sometimes for brands. It just depends on 2:57:21 how clean their data set is over time of 2:57:24 different flat percentage discounting so 2:57:26 that we can see historical units per 2:57:27 transaction change. So, what we're 2:57:29 really looking at here is as we move 2:57:31 across the x-axis, discount rate is 2:57:33 increasing. So, we're going from 10% to 2:57:35 20%, 30, 40, and then let's go 50 all 2:57:38 the way over here. On the y- axis, we 2:57:40 have contribution profit per order. Now, 2:57:42 what's going to happen is as you go and 2:57:45 discount by like 5 to 10%. You're 2:57:48 actually going to make less money. And 2:57:50 the reason why you'll make less money is 2:57:52 because this is such an unconvincing 2:57:53 offer, no one really cares about a 5% 2:57:56 discount or a 10% discount that it's not 2:57:58 going to increase people's units per 2:58:01 transaction. People aren't going to put 2:58:03 more items in cart because of a 5% 2:58:05 discount. It's meaningless to most 2:58:07 people. And so as a consequence of that, 2:58:09 your average unit retail or your unit 2:58:11 price will just compress by 5%. And 2:58:14 there'll be no upside. And so because of 2:58:16 that, your contribution profit actually 2:58:18 declines when you have these really 2:58:20 small discounts. Then at some point 2:58:22 there's an inflection and people start 2:58:24 caring about the discount size. And at 2:58:27 that point, your units per transaction 2:58:29 starts to increase. And so if we look at 2:58:31 average unit retail on this graph, this 2:58:34 is what average unit retail looks like 2:58:36 because you were just discounting the 2:58:37 prices of all your products. So the 2:58:39 price is going down, but the important 2:58:41 component becomes units per transaction. 2:58:43 Now units per transaction stays the same 2:58:45 here. And then there is a point in which 2:58:48 people start to care about the discount 2:58:50 and it goes up and then the reality is 2:58:51 as the discount gets more and more 2:58:53 aggressive, it continues to go up. But 2:58:56 if the compression in margin doesn't 2:58:58 outweigh the increase in units per 2:59:00 transaction, guess what? Total 2:59:02 contribution profit declines. And so it 2:59:04 is the function of units per transaction 2:59:06 and average unit retail that will 2:59:08 produce the contribution profit result. 2:59:10 You want to understand how aggressively 2:59:12 does up increase as we increase the 2:59:14 discount threshold. And then obviously 2:59:15 you just get to a point in discounting 2:59:17 past like 30 40%. in which it's just not 2:59:19 worth it for anyone because no matter 2:59:21 how much units per transaction 2:59:23 increases, you're making like no money 2:59:24 because you just compress margin down. 2:59:25 Now, why does this theoretical graph 2:59:27 matter at all? Well, because what you 2:59:29 want to ideally be doing when you're 2:59:30 thinking about what kind of discount are 2:59:32 we at least surfacing within the offer. 2:59:34 Now, it doesn't have to be a discount 2:59:35 offer, but what I showed you before was 2:59:36 a really well-curated offer that did 2:59:38 still have a percentage off that was 2:59:40 headlined. You want that percentage off 2:59:42 to sit where contribution profit per 2:59:44 order is maximized. At what point do we 2:59:47 give away some margin, but we maximize 2:59:49 the average order value lift? And it is 2:59:51 at this point that we want a discount. 2:59:52 And for this brand, this would be at 2:59:54 20%. Once again, you can model this out 2:59:55 yourself as long as you have some 2:59:57 historical data points on units per 3:00:00 transaction during different flat 3:00:02 discounting periods. And that just 3:00:03 depends on the consistency of flat 3:00:05 discounting offers that you've done over 3:00:06 time. So, let's run through the five 3:00:08 different offer mechanics. Number one is 3:00:09 product repositioning. What it changes 3:00:12 is the persona, the problem, or the 3:00:13 headline. It's the exact same skew. So 3:00:15 we're not actually changing the product 3:00:17 fundamentally. We're not doing any kind 3:00:18 of bundle, but we are repositioning the 3:00:20 product. The contribution margin impact 3:00:22 is nothing. In fact, sometimes it can 3:00:24 actually be positive. And it's best for 3:00:26 doing it on one skew or doing it on a 3:00:28 hero skew. So there's no better example 3:00:31 to do this in than on a supplements 3:00:33 brand. So let's go and put Groans. Now 3:00:35 what you could do, and what average 3:00:36 marketers would approach this brand 3:00:38 with, and for those that don't know, 3:00:39 they sell Daily Greens gummies. Average 3:00:41 marketers will position the product like 3:00:43 this. Daily Greens gummies get your 3:00:45 vitamins and that's the core angle. 3:00:47 That's the way that the product's 3:00:48 positioned. But instead, what Gruns does 3:00:50 very effectively is they actually 3:00:51 reposition the product into almost six 3:00:54 different products that serve six 3:00:57 different audiences and play to six 3:00:59 different mechanics or unique 3:01:01 mechanisms. Number one is gut health for 3:01:03 bloating. Number two is GLP1 support. 3:01:06 Number three is fiber for digestion. 3:01:09 Number four, hair health. Number five, 3:01:12 cognition dash focus. Number six, 3:01:16 multivitamin for business professionals. 3:01:18 And they have way more than this. What 3:01:20 we're really doing here is we're just 3:01:21 listing out different personas and 3:01:22 concepts that they're testing and the 3:01:24 core ones that work the best in the 3:01:25 account. But persona testing also flows 3:01:29 directly into the offer because we 3:01:31 create offers around high performing 3:01:33 personas and we can craft the product 3:01:35 title, the product position, every other 3:01:38 component of the offer around this. So, 3:01:40 as an example, and I don't know the 3:01:42 compliance on this, but I'm going to 3:01:43 throw out an idea, which is that let's 3:01:45 say the GLP1 support is one of their 3:01:47 highest performing angles, and they're 3:01:49 driving a few million dollars a month in 3:01:50 new customer revenue through this angle. 3:01:52 or they could then go and craft an offer 3:01:54 specifically for this angle on a 3:01:56 dedicated landing page which has 3:01:57 something along the lines of the GLP1 3:02:00 angle going into it repositioning maybe 3:02:02 the product title slightly and then if 3:02:04 you hop on a subscription on a 90-day 3:02:07 billing cadence so we collect 90 days 3:02:09 right up front to be able to improve the 3:02:10 profitability of acquisition then you 3:02:12 also get 25% off your GLP1s with our 3:02:17 registered partner and so you would 3:02:20 partner with an actual GLP1 provider 3:02:22 You would ask them if you can give 25% 3:02:24 off to your customers that flow through 3:02:26 from this funnel and then you give this 3:02:28 away to them. Now, cost of goods here, 3:02:30 $0. In fact, you could actually probably 3:02:32 make money on this offer because you 3:02:34 asked the GP1 provider to give you a 25% 3:02:36 off discount and then a 20% clip on all 3:02:39 future purchases or something like that. 3:02:42 And so, you would actually substantially 3:02:44 increase the profitability of this 3:02:46 whilst not changing cost of goods and 3:02:48 just giving something away for free. And 3:02:50 that's how you took the product, you 3:02:52 repositioned it, and then you crafted 3:02:54 the offer around it. Now, you could do 3:02:55 this for obviously all of them. So, if 3:02:57 we're repositioning the product into 3:02:59 hair health, as an example, one really 3:03:01 important piece of daily greens 3:03:03 impacting hair health might be that 3:03:05 really when you do double blind double 3:03:07 blind placebo experiments, the hair 3:03:10 health impact is only measurable post 90 3:03:13 days or post 120 days. And so you make 3:03:16 that unbelievably obvious in the front- 3:03:18 end offer. And so you go, "This is an 3:03:20 incredible product for hair health, but 3:03:22 it does take 120 days of consistent 3:03:24 usage." And so because of that, and 3:03:26 because our studies have shown that, 3:03:28 because our customers say that, here's 3:03:29 all the testimonials, we're giving you a 3:03:31 special offer to make sure that you can 3:03:33 achieve this goal, which is that our 3:03:35 normal 90-day subscription, we're going 3:03:37 to throw in a free extra 30 days in your 3:03:40 offer. Now, how are we actually 3:03:41 mechanically going to do this to be able 3:03:42 to preserve margin as a business? We're 3:03:44 not going to give you four in the first 3:03:46 order. What we're going to do is we're 3:03:48 going to give you the three, but on your 3:03:50 second order, you get one free. So on 3:03:52 the second billing, when they get 3:03:54 charged for another 90, which takes them 3:03:56 to 180 days, one of these within the 3:03:59 second order is obviously discounted 3:04:01 100%. Which is giving them like a 33% 3:04:03 discount on the second purchase. And so 3:04:05 that's where you're using the product 3:04:06 repositioning and you're building it 3:04:08 into the offer for congruency. Now, none 3:04:10 of these, and we could go all day. I 3:04:12 could come up with 20 different offers 3:04:13 for each individual product persona 3:04:15 here. Really, the key thing to 3:04:16 understand here is none of this would 3:04:18 have been possible. None of these offers 3:04:20 could have been made unless we started 3:04:22 repositioning the product. Cuz if we 3:04:24 just kept it here and we removed this 3:04:25 and we just go, "Okay, it's a daily 3:04:26 greens gummy. What can we do?" We're 3:04:28 incredibly restricted and limited into 3:04:30 what we can do here. Yeah, we could, 3:04:31 sure, we could add on 25% off GLP1s, but 3:04:34 this is irrelevant to 90% of people that 3:04:36 are buying the product, right? 90% of 3:04:37 people aren't on GLP1s. Or maybe they 3:04:39 are. I don't know. And then the same for 3:04:40 hair health. 90% of people probably 3:04:42 aren't buying for hair health. And so it 3:04:44 completely cuts the ability to do any 3:04:46 kind of offer around the clinical 3:04:47 studies here and substantially restricts 3:04:49 us to just very broad, very generic 3:04:51 offers that are not going to cut through 3:04:53 on a specific addressable market. All 3:04:55 right, the next offer mechanic is using 3:04:57 a bundle. Now, what it changes is units 3:04:59 per transaction goes up obviously 3:05:01 because you're forcing people into 3:05:02 buying multiple products. You get cost 3:05:04 of goods efficiency specifically in the 3:05:06 shipping and fulfillment, not in the 3:05:08 actual cost of goods. And then 3:05:09 contribution margin should positively be 3:05:11 impacted if you craft the bundle 3:05:13 correctly. And then this is obviously 3:05:15 best for multisq brands. Unless you can 3:05:17 package one product multiple times like 3:05:19 we did before with the 90-day trials. If 3:05:21 it's a consumable, if you don't sell a 3:05:23 consumable and you only have one 3:05:25 product, you can't really do this. Now, 3:05:26 I really don't think I have to explain 3:05:27 bundles. I already went through 3:05:29 economics and discount mechanics across 3:05:31 discounting. So, the exact same same 3:05:33 thing applies here. Now, bundles isn't 3:05:35 just taking different SKs and putting 3:05:37 them together, but it can just be the 3:05:38 same skew given away twice. And so, if 3:05:40 you sell board games for kids, buy two, 3:05:43 get 25% off. Like, this is a very common 3:05:45 offer that everyone's very familiar 3:05:46 with. This is probably what everyone 3:05:49 thinks when they think offers. In fact, 3:05:51 one of our interview questions that we 3:05:52 ask is, "Explain to me what a good offer 3:05:54 actually is." The most common answer is 3:05:56 just high discount percentages or some 3:06:00 kind of bundle because it is 3:06:01 advantageous to the advertiser because 3:06:03 they protect contribution margin which 3:06:05 is correct. The issue with bundles is 3:06:07 that you could do a lot more and so a 3:06:09 bundle is pretty simple. It's an easy 3:06:11 offer. It makes sense as long as you do 3:06:13 the economics correctly, but you can 3:06:14 take it a lot further and be a lot more 3:06:16 clever with how you're putting these 3:06:17 offers together. The next mechanic, 3:06:19 which is one of my favorite to be 3:06:21 honest, is gift with purchase. What it 3:06:23 changes is up if you force units per per 3:06:26 transaction up with the gift with 3:06:28 purchase with pretty much no change in 3:06:31 contribution margin. So there's a zero 3:06:33 impact. And this is for anyone that has 3:06:35 a secondary skew that they can give away 3:06:38 for free that's incredibly low cost and 3:06:40 pairs well. I'll show you the economics 3:06:42 of why I actually like this so much. So 3:06:44 let's say you sell towels, beach towels. 3:06:46 When someone buys one, it's $50. When 3:06:48 someone buys two, it's $100. But when 3:06:50 they buy two, they get a free beach bag 3:06:52 towel. Sorry, they get a free bag that 3:06:54 holds the towels as well. And then 3:06:56 because beach towels might be bought by 3:06:58 families, you could probably do a buy 3:07:00 four offer as well for families of four. 3:07:02 And you could give a further tiered 3:07:04 giveaway, but let's just keep it simple. 3:07:05 Now, the cogs on the tow is $15 a unit. 3:07:07 So, it's 15 over here, it's 30 here. 3:07:09 Now, the cogs on the bag is $0 here cuz 3:07:12 they're not getting a bag on this one. 3:07:14 They are. So, it adds to our cost of 3:07:15 goods, but it's super cheap. It's $4. 3:07:18 Now, shipping and fulfillment fees plus 3:07:20 3PL is $850 here on one unit. Over here, 3:07:23 it's $1250. Which means our contribution 3:07:25 margin is $26.50 3:07:28 and $53.50. 3:07:30 Meaning, if we do a break even 3:07:32 calculation on rorowaz here, the break 3:07:34 even rorowaz on just selling the towel 3:07:37 is 1.89. The break even rorowaz on this 3:07:40 offer is 1.87. 3:07:43 So, contribution margin almost doubled. 3:07:47 In fact, it did double. It over doubled. 3:07:48 And our break even return on ads, the 3:07:50 efficiency that we have to operate at, 3:07:52 decreased. So, we can be at a worse 3:07:54 efficiency and make double the amount of 3:07:56 money, which is a crazy position to be 3:07:58 in. And this is why gift with purchase 3:08:00 offer can work so incredibly well as 3:08:02 long as it is convincing. Now, this 3:08:04 offer, the economics of it look 3:08:06 incredible. Where it falls apart is do 3:08:08 people really care about getting a bag? 3:08:10 And that's why the offer actually has to 3:08:12 be well thought through. You have to 3:08:13 understand the customer correctly and 3:08:15 you have to be giving something that has 3:08:16 a high perceived value. Now, how would 3:08:18 you make the perceived value of this bag 3:08:20 increase? Well, you would actually sell 3:08:21 it on the website and you would sell it 3:08:23 for like $60. And so then when you 3:08:25 present this offer, you don't present it 3:08:26 as $100 plus you get a free bag. You 3:08:29 present it as $160 has been discounted 3:08:32 to 100. And so you're getting like 40% 3:08:34 off by buying two. What's even better 3:08:36 about this is what I showed before, 3:08:38 which is that you can implement gift 3:08:39 with purchases in conjunction with other 3:08:41 mechanics in the offer. So, this doesn't 3:08:43 just have to be the offer. You can layer 3:08:45 more stuff in. And so, on top of this, 3:08:47 you would take whatever the biggest 3:08:49 objection is or whatever the feedback is 3:08:51 from customers. And by the way, the 3:08:53 offer always starts at the customer. It 3:08:55 doesn't start at you. And this is where 3:08:57 people really get offers wrong is they 3:08:59 go, "How do I make the most amount of 3:09:00 money? How do I set this thing up so my 3:09:03 contribution margin and break even 3:09:04 rorowaz is as low as possible and as 3:09:06 high as possible. If you do that, you'll 3:09:08 just fail because nobody cares what you 3:09:10 want. It matters on what the customer 3:09:12 wants. And so when we then go and add 3:09:14 another variable to this offer, we're 3:09:16 not just going to arbitrarily throw 3:09:17 something in like, oh, let's just tack 3:09:19 on like a prize pool that you can win 3:09:22 like a holiday to Fiji that's worth 3:09:24 $5,000 and then we'll give this away 3:09:26 once a quarter. And so there's also this 3:09:28 perception of you could win a holiday 3:09:30 when you buy this offer as well. Cool. 3:09:31 That's great. But does our target 3:09:33 demographic care at all about going into 3:09:35 a prize pool for some holiday with no 3:09:37 known nods and no no information about 3:09:40 it at all? Maybe, maybe not. And so this 3:09:42 actually starts with understanding the 3:09:43 customer and directly surveying them and 3:09:45 asking them, "What would you like to see 3:09:46 from us? What held you back from 3:09:48 purchasing originally? What was the 3:09:50 concerns when you first bought the 3:09:51 product?" And you might find out, well, 3:09:53 the concern was that we actually thought 3:09:54 the towels would be relatively low 3:09:56 quality and that they wouldn't last, 3:09:58 particularly for kids in the family who 3:10:00 have ruined towels in the past. It's 3:10:02 like, okay, great. Well, when you buy 3:10:04 two, we're going to put in a lifetime 3:10:06 guarantee. And so, if anything ever goes 3:10:08 wrong with the towel, we'll send you 3:10:09 another one. Now, if that is the biggest 3:10:10 objection from people buying and you 3:10:13 then place that in the higher average 3:10:15 order value offer, well, guess what? Not 3:10:18 only you probably going to increase 3:10:19 conversion rates, but you're going to 3:10:21 increase conversion rates on this offer, 3:10:23 not this one. So, you're going to give 3:10:24 people even more reason to go up and 3:10:27 spend more. And then once again, this 3:10:29 for you might not matter at all. And 3:10:31 this is why offers need to be curated to 3:10:33 you specifically because a lifetime 3:10:34 guarantee on some businesses means 3:10:36 absolutely nothing. Nobody cares about a 3:10:37 lifetime guarantee. If I'm buying some 3:10:39 creatine gummies, I'm going to eat them 3:10:40 and then they're done. And so, we need 3:10:42 the lifetime guarantee or we need 3:10:43 whatever the offer is to actually meet 3:10:45 the customer where they are. So the next 3:10:47 mechanic is tiered offers. This 3:10:50 increases up because we're saying that 3:10:52 as you spend more, you get something. It 3:10:54 generally improves contribution margin 3:10:56 if it's structured correctly. And this 3:10:57 is best for people that have a binomial 3:10:59 average order value distribution. Now 3:11:02 what that means is that if we take 3:11:04 average order value and we put it on the 3:11:05 y ais, if we take average order value 3:11:08 and we look at all the different orders 3:11:11 within the business and we put these 3:11:12 across here, then the amount of 3:11:14 customers that are buying. So you could 3:11:16 do this as a bar chart. So people that 3:11:17 are purchasing $0 to $5, this is how 3:11:19 many people, this is how many people, 3:11:21 etc. I'm going to do it as a line graph 3:11:23 cuz it's a little bit easier for me, 3:11:24 which is that average order value will 3:11:27 generally look something like this for a 3:11:29 lot of brands, which is that there is a 3:11:31 point in which most people buy within 3:11:33 this average order value range and then 3:11:35 there's some people up here and there's 3:11:36 some people down here. Now, for quite a 3:11:38 few brands as well, you will have a 3:11:40 binomial average order value 3:11:42 distribution. And this could be based on 3:11:44 the actual current offer structure which 3:11:46 is that people can buy one or they can 3:11:48 buy three or this is very common when 3:11:50 you sell a lot of products. If there's a 3:11:52 very large skew count like in fashion, 3:11:54 people can kind of buy whatever they 3:11:55 want. They can put their cart together 3:11:56 in any way. And so what you'll end up 3:11:58 seeing is there'll be a peak and then 3:12:00 there'll be another peak. This is called 3:12:01 a binomial distribution. There is two 3:12:03 peaks that exist within the distribution 3:12:05 set. So where does this start to cause 3:12:07 issues? Well, if we look at average 3:12:09 order value in Shopify, average order 3:12:12 value might say that it's right here. 3:12:14 This is your AOV. Now, the issue is if 3:12:16 you go and set an offer based on this 3:12:18 average order value number, it's not 3:12:20 really going to do much because let's 3:12:22 say that you set an offer, this average 3:12:24 order value is at $100 and you want to 3:12:27 do some offer at $ 110 to try to push 3:12:29 people up. So, maybe it's you get free 3:12:31 shipping at $110. If you set free 3:12:33 shipping as an offer here at $110, well, 3:12:36 it's not convincing any of these people 3:12:38 to spend more because they're already 3:12:40 spending over $110. So now you're just 3:12:42 giving them free shipping for free and 3:12:43 you're just losing margin on them. And 3:12:45 these people aren't going to be 3:12:46 convinced to go up to $110 because they 3:12:49 only sit at like 70. So you're 3:12:51 effectively telling them to almost 3:12:52 double their average order value to get 3:12:54 free shipping. They're not going to 3:12:55 double their average order value. And so 3:12:57 the only people that you're convincing 3:12:58 to spend more is just really these 3:13:00 people right here. And this is a tiny 3:13:02 fraction of the total amount of 3:13:04 customers. This is maybe 10% of 3:13:05 customers. And so you've just given away 3:13:06 enormous amounts of margin on all of 3:13:08 these users, all of these customers. And 3:13:11 then none of these people are getting 3:13:12 convinced to actually spend more. So 3:13:13 it's a bad offer. Hence why if you have 3:13:15 a binomial distribution, you need to 3:13:18 instead look at the distribution curve, 3:13:19 not look at any averages, and then build 3:13:21 an offer around the two peaks. And so in 3:13:23 this case, we would want to set an offer 3:13:25 somewhere here that's targeted at these 3:13:28 guys. And then we would want to set an 3:13:29 offer somewhere here targeted at these 3:13:32 guys. And the idea is we want to shift 3:13:34 both of these up. Now once again, you 3:13:36 don't have to do this in isolation. This 3:13:37 doesn't have to be your whole offer 3:13:38 structure. You can use all of these 3:13:40 mechanics together. And so you can have 3:13:42 the gift with purchase to incentivize 3:13:44 one of those peaks up. And then you can 3:13:45 have a second offer for the second peak 3:13:47 that uses a different mechanic that 3:13:49 we've gone through. Then you can also 3:13:50 just straight discount. And the idea 3:13:52 here is that it will decrease your cost 3:13:54 to acquire a customer. the contribution 3:13:55 margin impact will probably be negative 3:13:58 unless uptweighs the margin erosion 3:14:01 which obviously you could do this 3:14:02 correctly. I showed you the graph before 3:14:04 where you want to find that point in 3:14:05 which units per transaction increases 3:14:07 and it maximizes your contribution 3:14:08 margin. So if you do this right you 3:14:09 actually do make more money. If you do 3:14:11 it wrong you make less. And this is best 3:14:12 for two scenarios. Either you have a few 3:14:15 SKs that have incredibly high gross 3:14:17 margin and then you just discount these 3:14:20 or you have grade C inventory. to 3:14:23 inventory that just isn't moving and you 3:14:24 need to get rid of it and you need to 3:14:25 turn it back into cash and you focus the 3:14:27 discounting here so that you can turn it 3:14:29 over. Now, to do this well, you want to 3:14:31 do a couple things. Number one, you want 3:14:32 to model average order value lift based 3:14:35 on previous discounts and how they have 3:14:37 impacted average order value and units 3:14:39 per transaction. That way, you don't 3:14:41 over discount and just lose a bunch of 3:14:42 money, but you discount to the correct 3:14:44 amount in which you maximize AOV and 3:14:46 maximize profit contribution. Then you 3:14:48 also model out the target return on ad 3:14:50 spend needed at this new discount 3:14:53 threshold. So not only do we know how 3:14:54 much average order value we should 3:14:56 expect, we also know what does our 3:14:57 target efficiency need to be on the 3:14:59 platforms to be able to hit the profit 3:15:01 contribution targets that we're after 3:15:03 and is this reasonable. Can we actually 3:15:05 hit these return on ad spend, these 3:15:07 efficiency numbers? Number three is you 3:15:08 want to do this very rarely, one to two 3:15:11 times a year. Flat discounting is 3:15:13 obviously going to train the entire 3:15:14 customer base and anyone that's warmer 3:15:16 in market to the fact that you discount 3:15:17 all the time. So you probably want to 3:15:19 preserve brand equity and perception in 3:15:22 market. And then number four is that 3:15:24 you're using it to turn stock into cash 3:15:27 through products that aren't moving as 3:15:29 opposed to just trying to rescue topline 3:15:32 revenue. These are just fundamentally 3:15:33 bad businesses which is the ones that 3:15:35 have to discount to achieve a topline 3:15:37 revenue figure. They'll always have 3:15:39 worse margins. They'll always be in a 3:15:40 worse position. and the cash flow is a 3:15:41 lot worse. Unfortunately, this is most 3:15:43 businesses though. Cross- sales and 3:15:44 upsells sit out separately from the five 3:15:47 offer mechanics. And the reason being is 3:15:48 that they sit downstream from the 3:15:51 primary offer. The main thing you always 3:15:52 want to be calculating and modeling out 3:15:54 when it comes to a crossell and an 3:15:55 upsell is the take rate. And what we're 3:15:57 really optimizing for here is 3:16:00 incremental contribution profit per 3:16:02 order which equals your take rate 3:16:05 percentage times by the average order 3:16:07 value that is going to be associated 3:16:09 with what is ever what is being taken 3:16:11 and then contribution margin or gross 3:16:14 margin percentage. So the take rate 3:16:16 percentage is the number of customers 3:16:17 that are exposed to the offer and then 3:16:19 actually take it. The average order 3:16:21 value for the upsell or crossell is the 3:16:24 dollar value of whatever it is that 3:16:26 they're adding to cart. So, if you're 3:16:27 saying, "Hey, here's an extra $20 item 3:16:29 you can add." This would be $20. And 3:16:31 then the gross margin of is obviously 3:16:32 whatever the gross margin is on that 3:16:33 product. And so, if we want to run 3:16:34 through a really clean example here, 3:16:36 let's say 10% of people take the upsell. 3:16:39 The upsell is $50 and the gross margin 3:16:42 is 50%. Then we have $2.50 in 3:16:46 incremental CP. Now, this doesn't seem 3:16:48 like a lot, right? You look at this and 3:16:50 you're like, "Ah, this isn't really even 3:16:52 worth it." But it is because at the 3:16:55 stage of the crossell or the upsell, the 3:16:57 order's already locked in. Okay? They've 3:16:58 already locked. They've already decided 3:17:00 on the primary purchase that they're 3:17:01 going to make. And at that point, right 3:17:03 before they're about to put in their 3:17:04 card details or often postcard details 3:17:07 on the actual post purchase page, you 3:17:09 give them an offer. You say, "Do you 3:17:11 want to add this in as well?" And then 3:17:12 this is just free additional incremental 3:17:14 profit if there's a high take rate and 3:17:16 if it's a good offer in the first place. 3:17:18 Let's say that you were actually only 3:17:20 making $20 in contribution profit on 3:17:23 acquisition. Well, now this upsell 3:17:26 increases that by over 10%. Which may 3:17:29 actually allow you to spend more on 3:17:30 acquisition. Now your CAT can be $2 3:17:33 higher to acquire even more customers 3:17:34 and take more market share, which allows 3:17:36 you to scale faster. Now, where people 3:17:38 make a massive mistake on the cross-ell 3:17:40 or the upsell, is that they focus on the 3:17:42 margin of the product. Let's push our 3:17:44 highest margin product as the upsell 3:17:46 because that's what will make us the 3:17:47 most amount of money. It's not true. So 3:17:49 let's say you have upsell A which is low 3:17:51 margin and then you have upsell B which 3:17:53 is high margin. Now on the low margin 3:17:55 product because it is more congruent 3:17:58 with what the customer actually bought. 3:18:01 The upsell makes more sense. It's 3:18:03 potentially just a better offer. Well 3:18:05 then because of that you have a 20% take 3:18:08 rate. But on the high margin products 3:18:09 people don't want this as much. is not 3:18:10 as good of a fit and therefore you have 3:18:12 a 10% take rate. Average order value 3:18:14 lift here is $40. Average order value 3:18:16 lift here is $30. Gross margin here is 3:18:19 60%. Gross margin here is 30%. So gross 3:18:22 margin here is terrible compared to 3:18:24 this. So what you need to understand 3:18:26 though is take rate isn't everything. 3:18:28 The thing that you need to be optimizing 3:18:29 for on an upsell is the incremental CP 3:18:32 per order. So let me run you through an 3:18:33 example. Let's say that you have upsell 3:18:35 A, which is low margin, and upsell B, 3:18:37 which is high margin. On upsell A, we 3:18:39 might have a way better take rate. Okay, 3:18:41 we have a 20% take rate compared to 10%. 3:18:43 Double the amount of people are taking 3:18:44 us. Our average order value lift is also 3:18:46 higher. We're making $40 extra rather 3:18:49 than 30. But on upsell A, we are low 3:18:51 margin. We're only 30% gross margin. And 3:18:53 that's maybe because we're giving away a 3:18:54 pretty heavy discount here. And also, 3:18:56 it's just a low margin product. On 3:18:57 upsell B, we have high margin. We're not 3:18:59 discounting it. We're giving it away at 3:19:00 the same price. Hence why the take rate 3:19:02 is way lower and hence why average order 3:19:03 value lift is way lower as well. Then 3:19:05 when we go and times all of these 3:19:06 together, which is what the formula is 3:19:08 up the top here, what do we get? We get 3:19:09 a $160 over here, we get a $180 over 3:19:12 here. So we actually make more money on 3:19:14 the half take rate upsell with lower 3:19:17 average order value lift because it had 3:19:19 better margin. And so don't just blindly 3:19:21 optimize your upsells and cross sales 3:19:22 based on take rate or average order 3:19:24 value or gross margin. Any of them 3:19:26 independently doesn't matter. It's how 3:19:27 they all flow into the mix to drive 3:19:29 incremental contribution profit. Quick 3:19:31 side note on exit intent offers as well. 3:19:33 Whether this is you're doing the popup 3:19:35 where someone goes to actually exit the 3:19:37 browser, it gives you a pop-up, which I 3:19:38 don't love. I'm not a huge fan of those. 3:19:39 They're pretty gimmicky. But the other 3:19:41 option is just an abandoned flow uh that 3:19:44 you can set up via emails as well. 3:19:45 There's three different stages of exit 3:19:47 intent. Someone can exit just when 3:19:48 they're browsing. Someone can exit when 3:19:50 they've already added something to the 3:19:51 basket or someone can exit when they've 3:19:53 gone all the way to checkout. What it 3:19:55 tells you is what stage in the funnel 3:19:56 they're actually at. So if they're 3:19:58 browsing and they exit, they have no 3:20:00 commitment. They haven't shown that they 3:20:01 actually want to buy anything. They're 3:20:02 actually still probably at the level of 3:20:04 product aware, not most aware. As they 3:20:06 have actually added something to basket, 3:20:07 they've showed that they want to buy and 3:20:09 as someone's reached checkout, they 3:20:11 pretty much are there at the purchase, 3:20:12 but there was some kind of friction that 3:20:14 stopped them. Potentially, they just 3:20:15 need a reminder to go back and buy or 3:20:16 they need some kind of risk reversal or 3:20:18 change in shipping offer. So, when you 3:20:21 think about the offer logic here, you 3:20:22 don't want to treat these customers the 3:20:23 same. You want to split out the offer 3:20:25 accordingly. So, if someone's just 3:20:26 browsing, you want to be generous. You 3:20:27 want to actually get them back. You want 3:20:29 to give them a reason to buy. You want 3:20:30 to put urgency associated with it. If 3:20:32 someone has added something to basket, 3:20:33 you don't want to go and give away all 3:20:34 your margin. This person might just be 3:20:36 taking some time and they'll buy in the 3:20:37 next one to two days. So yes, you can 3:20:39 give an offer, but you want it to be 3:20:40 small. And then if someone has reached 3:20:41 checkout, you don't want to give them an 3:20:43 offer at all. You want to do some kind 3:20:44 of risk reversal. Maybe you want to give 3:20:46 them free shipping. But if someone's 3:20:47 this far, I would be skeptical about 3:20:49 just giving away margin when it's 3:20:51 probably likely they're going to buy 3:20:52 anyway within the next 2 to 3 days. Now, 3:20:54 upsells, cross sales, where can you 3:20:55 actually put them? You can put them in 3:20:58 cart, you can put them at checkout, or 3:21:00 you can put them post purchase. Now, you 3:21:01 want to be thinking through number one, 3:21:02 the friction of where you're putting the 3:21:04 cross cells and upsells. Here, the 3:21:06 friction is medium because the buyer is 3:21:09 still deciding. So, if you're still 3:21:11 deciding whether you want to buy and 3:21:12 then I'm also throwing upsells and cross 3:21:14 cells at you, it could potentially make 3:21:16 you less likely to buy. Number one. 3:21:18 Number two is it's definitely going to 3:21:19 make you less likely to actually take 3:21:21 the offer cuz you're not bored in yet. 3:21:22 So, the typical take rate here is 3:21:24 between 5 to 15%. Now, once you get to 3:21:28 checkout, friction actually increases 3:21:31 further, which is counterintuitive. And 3:21:32 I used to think about this the other way 3:21:34 around, but our opinion has changed. 3:21:36 Take rates decline on a checkout offer. 3:21:38 And the reason why take rates declined 3:21:40 is because they already inherently 3:21:42 committed to the order and the price 3:21:44 that they have made. And so then when 3:21:46 you try to force a another offer onto 3:21:48 them while they're at checkout trying to 3:21:50 purchase, it ends up getting taken less. 3:21:52 They're still almost in a browsing 3:21:54 mindset here, willing to increase 3:21:55 average order value. But at this point 3:21:57 they are so committed that you trying to 3:21:59 just layer on and stack more things into 3:22:01 their cart often doesn't go down as 3:22:02 well. Now what actually helps here a lot 3:22:04 and how you can increase take rates and 3:22:06 do a good job at checkout is that you 3:22:07 don't try to sell them another product. 3:22:09 You try to sell them some kind of low 3:22:11 friction add-on. One thing that I used 3:22:13 to crush on back in the day with this 3:22:15 add-on is adding in shipping protection. 3:22:17 And so you add shipping protection as a 3:22:19 potential add-on at checkout. It's an 3:22:21 additional $4.99. 3:22:23 you'll get like a 30% take rate on 3:22:25 shipping protection and this is 100% 3:22:28 margin. You almost will never actually 3:22:30 refund on shipping and if you do 3:22:33 normally the carrier will pay for it 3:22:34 anyway. And so this is a really easy way 3:22:36 to add $1 to $2 of contribution margin 3:22:39 or gross profit to all orders with no 3:22:41 overheads. On incart you want to be 3:22:43 adding usually threshold unlocks are 3:22:45 really common. So you have the bar at 3:22:46 the top and it's like add one more 3:22:48 product and you'll get a particular 3:22:49 price. That's where you would place 3:22:51 those offer mechanics or you add samples 3:22:53 or you add like a complimentary product 3:22:56 and this is if you hit the threshold. 3:22:57 You then have post purchase. Now there's 3:22:59 pretty much zero friction here because 3:23:01 they've already made the order. 3:23:02 Payment's already been captured and if 3:23:04 it's set up correctly this is a 3:23:05 one-click ad. So it's do you want this? 3:23:07 Here's the offer. And if you click add, 3:23:09 instantly charges to their card and gets 3:23:12 added. So this is super low friction. I 3:23:14 would always have this. Take rates on 3:23:16 this as well can be unbelievably high. I 3:23:18 actually know people running 35 40% take 3:23:21 rates on post-purchase offers which is 3:23:22 so crazy. The other crazy thing is you 3:23:24 can string post-purchase offers together 3:23:26 which as a consumer gets really 3:23:28 annoying. I hate when people do this to 3:23:29 me but as a advertiser I love it because 3:23:31 you can give someone a post-purchase 3:23:33 offer which is this additional product 3:23:35 short time next 60 minutes discount of 3:23:37 30%. you get it taken, then you hit them 3:23:39 with another offer and you go because 3:23:41 you took this one, we have another one 3:23:43 for you which is 60% off and it's 3:23:45 another one and you have another take 3:23:46 rate percentage on the take rate. And so 3:23:49 you might have 25% take rates on the 3:23:51 first and then 10% take rates on the 3:23:53 second. And once again, I've seen some 3:23:55 uh supplemented CPG brands that have 3:23:56 crazy take rates post purchase. It's 3:23:58 allowed them to increase average order 3:24:00 value by 50 60% which just makes them 3:24:02 way more aggressive on acquisition and 3:24:04 it allows them to spend double the 3:24:05 amount on Facebook. What you can sell 3:24:07 here is literally anything. This is just 3:24:10 where some good AB testing goes far. I 3:24:12 just realized we've gone through this 3:24:13 entire segment, but I haven't even 3:24:14 defined what the difference is between a 3:24:16 cross-ell and an upsell. A cross-ell is 3:24:18 a different complimentary product 3:24:20 alongside what you're already buying. 3:24:22 So, an example is that if you buy a 3:24:23 t-shirt, we go and offer matching socks 3:24:26 or matching shorts. This is best for 3:24:29 multisq brands because you're selling 3:24:31 into other products. But an upsell is 3:24:33 just selling more of the same. So, if 3:24:35 you're buying the one month supply, 3:24:36 we're going to offer you a 3mon supply 3:24:38 with some kind of discount. Now, an 3:24:40 additional metric that you can track 3:24:41 here for post-purchase offers is what's 3:24:44 called RPV, which is revenue per visit. 3:24:46 You take post purchase revenue captured 3:24:49 and you divide by post purchase funnel 3:24:54 visits. You can also calculate this by 3:24:56 average upsell value times by conversion 3:24:59 rate. Now, the idea here is this is just 3:25:01 telling you how many dollars extra are 3:25:03 you making out of every person that goes 3:25:05 into the funnel as a function of the 3:25:06 post-purchase upsell. Now, a good 3:25:07 benchmark that you want to aim for is 10 3:25:10 to 15%. If you don't currently have 3:25:13 post-purchase upsells, well, the crazy 3:25:15 revelation if you made it this far in 3:25:16 the video is that you can add 10 to 15% 3:25:19 to your average order value overnight 3:25:20 just by adding post-purchase upsells if 3:25:23 they are done well. It's obviously the 3:25:24 caveat of the if it's implemented 3:25:26 correctly. So for context, if you have a 3:25:28 $200 average order value with an RPV of 3:25:32 10%. Then your RPV in dollar value is 3:25:36 going to be $20. So you can make an 3:25:37 extra $20 on every order through a good 3:25:39 post-purchase offer. Now, how should you 3:25:42 string a post a post-purchase offer 3:25:44 together? You want to always anchor high 3:25:46 and then downell below it. And most 3:25:48 brands end up inverting this where they 3:25:50 lead with the cheapest offer because 3:25:52 they think it'll convert better. But 3:25:53 what actually happens is that they price 3:25:55 anchor at $20 and then every highv value 3:25:58 upsell that they try to do after that 3:25:59 looks super expensive. So you want to 3:26:01 end up reversing the sequence. You show 3:26:03 an upsell of $50 to $60 post purchase. 3:26:06 You anchor there. If they don't take it, 3:26:07 you downell them into 20 to 30. If they 3:26:10 do take it, you bring them over to 3:26:12 another upsell which is slightly lower 3:26:14 at 30 to 40. And if they take that, you 3:26:16 can flow them again into downell number 3:26:18 two. or if they don't take this, you 3:26:21 flow them from downell one to downell 3:26:24 two. The five rules of upselling in cart 3:26:26 is number one, you only want to be 3:26:28 having one offer per slot in the drawout 3:26:31 cart. Or else, if you have three offers, 3:26:33 you just cause decision fatigue, which 3:26:35 is actually going to decrease your take 3:26:37 rates more than it will help it. Number 3:26:38 two is you want to focus on relevancy 3:26:40 over price, which is that you can pull 3:26:42 in a good price offer any day of the 3:26:43 week, but if it's completely not 3:26:45 contextual to what the person actually 3:26:46 has in their cart, then it doesn't 3:26:48 matter at all. they're not going to take 3:26:49 it. Number three is you want to focus on 3:26:50 high margin SKs. You don't really want 3:26:53 to be upselling into something that's 3:26:55 only going to get you an incremental 3:26:56 $45$5 in gross profit when instead you 3:26:59 could have something in here that's 3:27:00 relevant, that's a good offer that also 3:27:02 has high margin. Number four is you want 3:27:04 an easy to understand offer, 3:27:06 particularly on cross sales and upsells. 3:27:08 Unfortunately, you can get as complex as 3:27:10 you want in the front-end offer. And 3:27:12 that's what the majority of this video 3:27:13 has been about is that you actually want 3:27:15 to increase the complexity of the offer 3:27:17 because there's way more stuff that you 3:27:18 can do than just straight discounting. 3:27:20 But when it comes to cross sales and 3:27:21 upsells, unfortunately, you don't want 3:27:23 to get too complicated because you've 3:27:25 got really two to three seconds where 3:27:27 the offer is legible. They're going to 3:27:28 read it and then they're going to move 3:27:29 on. And so you want to make it 3:27:31 abundantly clear what the offer is, 3:27:33 which usually unfortunately has to be in 3:27:35 the form of some kind of flat discount 3:27:37 when it's added to cart. And then number 3:27:38 five is you want to measure the success 3:27:40 of upsells in cart not by average order 3:27:42 value lift but by gross margin lift. Or 3:27:45 else you will end up over prioritizing 3:27:47 into low margin SKs rather than high 3:27:49 margin SKs that might actually have a 3:27:51 lower average order value. So moving 3:27:52 into how you select the right offer for 3:27:55 you. There's four questions you want to 3:27:56 ask. Number one is what is the gross 3:27:58 margin flaw? You want to start designing 3:28:00 offers and understanding what the 3:28:01 economics of them actually looks like 3:28:03 and whether they will improve the 3:28:04 profitability of the business or whether 3:28:06 you're just going to erode margin for no 3:28:08 net benefit. Number two is what does the 3:28:10 average order value distribution look 3:28:11 like of your customers? Is there two 3:28:13 peaks? Should you be splitting the offer 3:28:16 across those thresholds then to move 3:28:17 those customers up or is there just one 3:28:20 peak and you should focus on moving one 3:28:21 core audience up into spending more like 3:28:23 in supplements where you might just have 3:28:24 a 30-day supply and you're trying to 3:28:26 move them up to 90. Number three is do 3:28:28 you have a single hero skew or do you 3:28:30 have multiple SKs cuz this completely 3:28:31 changes the dynamics of how you put the 3:28:33 offer together. In single SKS you want 3:28:35 to be leaning on product repositioning 3:28:36 as the highest lever then gift with 3:28:38 purchase and being disciplined with 3:28:40 discounting. But on multis you can do a 3:28:42 lot more. You can go into native 3:28:43 bundles. You can do cross category 3:28:45 upsells as well as bundles and you can 3:28:47 start to add in a bunch of threshold 3:28:49 mechanics that unlock different 3:28:50 products. And then number four is what 3:28:51 is the buyer's repeat behavior? As we 3:28:54 went through before, you can implement 3:28:56 subscription offers into the front-end 3:28:58 offer. You can start to really play 3:29:00 around with discounting the presentation 3:29:03 of gift with purchase, the time periods 3:29:05 in which you try to lock people in and 3:29:07 build them on in terms of cycles based 3:29:09 on the repeat behavior of the customer. 3:29:11 You might also have almost no retention 3:29:13 or repeat behavior because of the 3:29:15 industry that you're in. And so as a 3:29:16 product of that, you really want to 3:29:18 focus on actually sustaining gross 3:29:20 margin on first order, considering 3:29:22 that's where most of your gross margin 3:29:24 comes from. One last comment here on 3:29:25 selecting the right offer for you is 3:29:27 that there is 100 different offers that 3:29:28 you can run, there are probably 10 that 3:29:30 are good. And then of the 10 that are 3:29:32 good, no one can really say which one 3:29:34 will work best until you actually test 3:29:35 it in market. Then off the back of that, 3:29:36 once you have a large enough existing 3:29:37 customer base, you want to survey the 3:29:39 existing customers as much as possible 3:29:41 to be able to further iterate on the 3:29:42 offer. Offers also typically don't stay 3:29:45 static over time. So if you look at any 3:29:46 large CPG brand, the offer may stay 3:29:49 static for 6 months, 12 months, but on 3:29:51 the side, they are likely split testing 3:29:53 different offers on dedicated landing 3:29:55 pages and funnels to then find a better 3:29:57 one to rotate in. Because the unlock in 3:29:59 being able to find a better offer is one 3:30:00 of the most valuable things that you can 3:30:01 do in the front-end funnel as long as 3:30:03 creative and everything else is dialed 3:30:04 in. So then lastly, there's seven ways 3:30:07 that offers end up failing. Number one 3:30:08 is people just default to a blank 3:30:10 discount and they call it an offer. They 3:30:11 said that our offer going into this 3:30:13 period is 25% off, but it completely 3:30:15 negates everything that we've gone 3:30:16 through in this video, which is that the 3:30:17 offer is way more than just a discount 3:30:20 that you apply. Number two is people 3:30:21 miscalculate gross margin. They either 3:30:23 just include cost of goods sold and so 3:30:25 they don't include the impact of 3:30:27 shipping and fulfillment, which might 3:30:28 actually improve in bundles and as units 3:30:31 per transaction increase, or they 3:30:33 calculate it wrong in the other 3:30:34 direction. And so they don't understand 3:30:35 the true impact of gross margin on 3:30:38 efficiency targets like rorowaz and 3:30:40 therefore they discount way too hard to 3:30:42 where the efficiency will never be able 3:30:43 to reach the discount threshold to be 3:30:45 able to maintain margin. Number three is 3:30:47 that people run the same offer to new 3:30:49 customers and returning. Fundamentally, 3:30:51 these are two very different mechanics. 3:30:53 On new customers, we're optimizing for 3:30:55 first purchase economics and getting the 3:30:57 customer in to buy. on repeat offers. 3:30:59 We're really trying to deepen cohorts by 3:31:02 locking these people in to buy for a 3:31:03 sustainable period of time. And so the 3:31:05 offer that hits these two audiences 3:31:07 should look very different. Number four 3:31:08 is people set their thresholds based on 3:31:10 their mean average order value rather 3:31:12 than looking at the distribution curve. 3:31:14 Honestly, probably like 80% or more of 3:31:16 people do this and it means that their 3:31:18 offer is just completely set up wrong. 3:31:19 Number five is gift with purchase is 3:31:22 actually set up with the wrong gift. 3:31:23 either it's a gift that their customers 3:31:25 actually don't care about and it's not 3:31:26 materially even changing take rates on 3:31:28 this offer or this offer just isn't high 3:31:30 margin enough that it's not actually 3:31:31 changing the economics of the bundle 3:31:33 substantially enough to even matter and 3:31:35 so the economics and the math wasn't 3:31:37 ranked correctly behind this gift. It 3:31:38 was just thrown in because gift with 3:31:40 purchase sounds good but we didn't 3:31:41 actually consider the economics of it. 3:31:43 Number six is just running a 25week 3:31:45 promo calendar all year round. 3:31:46 continuing to just rotate out different 3:31:48 percentage discounts across different 3:31:50 product categories and calling that 3:31:51 offer rotation when this is incredibly 3:31:53 unsustainable and means that you just 3:31:55 have to do this forever. And then number 3:31:57 seven is treating Q1 as if it's a 3:31:59 seasonal dip and not an offer issue. You 3:32:01 can actually rotate offers into Q1 that 3:32:04 are contextually relevant to this point 3:32:06 in the year that allow you to not have a 3:32:08 dip. It's also one of the best points in 3:32:10 the year to be testing offers because of 3:32:13 the overall suppression in conversion 3:32:14 rates and market sentiment. So the five 3:32:16 things I'd be taking away from this 3:32:18 video is number one, the offer becomes 3:32:20 one of the most important levers in 3:32:22 acquisition. And so if you're not 3:32:23 testing your offer, you should be you 3:32:25 should be setting up separate funnels 3:32:26 with dedicated landing pages to test 3:32:28 different front-end acquisition offers 3:32:30 as it can substantially change the 3:32:31 profit on first purchase as well as what 3:32:33 retention looks like. Number two is the 3:32:35 offer isn't the discount. The discount 3:32:37 is one mechanic that sits inside the 3:32:39 offer, but it's usually the worst one to 3:32:41 lead with. Number three is that 3:32:43 discounting without understanding the 3:32:44 curve of how contribution profit and 3:32:47 average order value changes means that 3:32:48 you won't plan correctly and you'll just 3:32:50 set up discounts. You'll set up offers 3:32:52 without understanding the other side of 3:32:53 the coin which is what efficiency do we 3:32:55 have to hit on paid platforms to be able 3:32:57 to actually make this work. Number four 3:32:59 is the five offer mechanics which is 3:33:01 product repositioning bundles gift with 3:33:03 purchase tier threshold and then 3:33:05 disciplined straight discounting. And 3:33:07 then number five is you should have some 3:33:09 form of upsell, cross-ell or 3:33:10 post-purchase upsell. And it should be 3:33:12 structured around maximizing 3:33:14 contribution margin. And I gave you two 3:33:16 metrics as to how you can actually 3:33:18 measure success. Now, if you want a 3:33:19 calculator that shows you how your 3:33:21 discounting changes your efficiency 3:33:23 targets, we'll put a link in the 3:33:24 description below. If you're an 3:33:25 e-commerce brand, you can access it to 3:33:27 start planning out your offers and how 3:33:29 discounting is going to impact 3:33:30 efficiency. If you're an e-commerce 3:33:32 brand doing over $5 million a year, you 3:33:34 can also click the link in the 3:33:35 description, watch a quick two to three 3:33:36 minute video that'll run you through the 3:33:38 audit process. And then if you're a 3:33:39 performance marketer that has gotten 3:33:40 this far in the video, reach out to us 3:33:42 at hiringdigital.com.au. 3:33:44 We're always looking for new performance 3:33:46 marketers on the team. If you run meta 3:33:48 ads as an e-commerce brand, this is the 3:33:49 only video that you need to watch on 3:33:51 meta structure. Whether you're just 3:33:53 starting a new ad account or you're 3:33:54 spending a million dollars a month on 3:33:55 Meta, we'll be running through every 3:33:57 level of structure, everything you need 3:33:59 to know so that you can maximize 3:34:00 efficiency on the platform. There's 3:34:02 three claims on account structure that 3:34:04 are all true at the same time. And by 3:34:06 the end of this video, you'll understand 3:34:08 why three of these claims all make sense 3:34:10 contextual to your ad account. Claim 3:34:12 number one, consolidation will always 3:34:15 beat segmentation within any meta ad 3:34:17 account structure. So, if you have fewer 3:34:19 campaigns, fewer adsets, you'll 3:34:20 generally get better performance, but 3:34:22 it's not necessarily the right thing to 3:34:24 do. Number two is that media buying 3:34:26 actually still matters at higher ad 3:34:28 spend levels. Now, if you're spending 3:34:29 $100 a day, it's probably not that big 3:34:32 of a lever when it comes to growing the 3:34:34 account and growing the business. But if 3:34:35 you're spending $100,000, $200,000, 3:34:37 $300,000 a month, maybe you're buying is 3:34:40 still going to give you a 10 to 20% 3:34:42 efficiency lift if you do it correctly. 3:34:44 And then third is there is actually no 3:34:47 universal perfect account structure that 3:34:49 fits every business. The reason why 3:34:51 there is a thousand YouTube videos on 3:34:53 how to structure ad campaigns is because 3:34:55 there is a thousand different types of 3:34:57 businesses. And so dependent on the 3:34:59 nuances within your particular business, 3:35:00 it will change how you think through 3:35:02 structure. I'm going to be giving you 3:35:04 through the way that you should think 3:35:06 through the problem so that you yourself 3:35:08 can make the structure yourself. Over 3:35:10 the next 2 hours, I'll walk you through 3:35:12 how Meta decides where your money's 3:35:14 going. The three spend playbooks if 3:35:16 you're doing under 50k a month in spend 3:35:18 50 to 250 and 250 plus. The difference 3:35:20 between ABOS and CBOS and when you 3:35:22 should use each. Three settings that you 3:35:24 should be turning off right away as you 3:35:25 finish this video. How to design high 3:35:28 performing adsets. How much spend should 3:35:30 go towards existing customers and 3:35:32 different audiences. What's the testing 3:35:34 budget look like? How do you math that 3:35:36 out? How do you back propagate from your 3:35:38 goals? And then number four is I'm going 3:35:39 to give you four diagnostic questions 3:35:41 that you should always be asking 3:35:42 yourself. Meta is optimizing for one 3:35:44 thing in 2026 and honestly forever. It's 3:35:46 revenue per user per minute. Now there's 3:35:49 two ways that Meta can grow this number. 3:35:52 Number one is they increase ad 3:35:54 inventory. And so they simply put more 3:35:56 ads onto their platforms or they buy 3:35:58 more platforms where more ads can serve. 3:36:01 Now you've actually seen this be the 3:36:02 case over the course of the last six 3:36:03 years. I remember six years ago I would 3:36:05 get one ad in every five to six posts. 3:36:08 Now, sometimes you get triple ads. 3:36:10 You're scrolling and you'll get an ad, 3:36:11 another ad, and then another ad all in a 3:36:13 row. And that's meta increasing the ad 3:36:15 inventory so that they can maximize 3:36:18 revenue on the platform. And then the 3:36:19 second is if they can't increase ad 3:36:21 inventory any further, if the platform's 3:36:23 starting to get not enjoyable to use 3:36:24 because there's just so many ads. Well, 3:36:26 the other thing that they can do is they 3:36:28 can just increase the cost for the 3:36:30 advertisers to serve, which is your 3:36:31 CPMs, your cost per thousand 3:36:33 impressions. And so the same amount of 3:36:34 ad inventory, but let's make it more 3:36:36 expensive for everyone. And you see this 3:36:38 in year-to-year CPM inflation. We have 3:36:41 over 150 million in ad spend connected 3:36:43 to our business manager. And we can go 3:36:45 and do an aggregated ad report and look 3:36:47 at what CPMs have looked like over the 3:36:49 course of the last 3 years. And it's 3:36:51 cyclical with obviously Black Friday, 3:36:53 but every year it goes up. And what you 3:36:55 end up seeing is there's about 20 to 30% 3:36:57 inflation in CPMs. Now, obviously, 3:36:59 there's a little bit of natural 3:37:00 inflation to the dollar that you need to 3:37:02 factor out of that, but still CPMs are 3:37:04 going up faster than inflation is. And 3:37:06 the reason for that is that Meta has to 3:37:09 continue to publish good quarterly 3:37:11 earnings because they're a publicly 3:37:12 traded company. And the way that they do 3:37:14 that is they need to increase revenue on 3:37:16 their biggest product, which is the ad 3:37:18 product. Now, this seems all very doom 3:37:20 and gloom and like, "Oh, Meta is against 3:37:21 you. Everything's becoming more 3:37:22 expensive. It's a terrible platform. 3:37:23 Don't spend on it." That's not 3:37:24 necessarily true because if Meta 3:37:26 increases CPMs, everyone just becomes 3:37:28 unprofitable. So they can't do that. 3:37:30 They can't just make all their 3:37:31 advertisers unprofitable or else people 3:37:32 will stop spending. And so what they 3:37:34 have to do when they increase CPM is 3:37:36 they also have to increase expected 3:37:38 conversion rates or ROI of the platform. 3:37:42 So they need to effectively more 3:37:44 efficiently pull dollars out of users on 3:37:47 the platform and transfer them to 3:37:49 advertisers so that you get conversion 3:37:52 rates that outweigh the CPM increase. 3:37:55 The reason why all this matters and the 3:37:56 reason why this context matters is that 3:37:57 every time Meta rolls out an update, 3:37:59 whether it's Andromedor, whether it's 3:38:00 Gemin whether it's one of the smaller 3:38:02 updates that you've never even heard of, 3:38:04 what they are trying to always do is 3:38:07 increase the efficiency of ad serving so 3:38:09 that then they can make it more 3:38:10 expensive for you to actually place. 3:38:12 Now, a core principle of setting up 3:38:14 structure on the account is 3:38:15 understanding that what you see in the 3:38:18 platforms is not necessarily what Meta 3:38:20 is optimizing. And that's where you get 3:38:21 a lot of poor decision-m being made 3:38:23 within account structures or within 3:38:25 decisions as to where budget should 3:38:26 flow. There's a lot of stuff that you 3:38:28 see that's not necessarily reality. And 3:38:31 so an example is what you see in the 3:38:33 platform is lastclick attribution. 3:38:35 Meaning if a user clicks on multiple ads 3:38:37 and then buys the purchase the 3:38:39 conversion value will just go to the 3:38:41 final ad that they touched. But in 3:38:43 reality Meta is optimizing across 3:38:46 multi-touch attribution. Meta knows that 3:38:48 just optimizing towards the last click 3:38:50 is meaningless when there was all of 3:38:52 these prior clicks and prior 3:38:53 interactions that led up to the 3:38:55 conversion. And so what you end up 3:38:56 seeing is that if you have three ads 3:38:58 here and the clickfunnel looks like this 3:39:00 and then they ultimately buy, you on the 3:39:03 surface will go, "Ah, this is the ad 3:39:05 that's performing killer." But Meta will 3:39:06 still be distributing spend here because 3:39:08 it sees that a click actually occurs 3:39:10 here, click occurs here, and then the 3:39:12 final click occurs. And so in the back 3:39:13 end, it's likely allocating 1/3 credit 3:39:17 to each one, which is why spend's 3:39:18 getting distributed in the way that it 3:39:20 is. So another example of what you see 3:39:22 is that one ad gets all the credit. But 3:39:24 the reality is that Met is optimizing 3:39:26 across a chain of impressions. What you 3:39:29 see is ad level return on ad spend. So 3:39:32 you're going down here and looking at 3:39:33 rows on these individual ads. But what 3:39:35 Met is actually optimizing for is your 3:39:37 CPA target at the adset level. Because 3:39:40 the bidding and optimization actually is 3:39:42 inherited from the adset, which is why 3:39:44 you don't set cost caps at an ad level. 3:39:46 If you're going to set cost caps or any 3:39:47 kind of more complex bidding approach, 3:39:50 you're going to do it at the adset level 3:39:51 instead. Now, if we go back to this 3:39:53 concept of sequencing where a user might 3:39:55 pathway through multiple ads, all of the 3:39:57 ads, let's say, are spending $5,000, but 3:39:59 they're at very different returns. When 3:40:01 you look at this on the surface, you 3:40:03 make an obvious decision if these are 3:40:04 all sitting under the one adset, which 3:40:06 is let's turn this ad this ad off. Let's 3:40:08 put all the spend here. Now, besides 3:40:11 sequencing, besides the fact that, well, 3:40:13 these ads are probably doing a little 3:40:14 bit of heavy lifting, we need to think 3:40:16 about why is Meta distributing spend to 3:40:19 these two ads. Well, there's something 3:40:21 called the breakdown effect. And this is 3:40:23 public documentation by Meta. You can go 3:40:25 and Google for it right now and pull up 3:40:26 their actual uh page on this. And the 3:40:28 idea is that when you use any of the 3:40:30 breakdown features in Meta Ads, so you 3:40:33 can go to breakdown and you can click on 3:40:34 age or you can click on gender or you 3:40:36 can click on placement type or you can 3:40:38 click on region. Okay, there's all these 3:40:40 different options as to how you can 3:40:41 segment and break down the data. What 3:40:43 you will often see is weird stuff that 3:40:46 doesn't make sense. You'll look at a 3:40:48 breakdown, for example, on placements, 3:40:50 and you'll see that stories are at a 4x 3:40:53 row. Feed is at a 3x, but the feed is 3:40:57 getting all of the spend. It might be 3:40:59 holding 80% of all the spend in the 3:41:01 account. And you go, why is this 3:41:03 happening? Seems pretty easy to fix 3:41:05 this, right? We just launch a dedicated 3:41:07 campaign that only places on stories 3:41:09 because stories perform better. But it's 3:41:12 a flawed assumption because you are 3:41:13 looking at blended data that is not 3:41:16 counting in the incremental impact of 3:41:18 pushing more spend through that 3:41:19 particular channel or through that 3:41:21 particular placement. And so if we 3:41:22 simplify this back to the ad example up 3:41:24 here, the reason why this ad is getting 3:41:27 the same amount of spend as this ad is 3:41:30 because Meta has tried to spend more 3:41:32 here. It's tried to push it up past, 3:41:35 let's say, $300 a day to 320. But when 3:41:39 it does that, there's zero incremental 3:41:41 returns. You don't make any more money. 3:41:44 But over here, when this was, let's say, 3:41:46 a lot lower down at $200 a day, Meta 3:41:49 went, "Oh, we can't put spend here. 3:41:50 We're getting no returns." So, let's 3:41:52 see. Even though this has a lower base 3:41:53 return, do we get any incremental 3:41:55 returns here by putting more spend? And 3:41:57 it goes on layers 250 in and we make 3:42:00 more revenue. Now, it's not at a great 3:42:02 return. Maybe it's at a 2x or something, 3:42:04 but this is at a 0x. And so, budget goes 3:42:07 here. Same thing applies at a story and 3:42:09 a feed level. So Meta will try to put 3:42:11 more budget through stories. Obviously, 3:42:13 it's got a better return, but when it 3:42:15 does it, the incremental ROI is so poor 3:42:19 that it would rather just put spend into 3:42:21 the feed. So the distribution of spend 3:42:24 majority of the time at a breakdown 3:42:26 level is actually accurate and you 3:42:28 should trust Meta. Now, there are cases 3:42:31 where that isn't the case, and that's 3:42:32 where you have to be a good media buyer 3:42:33 and understand what breakdowns matter, 3:42:35 what don't, and how you should be 3:42:36 thinking through distributing spend. But 3:42:38 majority of the time, if you're a 3:42:40 beginner and you get too deep into 3:42:42 breakdowns, you'll just make a bunch of 3:42:44 segmentation decisions that actually 3:42:45 aren't commercially aligned with what 3:42:46 the platform wants you to do, and you'll 3:42:48 get worse performance. So unless you are 3:42:49 like an expert six year, sevenyear in 3:42:52 media buyer, you should not be 3:42:54 implementing changes based on complex 3:42:56 breakdowns or your limited understanding 3:42:59 of return on ads spended an ad. Now it 3:43:01 wouldn't be a Facebook ads long form 3:43:03 video if there wasn't a mention of the 3:43:05 buzzword Andromeda. So here's a 30-cond 3:43:06 explanation which is that previously you 3:43:08 would choose interest, you would choose 3:43:10 audiences and then you would load up a 3:43:12 bunch of creative and that creative 3:43:13 would serve to that interest. Now you 3:43:15 load in your creative, you leave 3:43:17 everything broad and Meta looks at the 3:43:19 creative and based on its understanding 3:43:22 serves it to a relevant audience. That's 3:43:24 effectively the retrieval system chain. 3:43:26 Now you might be thinking, well, does 3:43:27 that mean interests are dead? Does that 3:43:29 mean lookike audiences are dead? We 3:43:31 shouldn't use these anymore. Generally 3:43:32 speaking, yes, interest targeting is 3:43:34 super flawed. And if you actually go 3:43:36 down this rabbit hole, you'll learn a 3:43:37 bunch of reasons as to why interest 3:43:38 targeting is not good. I'll give you one 3:43:40 of them, which is that interests don't 3:43:42 take into consideration intent. And so 3:43:44 if I said I hate dogs on a Facebook 3:43:47 post, Meta would group me into being 3:43:49 interested in dogs. And so if you go and 3:43:51 target the dogs interest, you would 3:43:52 target me. But I actively went out there 3:43:54 and said, "I hate dogs." And I commented 3:43:56 on a bunch of posts. And so it isn't 3:43:58 intent driven. It's simply interest 3:44:00 driven. And so because of that, Meta put 3:44:03 out a report or someone put out a report 3:44:04 saying that 30% of the people that are 3:44:08 inside of an interest group aren't meant 3:44:10 to be there. They were incorrectly 3:44:12 assigned. And so interest groups in 3:44:14 itself is a poor way of categorizing 3:44:16 users. Now, if you really want to go a 3:44:18 little bit deeper on this topic, the way 3:44:20 that interest targeting worked 3:44:21 conceptually is it was effectively 3:44:23 labeling. And so you would be interested 3:44:26 in a pets post. And once again, that 3:44:28 just means an interaction, comment, a 3:44:29 like, something. Meta would tag you with 3:44:32 pets. And now anyone that wants to 3:44:33 target pets would target you. Very 3:44:36 rudimentary. When you think about Meta 3:44:37 being a trillion dollar business, you're 3:44:40 like, "What? They're just putting labels 3:44:41 on people based on the post that they 3:44:42 interact with. That doesn't seem very 3:44:44 sophisticated and it's not, which is why 3:44:46 it doesn't perform as well as targeting 3:44:48 broad. What happens when you target 3:44:49 broad? Instead, you can think of it as 3:44:52 if it's a vector space. Now, I'm just 3:44:54 drawing three axes here, but in reality, 3:44:56 this is like 10,000 dimensions. But what 3:44:59 happens is that when I go and interact 3:45:00 with a pet's post, I get plotted on the 3:45:03 graph. And then depending on what I 3:45:04 interact with in real time, I will get 3:45:06 moved around on this chart in a certain 3:45:08 direction. So if I interact with cats as 3:45:11 an example, I might get moved up in this 3:45:12 direction. But if I interact with dog 3:45:15 posts, I get moved over in this 3:45:17 direction. And then what ends up 3:45:19 happening is people will naturally 3:45:20 cluster through this three-dimensional 3:45:22 space. And then when you go and target 3:45:24 broad, instead of targeting people with 3:45:25 labels, you're just targeting this big 3:45:27 space and like sprinkling your ad out 3:45:29 all over the place and going who 3:45:31 interacts with it. And there'll be these 3:45:32 hotspots and over here the people in 3:45:35 this area actually start buying from 3:45:37 you. and Meta goes, "Okay, well, let's 3:45:38 just target this area." And then it will 3:45:40 go and target this area right here. And 3:45:42 then as you start to scale, what happens 3:45:44 is the area increases around this spot. 3:45:47 And so you target colder and colder 3:45:48 audiences. And this is why most ads 3:45:50 fatigue. This is why most campaigns die 3:45:51 when you try to scale them up because 3:45:53 you go from a hyperspecific target 3:45:55 demographic and you try to scale out of 3:45:57 it. And these people out here actually 3:46:00 aren't convinced enough to buy your 3:46:01 product cuz your ads might not be good 3:46:03 enough. Now, the caveat with this whole 3:46:04 model is that it isn't 3:46:06 three-dimensional. it's like 10,000 3:46:08 dimensions and so it can be much more 3:46:10 specific in the actual clustering of 3:46:12 users. So the first core thesis to 3:46:15 understand is that consolidation beats 3:46:18 segmentation. Really important concept 3:46:20 to always be thinking about anytime 3:46:21 you're structuring a meta account. Now I 3:46:23 want to run you through the timeline of 3:46:25 why this used to not be the case and why 3:46:26 it's counterintuitive and why a lot of 3:46:28 agencies particularly legacy agencies 3:46:30 are still hyper segmenting all over the 3:46:32 place and it's probably ruining your 3:46:33 performance. Now, what's also really 3:46:35 unfortunate as a little side note is 3:46:37 that the legacy outdated agencies are 3:46:40 typically the very cheap agencies, which 3:46:42 means they're the agencies that work 3:46:43 with smaller businesses. And so, as a 3:46:45 product of that, a lot of smaller 3:46:47 businesses that might be watching this 3:46:48 video that are just starting or maybe 3:46:50 spending $10,000 or $20,000 a month on 3:46:52 Meta, if you're with an agency and 3:46:54 they're cheap, they're probably running 3:46:56 legacy structures. And so, this is 3:46:59 probably relevant for you so that you 3:47:01 can push towards what actually works. 3:47:03 these days. So 2018 was actually the 3:47:05 first time I opened up a Metat account 3:47:07 and I was spending my own money at the 3:47:08 time. Now back here you could segment by 3:47:11 interest. In fact, it was favorable. 3:47:14 Interesting was actually one of the 3:47:16 biggest levers in the account back here. 3:47:18 Now obviously creative still mattered. 3:47:19 Obviously the website still like all of 3:47:21 these things played a part. But there 3:47:23 was an additional lever which is that if 3:47:25 you could take a creative and you could 3:47:27 find the interest that it performed on, 3:47:29 you could achieve scale. and creatives 3:47:31 would not perform on some interests and 3:47:33 they would perform on others. In fact, 3:47:35 you could duplicate adsets with the same 3:47:36 interest, same ads, and sometimes 3:47:38 they'll work, sometimes they won't. And 3:47:39 so, there was all these different 3:47:40 nuances in the account that would 3:47:43 actually allow you to see performance. 3:47:45 And the reason being is that the 3:47:47 platform was built around this. How it 3:47:49 used to work is every time you would 3:47:50 launch a new adset, the adset would go 3:47:53 out and it would serve to a thousand 3:47:55 random people and based on the initial 3:47:57 intent signals, so who would click, who 3:47:59 would interact, potentially who would 3:48:01 buy, meta will then zone in on those 3:48:04 types of users within the interest 3:48:06 group. And so you might have had that 3:48:08 when you launch this adset, for whatever 3:48:10 reason, moms age 40 to 50 interact. And 3:48:13 so it starts going after that audience. 3:48:15 You could launch the exact same ad set a 3:48:17 second time, same ads, same interest. 3:48:21 But on these thousand people, for 3:48:23 whatever reason, men aged 30 to 35 are 3:48:26 the ones that interacted because it's a 3:48:27 very small sample size. So you can have 3:48:29 bias in these small sample sizes out of 3:48:30 the gates. As a product, this adset 3:48:33 starts going off and optimizing towards 3:48:34 those types of audiences and you end up 3:48:37 with very different results on each and 3:48:38 you end up with very different 3:48:39 audiences, too. And that was just a 3:48:41 product of the way that the ad serving 3:48:43 worked at the time. And so you were 3:48:45 favored to Seg. It was a good thing to 3:48:49 have a ton of adsets where you were 3:48:51 constantly doing testing across 3:48:52 different interests, across different 3:48:54 audiences, across lookike audiences so 3:48:56 that you could try to squeeze out more. 3:48:58 Then what ended up happening? Well, Meta 3:49:01 improved the ad product, right? We're 3:49:02 looking at an 8-year time horizon here. 3:49:05 Some stuff happens. iOS 14 popped up 3:49:08 which substantially impacted the ability 3:49:10 for Meta to actually do interest groups 3:49:12 because of a lot of the interest 3:49:13 grouping was done using offsite pixel 3:49:16 data from blogs from whatever they were 3:49:17 interacting with on the internet. So 3:49:19 they had to start to change the way that 3:49:22 their retrieval system was working as 3:49:23 well as their ranking system for users. 3:49:25 Then AI started to pop up over here and 3:49:28 that's when Meta became meta. Went from 3:49:30 Facebook to Meta and Zuckerberg went and 3:49:32 made that big play into the metaverse. 3:49:34 bought a ton of chips off Nvidia. They 3:49:37 ended up not being able to do anything 3:49:38 with those chips. So what did they do? 3:49:40 They rolled them into inference to be 3:49:42 able to better train the actual ad 3:49:44 surfing platform. And so then the ad 3:49:46 serving platform started to improve even 3:49:48 further and they started to be able to 3:49:49 squeeze more efficiency. But then as a 3:49:51 product of that consolidation was 3:49:53 preferred during iOS 14.2. Just a side 3:49:55 note, Meta lost they said about 30% of 3:49:59 the signal that they were using for 3:50:01 targeting at the time, which means that 3:50:02 they had to compensate for this by 3:50:04 finding better buyers by not relying on 3:50:06 interest targeting signals and instead 3:50:08 being able to use the creative and go 3:50:10 broad. Now, the fundamental reason why 3:50:12 the platform now prefers consolidation 3:50:14 is because conversion data is siloed at 3:50:17 the campaign level. This doesn't just 3:50:19 apply to Facebook, it applies to Google, 3:50:21 it applies to Tik Tok, it applies to 3:50:22 Pinterest, it applies to all of the ad 3:50:24 platforms. They're all structured in the 3:50:25 same way if you haven't noticed which is 3:50:27 you have campaigns, you have adsets and 3:50:29 you have ads. On Google you have 3:50:30 campaigns, you have ad groups and then 3:50:32 you have app. This is the same case on 3:50:34 all the platforms. And the reason being 3:50:35 is it's the way that the platform silos 3:50:38 out data in targeting. I won't go too 3:50:41 deep into it, but I think it's important 3:50:42 to understand at least the architecture 3:50:44 of an ad account, whether you're a 3:50:45 beginner or whether you're an expert 3:50:47 because most people never even think 3:50:48 through this problem. Um to be able to 3:50:50 better understand structure. So you 3:50:51 should always be challenging everything. 3:50:53 Why does this exist? Why are they doing 3:50:55 it in this? Why do campaigns and adsets 3:50:58 exist? Why are there adsets? Why aren't 3:51:00 they just campaign? In fact, why isn't 3:51:02 there just one camp? Why isn't the 3:51:04 platform just ads and you just input 3:51:06 your ads and there's no structure up 3:51:08 here? Like, why does all of this 3:51:10 structure exist? Why have they 3:51:11 introduced this additional complexity? 3:51:13 And it's because you do need a level of 3:51:15 segmentation within the account to 3:51:17 better commercially align with the 3:51:19 objectives of most businesses. The 3:51:21 reason why campaign segmentation exists, 3:51:23 the reason why you can make multiple 3:51:25 campaigns is because businesses often 3:51:29 need to flow budgets through different 3:51:32 campaigns over time. And so you might 3:51:34 have a promotion that rolls in that then 3:51:35 rolls into something else. And so you 3:51:37 need segmentation in the actual KPIing 3:51:39 and reporting. Number one. Number two is 3:51:42 different business units exist in a lot 3:51:45 of businesses, which is that you might 3:51:46 have a category for pets and then you 3:51:49 have a category for children. As a 3:51:52 product of that, you don't want the 3:51:54 campaign getting confused and not 3:51:56 understanding who to target when you 3:51:57 have slightly different demographics. 3:51:59 And so, as a product of that, you want 3:52:00 the ability to be able to segment the 3:52:03 account based on the actual segmentation 3:52:05 of personas within the business. And 3:52:07 this allows you to do it. And one of the 3:52:10 downsides of that is that conversion 3:52:12 data here is not shared at the same 3:52:15 level as it is at the adset and ad 3:52:17 level. And so is there some conversion 3:52:19 data sharing going on here? For sure. 3:52:21 Campaign one is pulling from some of the 3:52:23 signals of campaign 2 and vice versa. 3:52:25 But they're not directly sharing all of 3:52:28 the conversion data with each other. 3:52:30 Which means that if you start to 3:52:32 introduce a bunch of campaigns in the 3:52:34 account and you have let's say four 3:52:35 campaigns for a small ad account with 3:52:37 not much budget, what you're actually 3:52:39 doing is taking all your conversion 3:52:41 data. Let's say you're getting 100 3:52:42 conversions a month and you're splitting 3:52:43 it up across four campaigns. So now you 3:52:45 have 25 25. You're inherently going to 3:52:49 get worse results in the account due to 3:52:52 segmentation because now you have less 3:52:54 conversion data that these campaigns can 3:52:57 optimize on. And anytime you have less 3:52:58 conversion data, you will see worse 3:53:00 results due to small sample size bite. 3:53:02 Which simply means that if you don't 3:53:04 have a lot of conversions, meta doesn't 3:53:05 have a lot of signal and so it doesn't 3:53:06 really know who to touch. The more 3:53:08 signal you have, the more conversions 3:53:09 you have, generally actually the better 3:53:11 the efficiency is. Which is why you 3:53:12 often actually see that when you can 3:53:15 crack through 100 or 200 conversions a 3:53:17 month, the ad account starts performing 3:53:19 better rather than worse. And it's 3:53:21 because there's enough signal that Meta 3:53:22 now understands who to target properly. 3:53:24 And you can get out of that rut of Meta 3:53:27 not really understanding who to target 3:53:29 because there's not enough conversion 3:53:30 data. So you always generally want to be 3:53:32 consolidating up at the campaign level 3:53:34 cuz every time you segment, you're 3:53:36 chunking out the data. Same thing kind 3:53:38 of applies at the adset level, not as 3:53:40 much. There is more data sharing that 3:53:41 occurs here than at the campaign level. 3:53:43 Number one. Number two is that all the 3:53:45 data at the campaign level also inherits 3:53:47 down. So you don't really have to worry 3:53:48 as much about segmentation at the adset 3:53:50 level. You can kind of segment all you 3:53:52 want. It's not going to impact 3:53:53 performance that heavily. You can 3:53:54 consolidate up. It's not going to 3:53:56 perform impact performance that heavily. 3:53:57 We're talking about really 5 to 10% 3:53:59 efficiency swings here based on hyper 3:54:01 segmentation versus consolidation. It's 3:54:03 noticeable particularly at scale, but 3:54:05 it's also going to be context dependent 3:54:07 on the actual. Then at the ad level, 3:54:08 obviously you're going to have massive 3:54:10 segmentation because you can't stack ad. 3:54:12 Now you technically can stack ads and 3:54:14 the name of this always changes. So by 3:54:16 the time you're watching this, it might 3:54:17 be different. At the moment, it's called 3:54:18 flexible ads, which is where you can 3:54:20 load in like five different creative on 3:54:21 the same ad. The reason why we generally 3:54:23 don't like these is because you don't 3:54:25 get visibility into data insights. And 3:54:27 so you can go and load five creative up 3:54:29 into one ad, which is all well and good. 3:54:31 Nice. It's consolidated. Meta prefers 3:54:33 consolidation. So it seems like the 3:54:35 right idea. But the issue is when this 3:54:37 ad performs well, we don't know what 3:54:39 creative is actually performing well. Is 3:54:41 it creative one, two, three, four, five? 3:54:43 which of the creative that we put in 3:54:44 here is actually lifting performance 3:54:46 because we want to do more like that. We 3:54:48 want to make more ads like that. But if 3:54:49 we don't have the ability to read the 3:54:50 insight, well, it's kind of useless. 3:54:51 Yeah, we got performance, but now we 3:54:53 don't know what to do with it. There's 3:54:54 also a couple other benefits of 3:54:56 consolidation over segmentation. And 3:54:58 this is mainly letting the machine 3:55:00 decide where to distribute spend. Now, 3:55:02 this can be a bad thing and it can 3:55:04 overweight your account into a heavy 3:55:06 degree of risk, but can be a good thing 3:55:08 if you're just trying to maximize 3:55:09 efficiency in the short term, which is 3:55:10 that when we're talking about 3:55:11 consolidation, Meta is picking where to 3:55:14 distribute your spend. If you just have 3:55:15 everything sitting in a campaign that's 3:55:17 a CVO with a bunch of adsets with a 3:55:19 bunch of ads, Meta is just going to go 3:55:20 and swing budgets around based on what 3:55:22 it believes is most efficient. when you 3:55:24 have segmentation and let's say an 3:55:26 structure. So you're choosing where 3:55:27 budgets go at the adset level. Well, you 3:55:29 are deciding where the budget goes and 3:55:31 inherently meta will always in most 3:55:33 cases distribute budget more efficiently 3:55:36 than you will. However, Meta will also 3:55:38 therefore distribute budget on an 80/20 3:55:40 Purto sprint, which is 20% of your ads 3:55:42 will get 80% of the spend. Now, kind of 3:55:44 annoying if you're making a 100 new ads 3:55:46 a month and none of them are getting any 3:55:48 spend and they're not getting tested. 3:55:49 Also kind of annoying if all the spend 3:55:50 goes into just one creative and maybe 3:55:52 you're paying influencers for 3:55:54 partnership ads that are getting no 3:55:55 spend or maybe you know that creative is 3:55:57 going to fatigue and now you're like 3:55:58 we're in a risky position because we 3:56:00 have no backup ads that are doing well. 3:56:01 And so this can be good but it can also 3:56:04 hurt you. Whereas on segmentation you're 3:56:06 going to be forcing spend across a 3:56:08 structure in a way that you want which 3:56:10 technically might derisk you might put 3:56:11 you in a better position might 3:56:12 facilitate better testing. uh if you're 3:56:14 a novice and you're not doing this 3:56:16 thoughtfully and methodically and 3:56:18 carefully, well then you can just end up 3:56:19 burning a bunch of money here. And so a 3:56:20 good way to kind of think through this 3:56:22 problem is that if let's say you're an 3:56:23 in-house business, $10 million a year 3:56:26 and you're making an in-house media 3:56:27 buying hire. If you're hiring someone 3:56:29 that is in their first year and they 3:56:31 have kind of no idea what they're doing, 3:56:32 I would prefer they run this structure 3:56:34 cuz I would actually prefer Meta 3:56:35 Distributes budgets over them because 3:56:37 they don't know what they're doing. If 3:56:38 you know what you're doing and you're 3:56:39 experienced and you understand weighing 3:56:41 risk against profit and efficiency and 3:56:44 understanding the changes that need to 3:56:45 be made, I would much prefer they run 3:56:47 this structure or somewhere in the 3:56:48 middle so that we can still have control 3:56:50 over where spend's getting forced 3:56:52 carefully and thoughtfully, but we're 3:56:53 not just burning money. There's also the 3:56:55 extreme of this which is there is a 3:56:58 circle on the internet that is focused 3:57:00 around consolidate everything. Just run 3:57:02 one campaign, one ad set, throw all the 3:57:04 ads under it. That's all you need. You 3:57:05 don't need an agency. You don't need any 3:57:07 kind of media buying. You don't need any 3:57:08 kind of segmentation. You don't need 3:57:10 structure. Just consolidate. Just upload 3:57:12 ads into a campaign and you're good. 3:57:13 That's a terrible idea for 90% of 3:57:16 people. Now, 10% of brands can get away 3:57:18 with that and they'll be okay. But 3:57:20 there's three reasons as to why you 3:57:22 don't want to just consolidate 3:57:24 everything. Number one is different unit 3:57:25 economics. And so, if you're a large 3:57:27 business with a lot of SKs, generally 3:57:29 the unit economics or the gross margin 3:57:31 will change across different categories. 3:57:33 you will have 70% gross margin on this 3:57:36 particular product range. But on this 3:57:37 particular product range, maybe you have 3:57:38 60%. Or you will have grade A, grade B, 3:57:41 grade C, grade D inventory. So on your 3:57:43 grade A inventory, you have incredible 3:57:44 unit. On your grade C inventory, you 3:57:47 also have incredible unit. Your gross 3:57:49 margin is the same. It's great, but it's 3:57:50 not selling. And so because of that, you 3:57:52 generally have to discount it hard, 3:57:54 which compresses the actual margin post 3:57:56 discount. And so if we put both of those 3:57:58 products or both of those ads into the 3:58:00 one campaign, we would end up with 3:58:02 likely a lot of spend getting 3:58:03 distributed to the heavy sale items. But 3:58:05 the heavy sale items is not where we 3:58:07 want our budget to go because we have 3:58:08 compressed margin there. We're probably 3:58:10 going to end up with worse contribution 3:58:11 margin and it's also just bad branding 3:58:13 to be heavy pushing sales messaging 3:58:15 through our entire funnel and existing 3:58:17 customers. So when you have different 3:58:18 unit economics across the product 3:58:20 portfolio, you actually want 3:58:22 segmentation. Number two is you might 3:58:24 have different audiences. And so an easy 3:58:26 example of this means that you might 3:58:27 have an activewear brand and the 3:58:29 activewear brand sells to both men and 3:58:31 women. Now if you go and put all the men 3:58:33 and all the women ads into one adset 3:58:35 into one campaign, the adset is likely 3:58:37 going to get confused. And in fact, you 3:58:39 see this on ad accounts all the time. 3:58:40 This is an easy call out anytime I audit 3:58:43 a brand that sells to both men and women 3:58:45 through core different product ranges. 3:58:46 So there's a men's range and there's a 3:58:48 women's range is that you just go down 3:58:50 under the adset level and you just take 3:58:52 a look at all the ads. You use the 3:58:53 breakdown feature. you look at gender 3:58:55 split and what you'll end up seeing is 3:58:56 that women ads are going to men and men 3:58:58 ads are going to women. Now men ads 3:59:00 serving to women isn't as bad of a idea 3:59:03 because women do a lot of purchasing for 3:59:05 their partners and so you end up seeing 3:59:07 directional purchase behavior from women 3:59:09 to men but you don't see it as much in 3:59:12 the other direction dependent on the 3:59:13 category and so you generally don't want 3:59:14 to be pushing women's active wear ads to 3:59:17 men. Will you get a return? Sure. People 3:59:19 will buy. You still get a return on it. 3:59:20 Is it the most efficient use of your 3:59:22 capital in the business? Absolutely not. 3:59:24 You should serve your women's active 3:59:25 wear ads to women. And so if you just go 3:59:27 and put all of these in the one ad set 3:59:29 because targeting primarily exists at 3:59:31 the adset level, that is where you 3:59:33 choose the targeting. That is where the 3:59:34 targeting generally resides for most of 3:59:36 what's occurring under it. It gets 3:59:38 confused. It doesn't know where to serve 3:59:39 them. So, it's just serving them to 3:59:40 everyone. Number three is high average 3:59:42 order value or low conversion volume 3:59:44 accounts. We have some clients that we 3:59:46 work with where average order value is 3:59:48 $10,000. on a $10,000 average order 3:59:51 value. The issue is is that you just 3:59:53 don't do much order volume. And so this 3:59:55 brand as an example, I think they do 50k 3:59:57 a day or something like that, it's only 3:59:59 five orders a day. Not a lot of purchase 4:00:01 volume at all. And so what ends up 4:00:02 happening in an account like this is 4:00:04 that if you have a CBO as an example, 4:00:06 and this is a bunch of ad sets, so this 4:00:08 is ad set one, set two, etc., is that 4:00:12 Meta will distribute budget at the adset 4:00:14 level not based on purchase, not based 4:00:17 on row, not based on actual signal that 4:00:20 we care about. Instead, because there's 4:00:22 nowhere near enough signal, there's only 4:00:24 like five conversions max getting 4:00:26 attributed into the account per day. 4:00:27 What ends up happening is Meta needs to 4:00:29 go upstream in its optimization. And so, 4:00:31 it goes up and starts looking at CTRs 4:00:33 and CPCs and hold rates. And now sure 4:00:36 these are pre-intent pre-click signals 4:00:38 as to something that might infer future 4:00:40 performance. But unfortunately the 4:00:42 reality is when you look at very large 4:00:44 data sets CPCs are not correlated with 4:00:47 returns except for the extremity of 4:00:50 bounds. And so if CPCs are super high 4:00:52 yes ROI will be super low. If CPCs are 4:00:55 super high that actually doesn't 4:00:57 correlate with ROI. So what you end up 4:00:58 saying when you graph this out is you 4:01:00 just say something like this, which is 4:01:01 that cost per click has almost no 4:01:03 correlation or a very weak correlation 4:01:06 to actual return on the ad. And so if 4:01:08 we're optimizing budget distribution 4:01:10 based on cost per click or CTR, we're 4:01:12 optimizing it towards a metric that has 4:01:14 nothing really to do with the actual 4:01:16 core objective, which is to drive 4:01:17 revenue. And so in this case, we likely 4:01:20 don't want to use a CBO. We likely don't 4:01:22 want the campaign to distribute budgets 4:01:23 how it is. And we want to think through 4:01:25 a structure that's going to look very 4:01:27 different from what you would expect in 4:01:29 any other account which is sort of 4:01:31 getting us towards the position of every 4:01:33 account looks different depending on the 4:01:34 actual commercial objectives of the 4:01:35 business. What do the unit economics 4:01:37 look like? What do the audience profiles 4:01:38 and personas look like? What does the 4:01:40 average order value and conversion 4:01:41 volume look like? Because all of these 4:01:43 things will change how we approach 4:01:44 consolidation or segmentation. I'll give 4:01:46 you a tactical example of where 4:01:48 consolidation on an audit I did a few 4:01:51 weeks ago was very relevant and not a 4:01:53 good idea which was this was a furniture 4:01:55 brand about 20 to 30 million they had a 4:01:58 cold campaign which was consolidated 4:02:00 that's all they had and then it was a 4:02:02 CBO and they had a bunch of adsets down 4:02:04 here now the issue was 57% 4:02:07 of the meta budget was going into the 4:02:10 top adset now what was this adset it was 4:02:12 just a dynamic product app now the issue 4:02:14 is return on ad spend looks really good 4:02:16 on service it was like a seven rorowaz 4:02:17 but once you break down to 7-day click 4:02:19 or you break down to incremental 4:02:20 attribution either one gave the same 4:02:22 read returns was actually a 1.9 4:02:24 everything else was performing better 4:02:26 than this on an incremental or a 7-day 4:02:28 click rate and so this was a good 4:02:29 example of where them just consolidating 4:02:32 up and putting DPAs in with everything 4:02:33 else not a good idea particularly in 4:02:35 that industry because you'll always end 4:02:36 up with overspend into an ad type that 4:02:39 doesn't actually perform that well on 4:02:40 cold audiences let's talk about what in 4:02:42 media buying actually died what doesn't 4:02:44 matter anymore and then what still 4:02:46 matters because there's a lot of stuff 4:02:48 that is irrelevant these days. And when 4:02:49 I say media buying still matters, I'm 4:02:51 talking about a very specific subset of 4:02:53 things within the account. There's a lot 4:02:55 of stuff that you can push to the side. 4:02:56 What 99% of people shouldn't be doing is 4:02:59 daily bid tweaks. Going into the account 4:03:01 using cost caps or bid caps and just 4:03:03 changing bids every single day. Now, 4:03:05 that used to be a strategy four or five 4:03:07 years ago that a lot of people did quite 4:03:09 well with, but these days you're really 4:03:10 moving the needle on a variable, on a 4:03:12 lever that has nowhere near as much 4:03:14 leverage as other things that you could 4:03:16 be focusing on. Now, could you 4:03:17 technically squeeze 5% more out of an 4:03:19 account using daily bid adjustments 4:03:20 every single day and dropping an hour a 4:03:22 day into this? Sure, probably. Could you 4:03:24 spend the same amount of time just 4:03:25 making better creative and double the 4:03:27 account or triple the account? 4:03:28 Definitely. And so if we look at return 4:03:30 on capital being time allocation, you're 4:03:33 way better off just making better, more 4:03:35 creative than going in there and just 4:03:36 tweaking with bids all day. Now, what 4:03:37 still matters is account structure and 4:03:39 account architecture. So thoughtfully 4:03:40 thinking through how we're segmenting 4:03:42 and setting up the account in alignment 4:03:43 with the commercial objectives of the 4:03:44 business. The next one, for smaller 4:03:46 accounts, this is applicable on very 4:03:47 large accounts, but for smaller 4:03:48 accounts, daily budget pacing. So 4:03:50 changing budgets every single day in 4:03:52 alignment with let's say expected 4:03:53 conversion rates across the week or just 4:03:55 in alignment with daily performance, not 4:03:57 a good idea. You don't want to be going 4:03:59 in and tweaking budgets all the time. 4:04:00 You're going to mess around with 4:04:01 campaigns. You're probably making these 4:04:02 decisions on very limited sample sizes 4:04:05 of data. Once again, was this the 4:04:06 strategy that you would do back in 2019, 4:04:08 2020? Yeah, for sure. Is this something 4:04:10 that matters these days? Not at all. 4:04:11 What still matters? Concept level 4:04:13 segmentation at the adset level. We're 4:04:15 going to go into this in a lot more 4:04:16 detail later on in this video, but this 4:04:19 matters a lot. What doesn't matter on 4:04:20 the other hand is interest targeting. 4:04:22 Very strongly of the belief this 4:04:24 shouldn't be in your account. You're 4:04:25 wasting resources. you're wasting tests 4:04:27 on something that doesn't even really 4:04:29 work anymore. What's a better use of 4:04:30 time is instead of going in and doing 4:04:32 interest testing and interest targeting 4:04:34 is you can do page testing instead. So 4:04:36 split testing different landing pages, 4:04:37 split testing CRO on the website. This 4:04:40 matters a lot. You can make an 4:04:41 inflection in conversion rates. That's a 4:04:43 much better test than going and just 4:04:44 messing with interest targeting on the 4:04:46 platform. I'm also going to throw in 4:04:47 here lookike audiences don't matter 4:04:49 anymore. They don't work. I could once 4:04:51 again talk for 5 minutes about all the 4:04:52 deficits of lookalike audiences and why 4:04:54 you should just be going broad instead. 4:04:56 What does matter on the other hand is 4:04:59 thinking through and controlling 4:05:01 percentage or dollar allocation to 4:05:03 existing customers. This is something 4:05:05 that you should really be thinking about 4:05:07 and controlling over time. Before I 4:05:09 start to give you the actual account 4:05:10 structure that you should be using at 4:05:11 each level of spend, I want to quickly 4:05:13 talk about page strategies. We as an 4:05:15 agency only work with eight and nine 4:05:17 figure businesses. And so as a product 4:05:19 of that, the content is more tailored to 4:05:21 very large businesses and people that 4:05:22 are in a position to work with us. If 4:05:23 you're small, this is going to be kind 4:05:25 of irrelevant. If you're big, this is 4:05:26 going to be super relevant, which is 4:05:27 that Meta caps the amount of ads that 4:05:30 you can have live on a page at about 3 4:05:33 to 500 ad. And so what this means is 4:05:35 that if you launch a lot of creative 4:05:37 every single week, every single month, 4:05:38 you very quickly pass this cap. If 4:05:41 you're even a somewhat big account, you 4:05:43 then have two options. Option number one 4:05:45 is you just start turning ads off 4:05:46 because you can't have more than 500 4:05:48 active ads in the account which is 4:05:49 really annoying. Or option two is you 4:05:51 make more pages. And so I'm going to 4:05:52 give you three options here that we 4:05:54 recommend to all clients when we run 4:05:55 into this position. And we run into this 4:05:57 position with pretty much like every 4:05:58 single client because all of our clients 4:06:00 are launching enormous amounts of 4:06:01 volume. Number one is you create 4:06:03 multiple duplicate pages. So someone 4:06:05 that does this quite well and I've done 4:06:07 a reel on this on Instagram is Grooms. 4:06:10 They have like 15 grooms pages and the 4:06:12 only difference is that the logo is a 4:06:14 different color and then they're just 4:06:15 running ads to a bunch of these 4:06:16 different pages. Now, personally, if I 4:06:18 was running an e-commerce brand, I would 4:06:20 just do this. I would just spin up a 4:06:21 bunch of pages for the sake of ads. I'm 4:06:23 not really too concerned about 4:06:24 consolidation onto a single page, but I 4:06:27 know some people are in which case your 4:06:28 other option is whitelisting or 4:06:30 partnership ads. And so, you just have 4:06:32 to lean more into running ads through 4:06:34 other people's handles that aren't 4:06:35 yours. Now, worth noting there is a core 4:06:37 difference between whitelisting and 4:06:39 partnership ads. Partnership ads is 4:06:40 where you're using an influencer or a 4:06:42 creator or someone that actually exists 4:06:44 out there and you're running the ads 4:06:45 through their page. Often you'll have to 4:06:47 pay them to run the ads through their 4:06:49 page if they have somewhat of a 4:06:50 following and they have leverage. 4:06:51 Whitelisting on the other hand is you 4:06:53 are taking a third-party page that you 4:06:55 can create and you're running ads to it. 4:06:57 So in this case, rather than Grunes 4:06:59 making another page called Grunes, 4:07:02 instead they might make a page called 4:07:04 fiber health magazine or something like 4:07:06 that and then they're running the same 4:07:08 ads. I personally would change them a 4:07:10 little bit to be more advvertorial and 4:07:12 orientated around this third party 4:07:13 framing, but you could run the same ads 4:07:15 and instead you're running it through 4:07:16 the handle fiber health. And now the 4:07:18 obvious advantage there is that it looks 4:07:20 like it's coming from a third party. It 4:07:21 looks less like you're being sold to. 4:07:23 It's a really good strategy that we use 4:07:24 across quite a lot of clients and it's a 4:07:26 way that we can decrease the meta ad cap 4:07:28 on the primary page. And then lastly, 4:07:30 you can also do regional pages. Now, if 4:07:33 I was running a brand, I would do all 4:07:34 three of these. I would have 4:07:35 whitelisting. I would have as many 4:07:37 partnership ads as I can. I would be 4:07:38 having multiple different primary pages 4:07:40 and I would be going into regional 4:07:42 pages, which is that when you sell in 4:07:44 multiple countries, you hit the meta ad 4:07:46 cap way faster. The reason being is that 4:07:48 generally you should segment your 4:07:50 campaigns based on country. And so you 4:07:52 will have a UK campaign, you will have 4:07:54 an Australia campaign, you will have a 4:07:56 USA campaign. If each of these has a 100 4:07:59 ads in each, you hit the cap. Even if 4:08:00 they're the same ad, you hit the cap, 4:08:02 which is really annoying. And so the way 4:08:03 that you fix this is you have a Grun UK, 4:08:05 Grooms Australia, Gruns USA, or whatever 4:08:07 your brand is. In that way, you have new 4:08:09 caps per country. And so you don't run 4:08:11 into this issue. So the third core claim 4:08:13 I made at the start of the video was 4:08:15 that there is no universal account 4:08:17 structure. So, what are the rules and 4:08:19 what are the mental models that you 4:08:20 should use to be able to think through 4:08:22 building your own account structure? 4:08:23 Well, there's two non-negotiables, which 4:08:26 is number one, data integrity. Now, what 4:08:29 this means is that the structure that 4:08:31 you use, more specifically, the 4:08:33 structure must be readable and 4:08:35 actionable. So, whatever you're running, 4:08:37 it needs to produce data that is 4:08:39 readable and actionable. Most people 4:08:42 create account structures, set up their 4:08:43 Meta account, set up their Google 4:08:44 account, Tik Tok account in ways that 4:08:46 aren't readable and therefore aren't 4:08:48 actionable. And so when we try to make 4:08:50 decision loops where we look at the data 4:08:52 and then we make a decision and then we 4:08:54 make time go by and we make another 4:08:56 decision, etc., this isn't a viable 4:08:58 option because the account wasn't set up 4:09:00 in a way that had data integrity. Sounds 4:09:02 really obvious, but you'd be surprised 4:09:04 if you don't think through this as a 4:09:06 core non-negotiable of the structure. 4:09:08 you end up leading yourself into a 4:09:09 position where you regret the way that 4:09:11 the account's structured and you have to 4:09:12 do a restructure. Then the number two 4:09:14 non-negotiable is commercial alignment. 4:09:16 The structure needs to actually fit 4:09:18 towards what the business sees and 4:09:21 thinks about its products margin and 4:09:23 goals. If a campaign ends up hiding a 4:09:26 product or hiding a category that needs 4:09:28 spend or it's going to turn into grade 4:09:30 inventory, that needs to be presented 4:09:32 within the account structure. If there's 4:09:33 a product portfolio with a low margin 4:09:35 base that actually can't support paid 4:09:37 media acquisition, well, that needs to 4:09:38 be reflected within the account 4:09:40 structure. If there's a particular goal 4:09:42 across a category that needs rep 4:09:44 prioritization or KPIing separately, 4:09:46 that also needs to be baked into the 4:09:48 account structure. There's five 4:09:49 questions you should be asking yourself 4:09:51 when you're building out the account 4:09:52 structure. Number one is how many 4:09:54 product categories do we have? The more 4:09:56 product categories, generally the more 4:09:57 segmentation will need to be introduced 4:09:59 if there's different goals across those 4:10:00 categories. Number two is how wide is 4:10:03 the margin spread? Is all the margin 4:10:04 relatively the same across the different 4:10:06 product categories or does it vary 4:10:08 substantially category to category? 4:10:09 Number three, do we have multiple 4:10:11 personas that don't overlap? Now, it's 4:10:14 fine to have personas that overlap and 4:10:15 are kind of the same person, but we're 4:10:17 tweaking the audience slightly. But if 4:10:19 we're talking about very different 4:10:20 personas that don't overlap, that will 4:10:22 need to be introduced to some degree 4:10:24 into the structure. How many regions are 4:10:26 we selling in? As I said before, when 4:10:27 you're selling in multiple regions, you 4:10:28 generally want campaign segmentation so 4:10:30 that you can control spend distribution 4:10:33 and sell through rates. If you're 4:10:34 holding inventory in particular 4:10:36 countries and they have their own 4:10:38 revenue targets, well then you need the 4:10:40 ability to be able to distribute spend 4:10:42 and the only way to do that is to 4:10:43 segment. And then lastly, what's the 4:10:45 creative throughput? If you're putting 4:10:46 10 ads in the account per month, there's 4:10:48 honestly not much segmentation that can 4:10:49 be done. If you're putting 2,000 ads in 4:10:51 the account per month, it's a very 4:10:53 different volume of creative that's 4:10:54 flowing through. So we need to think 4:10:55 about how is that creative actually 4:10:57 going to be introduced into the 4:10:58 structure methodically have spend 4:11:00 allocation and then give us the ability 4:11:02 to move it into a scaling structure. So 4:11:04 let's go through two actual examples. 4:11:06 Two different brands very similar in 4:11:07 terms of revenue similar in terms of 4:11:09 gross margin but the account structure 4:11:10 out of the back of it is going to look 4:11:12 very different. So we'll build out what 4:11:13 the account structure will look like for 4:11:15 brand A and brand B. So brand A is in 4:11:17 CPG. So they're selling consumables. 4:11:19 There's only one product. They're doing 4:11:20 5 million a year. 70% gross margin which 4:11:22 is fairly indicative of CPG. There's 4:11:24 only one persona so far, which is 4:11:26 actually a good thing. Most people at 4:11:28 this revenue level will try to squeeze a 4:11:29 bunch of different personas out, but 4:11:30 they're only at five mill a year, so 4:11:31 they only really need one persona, and 4:11:33 that's Busy Moms. And they sell an AU in 4:11:35 US. In my opinion, they should only be 4:11:37 selling an AU at this revenue level, but 4:11:39 they've decided to expand. Brand B is in 4:11:41 fashion, 50 SKUs, same revenue, slightly 4:11:44 lower gross margin, which is likely 4:11:46 indicative here of them needing to 4:11:47 discount constantly. Their gross margin 4:11:49 shouldn't be this low in fashion, but 4:11:51 it's normally a product of the fact that 4:11:52 they're taking a lot of product to 4:11:54 discount. They've got four personas. 4:11:56 This is fairly typical in fashion, 4:11:57 having three to four core personas, but 4:11:59 they're selling an AU NZ and UK. So, how 4:12:02 do these end up getting structured 4:12:04 differently? On brand A, we likely just 4:12:06 want one advantage plus cold campaign. 4:12:09 We want to be doing creative testing at 4:12:11 the adset level based on concepts for a 4:12:14 majority of spend should be going into 4:12:16 the one concept and the one persona 4:12:17 that's actually performing well at the 4:12:19 moment. There should be one of these for 4:12:20 the US and then we should have another 4:12:22 one for AU. We want existing customers 4:12:24 excluded from both of these cold 4:12:26 campaigns particularly in CPG where 4:12:29 there's hopefully going to be a lot of 4:12:30 returning customer revenue that's going 4:12:32 to overattribute into cold and the 4:12:34 frequency on these campaigns will end up 4:12:35 being driven up and it will just 4:12:36 retarget. So we want existing customers 4:12:38 excluded. doesn't mean we shouldn't have 4:12:39 any spend towards existing customers. 4:12:41 There's likely a bit of incrementality 4:12:43 here in allocating a little bit of spend 4:12:45 to them. And so, we're going to throw in 4:12:47 a retargeting campaign on existing 4:12:50 customers, but it's going to be at a 4:12:51 very low spend to the point that we just 4:12:53 want to keep frequency at below a seven 4:12:56 on a 30-day rolling period. So, this 4:12:58 will end up at this size of business 4:12:59 being a very small campaign, probably 4:13:01 spending $20 a day. So, that's brand A. 4:13:03 What about brand B over here? Due to the 4:13:06 higher complexity of SKUs, there's 4:13:09 probably multiple categories here. In 4:13:11 fact, there's three. And so, we're going 4:13:14 to build out campaigns for each 4:13:16 different category, particularly because 4:13:19 they have different purchase orders, 4:13:20 different revenue goals, and they need 4:13:22 to sell through all of the categories or 4:13:24 else they have to go to sale and a road 4:13:26 gross margin. So, they're going to have 4:13:27 a pants campaign, toss campaign, dress 4:13:29 campaign. Now, for AU NZ, we're just 4:13:32 going to consolidate this into one 4:13:33 campaign. So, we're going to target 4:13:35 Australia and New Zealand together. The 4:13:36 nice thing about New Zealand as a 4:13:38 pairing country is that it will 4:13:39 automatically max out at 7 to 10% of 4:13:41 total spend. So, we don't have to worry 4:13:43 about it overspending and most New 4:13:45 Zealand orders will always get fulfilled 4:13:47 from an Australian warehouse with 4:13:48 relatively fine shipping crazes. And so, 4:13:50 we don't have the need. Now, this is 4:13:53 already a lot of segmentation for a 4:13:54 brand of this size. And so, we also need 4:13:56 to be thinking through, do we really 4:13:58 want the UK in a separate campaign? 4:14:01 because that's going to mean we have six 4:14:02 campaigns now because we have to 4:14:04 duplicate the whole structure. We really 4:14:06 have two options here. Either option 4:14:08 number one is yes, we do that. But 4:14:11 rather than having six campaigns because 4:14:12 that introduces way too much 4:14:14 segmentation in an account of this size, 4:14:17 instead we move the segmentation down to 4:14:19 the adset level. So we have one 4:14:22 Australia campaign, one UK campaign, but 4:14:24 then at the adset level, we have pants 4:14:26 adsets, tops adsets, dress adsets, and 4:14:29 that's where the segmentation of 4:14:30 categories occurs, and we make sure it's 4:14:32 an so we can control budgets here. 4:14:35 Or option number two is we keep this 4:14:37 structure, but we just consolidate all 4:14:40 the regions into the campaigns. Now, why 4:14:44 or when would you want to do that? It's 4:14:46 if we are shipping the UK orders from an 4:14:49 Australian fulfillment center because in 4:14:51 that case it doesn't matter other than 4:14:53 our targets and our goals as a business 4:14:55 whether the UK revenue goes up, goes 4:14:58 down, goes sideways because we don't 4:15:01 have any holding costs of inventory in 4:15:04 the UK market. All of our stock is held 4:15:07 in Australia. So we just need to sell 4:15:09 through the stock regardless of where 4:15:11 we're actually selling to. Now, there is 4:15:13 a bit of complexity here because it's 4:15:15 fashion, which is that Australia is 4:15:17 southern hem, UK is northern hem. And 4:15:19 so, they're going to be different 4:15:20 seasons. And so, what's selling well in 4:15:22 Australia is not going to sell well in 4:15:24 the UK and vice versa, unless there's 4:15:26 transseasonal products. And so, in this 4:15:28 case, you are probably making purchase 4:15:30 orders for the UK season, in which case 4:15:34 you do need to move inventory. So, 4:15:36 there's this nuance in understanding 4:15:37 what does this region actually look like 4:15:39 within the business? Are we holding 4:15:40 inventory? Is there risk? do we need to 4:15:42 hit particular sales targets because 4:15:44 that information is then going to infer 4:15:47 into the account structure which is why 4:15:49 understanding the business and the 4:15:50 complexity and the goals is so important 4:15:51 when we're trying to translate into a 4:15:52 structure. I'm likely going to go with 4:15:55 consolidating this down having the 4:15:58 segmentation at an adset level and we 4:16:00 have an Australia cold campaign and then 4:16:03 we have the same thing in the UK. The 4:16:07 reason being as well is that the UK 4:16:10 likely sits on a different subdomain and 4:16:13 so we need all of the UK ads driving to 4:16:16 different URLs than where the Australian 4:16:18 ads are driving. And so because of that, 4:16:20 we need the separate campaigns and 4:16:22 therefore we're going to move the 4:16:24 category segmentation down to the adset 4:16:27 level. Then in fashion, what does do 4:16:30 incredibly well is dynamic product ads, 4:16:33 particularly in retargeting. So, we're 4:16:35 going to throw in a DPA retargeting ad 4:16:38 that's going to go to engaged audiences, 4:16:40 which is just like your 90-day website 4:16:42 visitors. And then we're also going to 4:16:44 throw in a final campaign, which is your 4:16:46 retargeting on existing customers, which 4:16:49 is going to contain two adsets. We're 4:16:51 going to have new arrivals in there. So, 4:16:53 we're constantly putting new arrivals in 4:16:54 front of the existing customer base. And 4:16:56 we're also going to have a DPA. We're 4:16:58 going to have it as two separate adsets 4:16:59 because if we consolidate all the spend 4:17:01 will just go to the DPA and we'll never 4:17:02 be able to actually serve new arrivals 4:17:04 to existing customers. These campaigns 4:17:06 are obviously because they're cold going 4:17:08 to have existing customers excluded 4:17:10 which is why we need this. Um targeting 4:17:12 existing customers in fashion is also 4:17:14 much more incremental than any other 4:17:15 category that we've seen. So it actually 4:17:17 is worth the spend allocation which is 4:17:19 why you have on the surface two brands 4:17:23 at the same revenue level. But when it 4:17:26 then comes to understanding the business 4:17:28 and all the complexities, the gross 4:17:30 margin profile, the personas, the 4:17:31 countries, the skew count, it then 4:17:34 translates into a very different account 4:17:36 structure. All right. So, if you're 4:17:37 spending sub 50k a month, I'll give you 4:17:39 a generic structure that most people can 4:17:40 get away with. Will it be ideal? Will it 4:17:42 be custom to the business? No. I've gone 4:17:44 through all of the reasons as to why you 4:17:46 should ask yourself those five 4:17:48 diagnostic questions to be able to 4:17:49 better understand how you should 4:17:51 structure yourself. But if you just want 4:17:52 something generic, if you need to throw 4:17:54 something up, here's what you should do. 4:17:56 Number one is a dedicated testing 4:17:58 campaign. Now, at this spend level, this 4:17:59 is probably all that you need. Now, what 4:18:02 this actually looks like is either an 4:18:04 or a CBO, and we'll go through later 4:18:06 which one you should actually choose 4:18:08 with adsets sitting under it. Three to 4:18:11 five creatives per adset. Now, you can 4:18:13 go way more than this if you have the 4:18:14 creative volume to support it. Just most 4:18:16 people at this spend level probably 4:18:17 don't. And the creatives are being 4:18:19 segmented based on concept. Now concept 4:18:22 for people who haven't watched any other 4:18:24 video of ours is the intersection right 4:18:26 here in the middle of an angle, an 4:18:29 offer, and a persona. And so you're 4:18:31 putting a persona, angle, offer 4:18:32 together, and then you're making 4:18:33 creatives under that. And that sits in 4:18:35 an adset. When you then go and make more 4:18:37 creatives for the concept, you have an 4:18:39 option. You can either launch it in the 4:18:41 existing adset or you can launch it in a 4:18:42 new one. In terms of which you do there, 4:18:44 if the adset is not performing, throw it 4:18:47 in. Try to get the adset to perform. 4:18:48 Throw more creatives at. If it is 4:18:50 performing, if it's hitting KPI, if it's 4:18:52 doing well, don't touch it. Number one 4:18:53 rule of meta media buying is never touch 4:18:55 something that's working. If it's 4:18:56 working, don't touch it. Do something 4:18:58 next to it. Don't ruin the thing that's 4:18:59 working. So, if it's working, don't 4:19:01 touch it. Launch the concept as another 4:19:03 adset. So, you might end up in a 4:19:04 position where you have 10 adsets and 4:19:06 six of them are for just one concept and 4:19:09 it's a bunch of new creative that you've 4:19:10 made over time and everything's working 4:19:11 because that ends up being the case is 4:19:13 one concept ends up outshining 4:19:14 everything else. Now, it's not that 4:19:16 important at this scale of business, but 4:19:17 you do want existing customers excluded. 4:19:20 We don't want our testing getting skewed 4:19:22 around based on existing customers 4:19:24 coming through a few ads and making 4:19:25 something look better than it actually 4:19:26 is. You also want to make sure that your 4:19:28 attribution setting is on 7-day click so 4:19:30 that you have data integrity in the 4:19:32 numbers that you're reading. Or else, 4:19:33 you also end up in a position where some 4:19:35 things might look better than others 4:19:36 just because it's claiming conversions 4:19:37 that has nothing to do with it. Then, 4:19:39 you can layer on a scaling campaign. Now 4:19:42 the idea here, the premise of having a 4:19:45 scaling campaign is that one of these 4:19:47 adsets performs well. So let's say this 4:19:49 adset at the top is performing well 4:19:51 above your target return of 5x and all 4:19:54 the other soft metrics of really good 4:19:56 too. You start scaling this up. You put 4:19:58 more spend more spend over time and 4:20:00 you're ramping it up and then eventually 4:20:02 return on ad spend drops to a point that 4:20:04 is unprofitable and so you need to pull 4:20:06 back spend. Now, when you do that, you 4:20:08 effectively find an equilibrium where 4:20:10 you find the daily spend level where 4:20:12 that creative can keep sitting there, 4:20:13 continue churning over results that 4:20:15 you're happy with. You can't push it any 4:20:16 further. Anytime you do, results drop. 4:20:18 So, you have to come back. And because 4:20:19 of that, you're kind of not happy with 4:20:20 it. You're like, "Ah, we could only take 4:20:22 this creative or this set of creatives 4:20:24 to $300 a day." But obviously, I want to 4:20:26 spend way more and I want to scale. So, 4:20:28 what do you do? It's at this point where 4:20:30 people often turn and go, "Well, we 4:20:32 can't scale this adset anymore. It's not 4:20:34 working. So, how can we take these 4:20:36 creatives and just launch them elsewhere 4:20:38 within the account structure to be able 4:20:40 to get more spend through them? And 4:20:41 that's effectively where scaling 4:20:42 campaigns are born, which is you take 4:20:44 creatives that have been maxed out. You 4:20:46 can't get any more spend through them, 4:20:47 and you go and just dump them in a new 4:20:48 campaign. And you hope that the new 4:20:50 campaign can get even more spend through 4:20:52 that ad. And a lot of the case, it 4:20:54 can't. For whatever reason, you take the 4:20:56 creative, you launch it in a new 4:20:57 campaign, you keep this on, keep this 4:20:59 spending, but you go and launch it 4:21:00 elsewhere, and you can likely get a 4:21:02 little bit more spend through that ad. 4:21:04 And so rather than the ad within the 4:21:06 account holding $300 a day, maybe you go 4:21:08 at this point and you layer it in on a 4:21:10 scaling campaign and a scaling campaign 4:21:12 can get it to hit. And so total spend 4:21:14 across the entire account on this asset 4:21:16 is now $400 a day because it's being 4:21:18 propped up by the scaling campaign. 4:21:20 That's the idea of why you've launched a 4:21:22 scaling campaign. There's absolutely no 4:21:23 reason to have one if you don't have ads 4:21:25 that are working well. like there's no 4:21:27 point in adding this complexity unless 4:21:28 you have creatives that are doing well 4:21:30 that have been cranked to their maximum 4:21:32 potential and then you're just trying to 4:21:33 squeeze even more out of it. A big 4:21:35 mistake I see is that people will have 4:21:37 something that works and they get it to 4:21:39 like $60 a day. So like a meaningless 4:21:42 amount of ad spend and then they'll go, 4:21:44 "Oh, we can't push it any further. Let's 4:21:45 roll it into scaling." This is nowhere 4:21:47 near enough spend to be able to start 4:21:49 trying to scale the ad even further. 4:21:51 Good way to conceptualize this is that 4:21:53 this media buying move of taking a 4:21:55 winning ad and then trying to get more 4:21:57 spend into it using an adjacent campaign 4:21:59 will get you on a good day an extra 20% 4:22:03 daily spend through the app. So if you 4:22:05 have an ad that's spending $60 a day, 4:22:07 this kind of move in the account is 4:22:09 going to get you like an extra $10 in 4:22:11 daily spend. Not worth it. Just make 4:22:13 better ads. Fix other things in the 4:22:15 business. There's probably a landing 4:22:16 page issue. There's probably an offer 4:22:17 issue. You still probably haven't found 4:22:19 a concept that's working. There's like a 4:22:21 million other things to put your effort 4:22:22 towards rather than trying to media buy 4:22:24 your way to an extra 10% of spend on a 4:22:26 really low base. If your ad spending 4:22:28 $400 a day actually worth just taking it 4:22:31 and putting it in a scaling, it's a 4:22:32 relatively loweffort move. It doesn't 4:22:34 take much time and yeah, you're going to 4:22:36 unlock an extra $80 per day in daily 4:22:39 spend. If you do this across like 10 ads 4:22:41 all at once, you take 10 of your highest 4:22:43 performing ads and you dump them in a 4:22:44 scaling campaign, it might allow you to 4:22:45 unlock an additional $1,000 day spend. 4:22:47 And that actually is worth it. That's a 4:22:49 decent unlock. it's worth having this 4:22:51 increased degree of segmentation and 4:22:53 complexity in the account. And then 4:22:55 lastly, you have the retargeting 4:22:56 campaign. This is absolutely an optional 4:22:58 campaign. When we talk about 4:23:00 retargeting, there's two different types 4:23:02 of retargeting. There's website visitor 4:23:04 retargeting. So, effectively warm 4:23:06 audiences, people that have shown 4:23:07 intent, they've taken some kind of 4:23:09 action, but they haven't purchased. And 4:23:10 then we have existing customer 4:23:12 retargeting. So, this is someone that's 4:23:14 actually purchased from us. They're all 4:23:16 the way down at most aware in the stages 4:23:18 of awareness cuz they're a customer and 4:23:20 we're trying to effectively get them to 4:23:21 buy a second time or a third time. Now, 4:23:23 for most accounts, 90% of them, you 4:23:26 don't need a website visitor retargeting 4:23:29 campaign. And the reason being is that 4:23:31 the testing or the scaling or whatever 4:23:33 currently exists will allocate a good 4:23:36 amount of spend anyway towards 4:23:38 retargeting website visitors. And you 4:23:40 see this by doing an audience segment 4:23:42 breakdown within the account. So, if you 4:23:44 hit breakdown, you hit audience 4:23:45 segments, you'll be able to see as long 4:23:47 as this is set up within your advertiser 4:23:50 settings in your audience segments. So, 4:23:51 make sure that's set up. You'll be able 4:23:52 to see spend, rorowaz, etc. on new 4:23:56 audiences as well as engaged and then 4:23:58 existing. Now, you shouldn't have any 4:24:00 spend to existing cuz they should be 4:24:01 excluded, but you'll be able to see your 4:24:03 spend towards engaged, which is your 4:24:04 website visitors. And there'll likely be 4:24:06 enough spend here at a high enough 4:24:08 frequency that you don't need a 4:24:10 dedicated campaign to force more spend 4:24:12 through that elites. It's okay on 4:24:13 existing purchases because you have it 4:24:15 excluded here. You probably want some 4:24:18 spend towards existing customers. It 4:24:20 just depends on the retention dynamics 4:24:21 in the business, the category, and the 4:24:23 stage of business that you're at. If 4:24:25 you're a new business, which is sub 50K 4:24:26 a month, you probably are. You're also 4:24:28 probably a small business. You don't 4:24:30 have that many existing customers. And 4:24:31 then depending on the product portfolio 4:24:33 and the retention portfolio, so is this 4:24:35 furniture where you're like going to 4:24:36 have no repeat purchases really. It's 4:24:38 very low. It's very infrequent. or is 4:24:40 this supplements low average order value 4:24:42 which has super high repeat rates in its 4:24:43 subscription model. Okay, well different 4:24:45 story. We probably want some spend 4:24:46 there. So you need to contextualize this 4:24:48 to the actual business model and the 4:24:49 expectations on repeat purchasing. If 4:24:51 there's an expectation of high repeat 4:24:52 purchasing, put some existing customer 4:24:54 spend in there. Keep it low, keep it 4:24:56 controlled. Look at frequency as the 4:24:57 measure. If you don't expect repeat 4:24:59 purchases, don't have. Now, the reason 4:25:01 why this structure works at this spend 4:25:03 level is number one, it's consolidated. 4:25:05 You probably have not a lot of 4:25:06 conversion volume at this spend level. 4:25:08 And so a product of that means that you 4:25:10 shouldn't have a lot of segmentation. 4:25:11 You want all the conversions 4:25:12 consolidated up. You need structured 4:25:15 creative testing at this spend level 4:25:16 because you probably don't have anything 4:25:18 that's a very large winner or else you'd 4:25:20 be spending more. And so you want an 4:25:22 structure where you can force spend 4:25:24 through adsets and start to learn stuff. 4:25:26 And then number three, at this spend 4:25:28 level as well, you probably don't have a 4:25:29 large product portfolio. You probably 4:25:31 don't have too much complexity in the 4:25:34 business model as it stands. And so we 4:25:36 can keep things relatively simple. In 4:25:37 terms of KPIing this structure on the 4:25:40 testing campaign, you want to KPI at the 4:25:42 adset level. Did the adset hit our 4:25:44 target cost per acquisition or our 4:25:46 target return on ad spend over a roll-in 4:25:49 window, which is reasonable. Okay? And 4:25:51 what window in which you read data is 4:25:53 dependent on conversion volume. So you 4:25:54 can't look at three-day windows if 4:25:56 you're only doing three conversions in a 4:25:57 3-day window. You can look at two-day 4:25:59 windows if you're doing a thousand 4:26:01 conversions a day. So, it's all based on 4:26:03 contextualizing uh the windows that 4:26:04 you're looking at for performance based 4:26:06 on how much volume the business is 4:26:07 actually doing. At this kind of scale, I 4:26:09 wouldn't really be reading data on any 4:26:11 shorter than a 5day period. And then if 4:26:13 the adset is hitting KPI, budgets go up 4:26:15 in 20% increments per day. If it's not 4:26:17 hitting KPI, we diagnose why. What do we 4:26:19 think was wrong in the creative? Let's 4:26:21 then go and make more ads, more 4:26:22 concepts, and let's either wind that 4:26:24 adset down or turn it off completely. On 4:26:26 the scaling campaign, you want to just 4:26:28 be KPIing all the way up at the campaign 4:26:29 level. There isn't any adset 4:26:31 segmentation on a scaling campaign, at 4:26:33 least at this level. And then on 4:26:35 retargeting, you want to be KPIing based 4:26:37 on incremental return on ad spend by 4:26:38 doing an attribution breakdown. And also 4:26:40 frequency shouldn't jump to above a 4:26:42 seven on a 30-day window. The four 4:26:44 common mistakes at this level of spend 4:26:46 is that people test inside the scaling 4:26:48 campaign. There's no exclusions on the 4:26:51 scaling campaign. They kill testing way 4:26:53 too fast. So the window in which they're 4:26:55 looking at conversion data is way too 4:26:56 little. And then they set budgets on the 4:26:58 adset level based on the 50 conversions 4:27:01 in a 7-day window rule, which ends up 4:27:03 with just way too much budget for their 4:27:05 particular business. So you actually 4:27:06 need to back propagate budgets at the 4:27:08 adset level based on your average order 4:27:10 value and expected CPA. So let me 4:27:12 quickly break that down for you so you 4:27:13 understand what kind of budgets you do 4:27:15 need to be setting on the adset level. 4:27:16 So when you're looking through adset 4:27:18 level spending, there's really two 4:27:19 factors that you need to think about. 4:27:21 You need to think about time to outcome 4:27:23 of the test and then you need to think 4:27:26 about total budget required for the 4:27:29 test. So what does that actually mean? 4:27:30 Well, let's say that you have a target 4:27:33 cost per acquisition. So you want to be 4:27:35 getting customers at $50 per customer. 4:27:37 You then come up with a new concept. You 4:27:39 make 10 ads under it. We then go and 4:27:41 launch it at the adset level. What we 4:27:43 really want to figure out as quickly as 4:27:45 possible is is that concept working or 4:27:47 not? Now we need to set a barrier or a 4:27:49 threshold for how much spend are we 4:27:52 going to put through those ads through 4:27:53 that concept before we decide yes or no. 4:27:56 Now the general rule of thumb that 4:27:57 almost everyone uses is you take your 4:27:59 target cost per acquisition. So this is 4:28:01 our target. You times it by three and 4:28:04 that is how much you should spend before 4:28:06 you decide whether to keep the ad on or 4:28:07 off. If you get three conversions in 4:28:09 this time, if you get four, even better. 4:28:11 Keep running it. If you get two, uh, 4:28:13 yellow light, let it run a little bit 4:28:14 longer. Let it run another $50 and then 4:28:16 we'll decide. If you get zero or one, 4:28:18 cut it. Now, that as a rule of thumb is 4:28:20 pretty decent. Now, I would always let 4:28:22 it spend a little bit more. I'm always 4:28:24 willing to be a little bit more gracious 4:28:26 because I know that the initial spend is 4:28:28 really at trying to learn and figure out 4:28:29 who to target. And then once it figures 4:28:31 it out, the ball gets rolling and you 4:28:32 actually see your efficiency client. And 4:28:34 so, I actually prefer this rule if 4:28:36 possible. And if the client's okay with 4:28:37 it, they're more of a times five. So, we 4:28:39 wait till we spend 250 and then we make 4:28:42 a concrete decision. There's no yellow 4:28:43 lighting. It's either a yes or a no 4:28:45 because we have enough spend volume 4:28:46 here. Okay, cool. So, let's say that 4:28:48 $250 is what we want to spend. So, we 4:28:50 know the total budget of the test. We're 4:28:52 going to spend $250 on these new 4:28:54 creatives before we decide whether 4:28:55 they're a success or not and whether we 4:28:56 turn them off. Then, we have time. This 4:28:59 is how quickly do we want the test 4:29:01 outcome? Because technically, we could 4:29:03 just set $250 a day as the budget and 4:29:06 we'll know whether this adset in this 4:29:07 group of creatives has worked today. 4:29:09 We'll know by the end of the day. Now, 4:29:11 the issue with that is that there are a 4:29:13 couple components that will play into 4:29:15 the time duration that you need to allow 4:29:17 a test to run for. Number one is daily 4:29:20 seasonality. The reality is is that 4:29:22 there is going to be some days of the 4:29:23 week in which you do better and some 4:29:25 days of the week in which you do worse 4:29:27 as a business. If you go and launch a 4:29:28 test on a good day, you will see better 4:29:31 results. If you go and launch it on a 4:29:32 bad day, you will see worse results 4:29:34 irrespective of the actual test 4:29:36 variable, which is the creative. And so 4:29:38 if we run tests too quickly, we don't 4:29:40 get the ability to encapture the whole 4:29:42 seasonality of the week and therefore we 4:29:44 can get a biased view of performance. 4:29:47 The second thing that impacts time is 4:29:49 time to purchase. And so for a lot of 4:29:51 brands, people don't instantaneously 4:29:53 purchase the first time they ever see an 4:29:55 ad from. Okay? You don't just serve an 4:29:56 ad to someone and they go, "Oh, great 4:29:57 buy." Normally it takes two impressions, 4:30:00 three impressions, four impressions, 4:30:02 maybe a couple clicks, and then over the 4:30:04 course of a 3 to 4 day period of warming 4:30:06 that user up, then they make the 4:30:08 purchasing decision. Now, if we run this 4:30:09 test in one day, well, guess what? We're 4:30:11 going to show the ads to a bunch of 4:30:12 people. They might go, "That's amazing. 4:30:13 These ads are great. I'm now 4:30:15 interested." But then you kill the ads, 4:30:17 you turn the test off, other ads go and 4:30:19 target those users in retargeting. They 4:30:21 all end up converting to other creative, 4:30:24 but you killed the primary top ofunnel 4:30:26 ad that you want to test it because you 4:30:27 only gave it one day. And so you need to 4:30:29 not only give it a little bit of time 4:30:31 because of daily seasonality that might 4:30:33 skew the results, but also because 4:30:34 people take a little bit of time to buy. 4:30:36 True. And this applies not only to 4:30:37 Facebook, but Google and all the other 4:30:38 platforms as well. So because of that, 4:30:40 we need to balance these two variables. 4:30:42 We need to figure out what is the budget 4:30:43 for the test. And then how much time do 4:30:46 we want as a feedback loop? Ideally, as 4:30:48 fast as possible. We want to know the 4:30:49 outcome quickly so we can continue to 4:30:51 iterate because ultimately the speed of 4:30:53 growth in a business is a function of 4:30:55 the speed of cycles of feedback and 4:30:58 learning and iteration. And so we want 4:31:00 to iterate and learn as fast as possible 4:31:01 so that we can grow. But we want to give 4:31:03 it enough time that it encapsulates 4:31:04 daily seasonality and time to purchase. 4:31:06 What does that often end up coming out 4:31:07 to? One week. One week is generally a 4:31:10 good duration to run a test for. Now if 4:31:11 you're a big business with more 4:31:12 stability with not a long time to 4:31:14 purchase and cut that down to three to 4:31:15 four days. If you're a tiny business 4:31:17 with tiny budgets with a incredibly long 4:31:19 time to purchase and massive daily 4:31:22 seasonality, well, okay, maybe we need 4:31:23 to extend that to 10 to 14 days. Okay? 4:31:25 So, all once again, always going to be 4:31:27 respective to your particular business, 4:31:29 which is why in this whole video and all 4:31:31 the content that we put out, we talk in 4:31:33 frameworks rather than actuals. Rather 4:31:35 than saying this is exactly what you 4:31:37 should do, we say here's how you should 4:31:39 think through the problem. Here's the 4:31:40 two variables that impact it. Now think 4:31:42 through the problem yourself in your own 4:31:44 context and then come up with your own 4:31:46 solution because every solution, every 4:31:48 budget, every adset is different for 4:31:50 every single business. Now as you jump 4:31:51 to 50 to 250k a month in ad spend, if I 4:31:54 was to prescribed you with a campaign 4:31:56 structure, which once again we don't 4:31:58 like doing cuz different based on all 4:31:59 the commercial objectives of the 4:32:00 business, but if I was to, you still 4:32:02 have the retargeting campaign, this is 4:32:03 likely still going to be just existing 4:32:05 customers. Very rare you need to layer 4:32:06 in website visitors here. You might, 4:32:08 probably unlikely. And then in terms of 4:32:09 the cold campaigns, this just builds out 4:32:11 a little bit further. So you're going to 4:32:12 end up with probably more adets at the 4:32:14 adset level cuz you're going to be doing 4:32:16 more creative testing and likely have 4:32:18 more creative volume, which is going to 4:32:19 introduce more adset volume. And then 4:32:21 potentially you're also going to extend 4:32:22 out to two maybe three campaigns. Now, 4:32:24 if we're looking at this as if it's a 4:32:25 CPG brand with one skew, this campaign 4:32:28 segmentation is actually going to come 4:32:30 from different funnels and different 4:32:31 approaches. And so you might have a 4:32:33 funnel or a persona or an approach 4:32:35 depending on the vernacular that you 4:32:36 want to use to explain it. But you might 4:32:38 have a funnel that is targeted at old 65 4:32:42 plus year olds for arthritis, right? And 4:32:45 this is a supplement that we can pivot 4:32:47 for that. In fact, let's just call it 4:32:48 fish oil, right? Fish oil applies for 4:32:50 old arthritis. All the ads as well are 4:32:52 going to be mirroring this type of style 4:32:54 that works with this age demographic, 4:32:55 which is going to be VSSLs. It's going 4:32:57 to be native statics. It's going to be 4:32:58 testimonials from an older demographic. 4:33:00 and then actually probably throw on like 4:33:02 TV style ads as well that you're 4:33:04 chopping into UGC format. Then you might 4:33:07 have the product repositioned. All of 4:33:09 the ads are now for young people 25 to 4:33:12 35 focused on like brain health or 4:33:16 performance at work or something 4:33:17 orientated around that. Now you could 4:33:18 also get this funnel working. The 4:33:20 landing pages are going to look very 4:33:21 different because you want different 4:33:22 type of demographics and photos on the 4:33:24 landing pages. Um the ads are going to 4:33:25 look very different. Probably the 4:33:27 profiles that you even run this through 4:33:28 might look very different as well. This 4:33:30 might be heavier on whitelisting. This 4:33:32 might be on the native page. But because 4:33:34 this is almost two different businesses 4:33:36 in itself because the funnels look very 4:33:38 different as a product, this is two 4:33:39 separate campaign. So that's what that 4:33:40 would look like in CPG with one skew. If 4:33:43 we're talking about like fashion with a 4:33:44 bunch of SKs, this just ends up being 4:33:46 some kind of category segmentation. It 4:33:48 might be new arrivals segmentation. So 4:33:50 you have your normal existing cold 4:33:52 campaign that you had at lower spend, 4:33:53 but now you just go and bolt on a new 4:33:55 arrivals campaign um for the sake of 4:33:58 pushing new arrivals and having higher 4:33:59 sellrough rates here. This might be a 4:34:01 particular product category. When we do 4:34:03 an LTV analysis, we find out that this 4:34:05 particular product category, let's call 4:34:07 it pants, ends up with not only a higher 4:34:09 average order value on first purchase, 4:34:10 but much better repeat rates and 4:34:12 retention. And so as a product of that, 4:34:14 we make a second campaign just for pants 4:34:17 where we want to have a higher cost per 4:34:20 acquisition KPI. So we can be more 4:34:22 aggressive in acquiring customers 4:34:23 through here because we know that our 4:34:25 pants category acquires really high 4:34:27 quality customers. And so as a product 4:34:29 of that, we get a second campaign. So 4:34:31 generally when you're going from the 4:34:32 sub50k range and then bridging into 50 4:34:34 to 250k. What ends up happening if 4:34:37 you're a good media buyer or a good 4:34:39 performance marketer is additional 4:34:41 campaigns will start getting layered in 4:34:42 that's has a very specific reason or 4:34:45 commercial objective that aligns with 4:34:47 the business's fundamentals. If you're 4:34:49 just layering in more campaigns for like 4:34:51 the sake you're like ah we're spending 4:34:52 100k a month right now. One campaign 4:34:54 seems weird. Let's do two let's do 4:34:55 three. you're breaking the fundamentals 4:34:57 that I went through before, which is 4:34:58 that every segmentation decision needs 4:35:01 to be done with the ability to read data 4:35:03 and make decision loops through an 4:35:06 underlying structure that has data 4:35:07 integrity. So, if you're just 4:35:08 introducing segmentation for the sake of 4:35:09 it, you're breaking all the rules that 4:35:11 we've gone through in this case, you 4:35:12 want to look at segmentation, but only 4:35:13 if it makes sense to better align with 4:35:15 the business's objectives. The main key 4:35:17 at this campaign level and where the 4:35:19 real skill unlock is and where some 4:35:20 people just smash straight through this 4:35:21 spend level and some people stay here 4:35:23 for a very long time is in continuous 4:35:25 testing of different concepts so that 4:35:28 you can find the 20% of concepts that 4:35:30 are going to do really well and are 4:35:31 going to be able to hold hundreds of 4:35:33 thousands of dollars a month in ad 4:35:35 spend. Now we have a YouTube video 4:35:37 called meta ads creative strategy in 4:35:39 2026 the full system where we spend 4:35:41 literally 30 minutes talking about how 4:35:43 to create concepts. So, I strongly 4:35:45 recommend you go and watch that video. 4:35:47 But if you're unfamiliar with the 4:35:48 concept, one minute run through. You 4:35:50 want to be building out personas, 4:35:52 angles, offers, and then ad types, 4:35:54 building this matrix out, and then 4:35:56 having your ad sets reflect this 4:35:58 structure, and then continue to test ads 4:36:00 under each concept. An example of this, 4:36:02 if you're selling teeth whitening, is 4:36:03 that a persona could be coffee drinkers. 4:36:05 The angle is whitening without 4:36:07 sensitivity issues. The offer is just 4:36:09 the product. There's nothing special 4:36:11 there. And then the ad type is 4:36:12 userenerated content. And obviously 4:36:14 under this ad set, if it's coffee 4:36:16 drinkers, whitening without sensitivity, 4:36:17 we can rotate in tons of different ad 4:36:19 types. Like the ad type is flexible. 4:36:20 This is infinite. The next angle is 4:36:22 brides. And you go, this is very 4:36:24 different, right? Coffee drinkers into 4:36:25 brides. How do brides relate to teeth 4:36:27 whitening? Well, we can frame this 4:36:28 around get ready for your wedding in 14 4:36:30 days. We can do a bundle offer that's 4:36:32 specifically curated for fast 14-day 4:36:35 turnaround to get white teeth for the 4:36:37 wedding. And then we can do this to a 4:36:39 testimonial of an actual person that 4:36:40 just had a wedding in the dress. They 4:36:42 can do a before and after. So they can 4:36:43 document the whole process. And the real 4:36:44 key here and where this really 4:36:45 accelerates is that once you get 4:36:47 something that's working, let's say this 4:36:48 bride idea does well and this ad does 4:36:50 really well. Well, you double down. You 4:36:52 start making a ton of creative around 4:36:54 this concept and you have a custom 4:36:56 landing page that has continuity through 4:36:58 it. So we might actually spin this 4:37:00 bundle off and call it like the wedding 4:37:02 in 14day bundle. have an own custom 4:37:04 landing page, have all these 4:37:05 testimonials on the landing page, build 4:37:07 the entire funnel around this angle, and 4:37:09 then this can scale to 2, 3x the volume, 4:37:11 and we can really saturate this market 4:37:13 while at the same time continuing to 4:37:15 work on other angles and other persona. 4:37:16 And the last one I have here is an 4:37:17 obvious one, which is smokers to remove 4:37:19 stains. You give them a subscription 4:37:21 offer because they're continuing to 4:37:22 smoke, so they'll have this issue 4:37:23 forever. And so you want to put them on 4:37:24 an offer that's relevant to that. And 4:37:26 then this is through founder head 4:37:27 talking content because maybe the 4:37:29 founder is a smoker and that's why they 4:37:30 started the business. So, the reason why 4:37:31 concept level testing becomes so 4:37:33 important at this tier of spend is 4:37:35 because each campaign should be 4:37:37 generating enough spend volume, enough 4:37:38 conversion volume to be able to support 4:37:40 a degree of segmentation at the adset 4:37:42 level. It allows you to then start to 4:37:44 KPI based on concept rather than just 4:37:46 the campaign as a whole, which allows 4:37:48 for better directional feedback in 4:37:50 creative. And this is really a core key 4:37:52 here, which is that this creates a 4:37:54 creative strategy because you have the 4:37:57 strategy, you come up with all the 4:37:58 creative, but then most people just 4:38:00 throw it in the account, and then they 4:38:01 just make more creative, and they throw 4:38:03 it in the account, and they make more 4:38:04 creative, but they don't actually look 4:38:06 at the account and go, "How is stuff 4:38:07 performing? How are we reading this 4:38:09 data? How are we drawing insights?" And 4:38:11 then, how is that informing the next 4:38:13 batch of creative that we're making? And 4:38:14 so the creative refreshes become 4:38:17 targeted, the losing concepts get more 4:38:19 attention, and the winning concepts get 4:38:22 more volume. You can also then begin 4:38:24 directing resources upstream to landing 4:38:27 pages, retention flows, product 4:38:30 portfolio expansion based on concept 4:38:33 level testing. As an example, let's say 4:38:35 that the bride's persona starts 4:38:37 performing incredibly well. What do we 4:38:38 do? We increase creative volume. We 4:38:40 create a custom landing page. We change 4:38:42 the offer accordingly. and potentially 4:38:44 as a byproduct of changing the offer. We 4:38:47 expand the product categories to meet 4:38:49 this audience as we're generating so 4:38:51 many existing customers through this 4:38:52 funnel. And so we might have a retention 4:38:54 win back flow off the back of this that 4:38:56 is post wedding. We send them a free 4:38:59 gift saying congrats on getting married. 4:39:01 Here is an additional offer for you that 4:39:03 pulls them into a different product line 4:39:06 that meets them at the stage of life of 4:39:07 where they are. So you can start to get 4:39:09 really creative in terms of the backend 4:39:11 retention, flow, sequencing, and product 4:39:13 portfolio based on direct feedback of 4:39:15 what type of customers are we acquiring 4:39:17 and from where. If you end up keeping 4:39:19 your concept targeting super broad and 4:39:21 you don't niche down in this way, you 4:39:23 don't get accurate persona insights and 4:39:25 that then bleeds into inefficiencies in 4:39:28 the rest of the business. So this is 4:39:29 really also telling you who your 4:39:30 customer is. Who is the customer? Who 4:39:32 are we acquiring? And in what percentage 4:39:34 allocations are they? Are we getting 40% 4:39:35 of our customers because they're 4:39:37 smokers? 40% because they're about to 4:39:39 get married, etc., etc., and then we can 4:39:40 start to craft the entire business 4:39:42 strategy around this. The last reason 4:39:43 why concept testing becomes critical at 4:39:45 this scale is it supports a large volume 4:39:48 of ads. So, at this spend level, you 4:39:50 should be launching really between 100 4:39:52 to 300 new ads per month if you want to 4:39:55 grow. If you don't want to grow, don't 4:39:56 launch that many ads. But if you do want 4:39:58 to grow, you should be launching around 4:40:00 about this volume. This volume becomes 4:40:02 actually quite sustainable for entering 4:40:05 into an account structure if you have 4:40:07 built out to let's say two campaigns 4:40:09 with 10 adsets within each campaign 4:40:11 because that's 20 adets which means on 4:40:13 the low end you've got five ads per 4:40:15 adset. On the high end you've got 15 ads 4:40:17 per ads set. Very reasonable. In fact, 4:40:19 you could actually have less campaigns, 4:40:20 less adsets and you're fine here. The ad 4:40:22 to adet ratio is completely manageable. 4:40:26 So what then happens at 250k plus? 4:40:28 pretty much everything at 50 to 250, 4:40:30 just more of it. And so when we come to 4:40:32 campaign segmentation at this budget, 4:40:34 and this is a enormous budget range 4:40:36 here, cuz I'm saying 250 all the way up 4:40:38 to $20 million a month in spend. Okay? 4:40:40 So it's an enormous range and therefore 4:40:42 there's an enormous degree of complexity 4:40:44 difference. And so because of that, 4:40:45 unlike the other two sections, I'm not 4:40:47 going to give you an exact account 4:40:48 structure that you should run because it 4:40:50 would just be ridiculous. It won't be 4:40:51 applicable to anyone at this spend 4:40:52 level. and instead I'm going to give you 4:40:54 a bunch of strategies, a bunch of 4:40:55 advice, a bunch of tips, and then I'm 4:40:57 going to run through an example so that 4:40:59 you can get as much value as possible on 4:41:00 how you should be structuring at this 4:41:01 spend level. Number one is you do want 4:41:04 to potentially consider layering in a 4:41:06 second ad account. Now, this strategy 4:41:08 changes a lot and what I would have said 4:41:10 6 months ago is different from what I'll 4:41:12 say today and what I say today will be 4:41:13 different in 6 months. And so, I don't 4:41:15 want to give you too much tactical 4:41:17 application here because it'll just be 4:41:18 outdated. But the idea here is that you 4:41:20 run the same pixel but you run different 4:41:22 bid logic. So if you're running maximize 4:41:24 conversions in the main account, you 4:41:26 might run maximize conversion value or 4:41:28 you might run cost caps or big caps in 4:41:30 this account on the same page. It could 4:41:31 be different page as well. And the idea 4:41:32 is that you'll start winning different 4:41:34 auctions from this ad account that the 4:41:35 primary account just isn't bidding on. 4:41:37 And so it will allow you to get more 4:41:39 volume through effectively the same 4:41:42 creative and the same pixel and the same 4:41:44 page, but you're entering different 4:41:45 auctions due to it being in a different 4:41:46 ad account. The disadvantage that you 4:41:48 could argue is it might increase CPMs 4:41:50 and ad costs for you because you might 4:41:51 be cross bidding against yourself. Now, 4:41:52 if you're using the same page and the 4:41:54 same pixel, you shouldn't really cross 4:41:55 bid that much, but it's definitely an 4:41:57 argument and no one has really been able 4:41:58 to prove whether this is the case or 4:42:00 not. The main other obvious advantage 4:42:01 here is that it derisks the business 4:42:04 enormously because if you have a second 4:42:06 ad account and the primary ad account 4:42:08 gets banned or the billing method goes 4:42:10 down for whatever reason, you don't just 4:42:11 lose all your new customer acquisition 4:42:13 in the business if you're overrelyant on 4:42:15 meta ads. Instead, you have two ad 4:42:16 accounts. One might only hold 10% of the 4:42:18 spend of the other, but if the main one 4:42:20 goes down, you can just crank up the 4:42:21 second one, and it prevents a doomsday 4:42:24 scenario where you might not have any 4:42:25 revenue for a week. Number two is that 4:42:27 at this spend level, you absolutely need 4:42:30 a page strategy. As I said earlier, 4:42:33 you're going to hit your ad limit 4:42:34 without a doubt. This is a guarantee. 4:42:36 And so, what is the strategy to be able 4:42:39 to avoid that from happening? And I gave 4:42:40 you the three options earlier in the 4:42:42 video. Number three is that you will 4:42:44 generally start to get complexity 4:42:46 getting introduced in more media buying 4:42:48 tactics at this level of spend. And this 4:42:50 is the level of spend where it actually 4:42:52 does start to make sense to be playing 4:42:53 around with the 3 4enters because 3 4% 4:42:57 when you're spending a million dollar a 4:42:58 month is actually quite mature and it 4:43:00 could add an enormous amount to bottom 4:43:01 line profit. And so having someone 4:43:03 dedicated on trying to media buy your 4:43:05 way to more efficiency is genuinely 4:43:07 worth the investment. This is where like 4:43:09 bidding complexity will start to be 4:43:10 introduced. You're not just running 4:43:11 maximize for conversions in the account 4:43:14 across everything, but you might start 4:43:15 introducing bid caps or you might start 4:43:17 introducing cost caps or you might have 4:43:19 maximize conversion value or target 4:43:20 rorowaz that's sitting next to these 4:43:22 campaigns. And the reason being is that 4:43:24 each different bidding strategy will 4:43:26 enter auctions differently with 4:43:28 different bids and you will generally 4:43:29 win more auctions as you start to 4:43:32 diversify the bidding strategies within 4:43:34 the account. The reason why I almost 4:43:36 never talk about this in any content is 4:43:38 because 99.9% of people are not spending 4:43:41 over 250k a month and so should not be 4:43:43 concerned whatsoever with bidding 4:43:45 strategies. Now counter to that 60% of 4:43:48 our client portfolio spends more than 4:43:50 250k a month. So for us internally the 4:43:52 bidding mix actually is a big deal and 4:43:54 this actually is something that we need 4:43:55 to think about and think through but for 4:43:57 most people ignore it. What will also be 4:43:59 the case at this spend level is you'll 4:44:01 generally have some kind of 4:44:02 international market expansion in which 4:44:04 case you need to start thinking through 4:44:06 the complexity that gets introduced into 4:44:08 the ad accounts from that international 4:44:09 expansion. Whether you run secondary ad 4:44:12 accounts for different regions, how you 4:44:14 deal with pages, how you actually deal 4:44:15 with the backend on Shopify or however 4:44:17 you're hosting the different regions. A 4:44:19 general word of advice is that I would 4:44:22 advise against segmenting countries out 4:44:24 at an ad account level. And the reason 4:44:26 being from an agency perspective is that 4:44:29 it will increase your costs because 4:44:31 working across multiple different ad 4:44:32 accounts increases labor dramatically. 4:44:34 And so I would always rather work on one 4:44:37 ad account than working across seven ad 4:44:39 accounts which we have some clients that 4:44:40 have seven ad accounts for seven 4:44:42 different regions. And it adds so much 4:44:44 additional labor and complexity into the 4:44:47 management across it. Now the reason why 4:44:48 you would have all those ad accounts is 4:44:50 really only one reason and it's that you 4:44:53 want to get build in the local currency 4:44:55 of that region. So if you have a US web 4:44:58 presence, you want the ad account to 4:45:00 bill you in USD because maybe you have a 4:45:01 US bank. If you have a UK presence, you 4:45:03 want all of your money flowing through 4:45:05 in great British pounds and so you need 4:45:06 a separate ad account. That's really the 4:45:08 only convincing argument I have seen for 4:45:10 introducing segmentation. Other than 4:45:12 that, everything else is solvable. You 4:45:14 could say, "Oh, reporting is better 4:45:15 because we can plug these ad accounts 4:45:16 into our dashboards and reporting." 4:45:18 Yeah, but you can just add filter rules 4:45:19 based on country segmentation, or you 4:45:21 can just add filter rules at a campaign 4:45:22 name level, and all of that's solved. 4:45:23 Like, you don't need to introduce ad 4:45:25 account complexity other than for the 4:45:27 reason of just getting built-in local. 4:45:29 Now, let me give you a worked example of 4:45:31 an actual real ad account that's 4:45:32 spending 25K per day, which is about 4:45:35 750K a month. So, closing up on a 4:45:37 million a month, and this performs 4:45:39 incredibly well for them. Now, note this 4:45:40 will not guarantee results for you 4:45:42 because your business has its own 4:45:44 complexities and its own differences and 4:45:45 you should think through everything that 4:45:46 we've gone through so far in terms of 4:45:48 how to structure it. Or you can 4:45:49 obviously always click the link in the 4:45:50 description, reach out to us. We will do 4:45:52 a free audit as long as you're doing at 4:45:53 least $5 million a year in revenue and 4:45:55 we can walk you through what that 4:45:56 account structure might actually look 4:45:57 like and we can provide it to you. So, 4:45:59 we have a testing campaign at the top. 4:46:01 It's an So, all tests are still 4:46:03 being done at the adset level. I have 4:46:05 actually seen accounts that are spending 4:46:07 $300,000 a day in budget and they're 4:46:10 still running tests. So I see a lot 4:46:12 as push back, oh you shouldn't run 4:46:13 testing once you're actually a big 4:46:15 account and you're spending a lot. 4:46:16 That's not true at all. Okay? Like you 4:46:17 can run CBO or It's up to you and 4:46:20 there's benefits of each one. It just 4:46:21 depends on how you want to manage the 4:46:22 account and also what the particular 4:46:24 nuances are of that business. ABOS at 4:46:26 this spend level work and you can 4:46:28 perform incredibly well and there's 4:46:29 reasons why you want to do them because 4:46:30 you want to force spend through new 4:46:31 tests. and CBOS can also work incredibly 4:46:33 well at this spend level. So this is 4:46:35 once again up to you, but this ad 4:46:36 account runs a testing campaign as an 4:46:38 Number two is then a scaling 4:46:39 campaign, which is a CBO, one adset, 4:46:42 cost caps. Every 1 to two weeks, the top 4:46:44 performers in the the post IDs are 4:46:46 taken and they're launched into the 4:46:48 scaling campaign. Now, they're not 4:46:49 turned off in the testing campaign. The 4:46:51 testing campaign is also used to scale 4:46:52 at the adset level. There's adsets in 4:46:54 here spending multiple thousands of 4:46:55 dollars a day. You still scale in the 4:46:57 testing campaign. It's just this is a 4:46:59 strategy to try to squeeze more out of 4:47:01 an existing post that's doing well. 4:47:03 Third campaign is a promo campaign. This 4:47:05 particular account runs promos every 4:47:07 three to four weeks. And so as a 4:47:08 function of that, we want it segmented 4:47:10 out. Why? Because the promos turn over a 4:47:12 lot. And so if you're launching promos 4:47:14 in the testing and the scaling campaign, 4:47:15 it will disrupt the learnings of the 4:47:18 campaign because you're constantly just 4:47:19 turning stuff off and launching new 4:47:20 stuff in it. You want to leave stuff 4:47:22 that's working and you want to add pipes 4:47:24 at this level of spend. So we want to 4:47:26 add this in separate to not impact the 4:47:28 performance over here. This will also 4:47:30 generally go after a different customer. 4:47:32 So it is a little bit different in 4:47:33 prioritizing price sensitive consumers 4:47:35 and so as a product of that we do want 4:47:37 it to optimize a little bit separately. 4:47:38 The fourth is an advertorial campaign. 4:47:41 So advertorials as an additional funnel 4:47:43 for this business started to do very 4:47:45 well in the testing. It started to do so 4:47:47 well to the point in which it was 4:47:48 consuming about 30% of total spend. And 4:47:50 so it actually made sense to just pull 4:47:52 it out and have it dedicated here. So it 4:47:54 can be KPIed on its own and it can be 4:47:56 looked at as a completely separate 4:47:58 funnel within the account. And then 4:48:00 lastly, a DPA campaign. This is 4:48:02 primarily for retargeting existing 4:48:03 customers as well as a little bit of 4:48:05 website visitors. This is running on 4:48:06 incremental attribution so that we're 4:48:08 attributing correctly based on its 4:48:10 actual incrementality. And this is kept 4:48:12 at a relatively low spend in line with 4:48:14 frequency. What doesn't change at this 4:48:16 spend level? What stays the same? Well, 4:48:18 number one is the two non-negotiables. 4:48:20 Everything that we introduce, every 4:48:22 extra bit of complexity needs to still 4:48:24 ensure that we have data integrity so 4:48:26 that we can read the data and then make 4:48:28 decisions. And it all needs to have 4:48:30 commercial alignment. Is this aligning 4:48:32 with the products, the margin, the 4:48:33 portfolio, the personas, or are we just 4:48:36 adding complexity for the sake of it 4:48:37 when we don't need it? Number two is the 4:48:39 breakdown effect still applies. Suddenly 4:48:41 just doing breakdowns or looking at the 4:48:43 ad level does not become more productive 4:48:45 here than it does at a lower spend. It 4:48:47 is the same thing. The principle still 4:48:48 stays. Number three, the concept 4:48:51 framework of launching creatives still 4:48:53 remains the same. You can have this 4:48:55 framework up at 300k a day and spend. In 4:48:58 fact, I would recommend it. Normally, 4:48:59 the big accounts that are spending those 4:49:01 levels are structuring creative and 4:49:02 testing it in this way. And number four, 4:49:04 you want to be KPIing at the ad set and 4:49:07 campaign level. This doesn't change. So 4:49:09 the premise really is that you're adding 4:49:12 structural layers into the account as 4:49:14 spend increases and business complexity 4:49:16 increases, but you're not giving away 4:49:18 the fundamentals that we spent the first 4:49:20 30 minutes of the video setting in 4:49:22 place. And then the question I've 4:49:23 alluded to throughout the whole video is 4:49:25 versus CBO. It's the wrong question. 4:49:27 It's honestly personal preference. And 4:49:28 the reason being is that this is just 4:49:30 what your risk portfolio is. Okay? If 4:49:32 you want more risk in the account, but 4:49:34 you want better efficiency, you go for 4:49:36 CBO. The reason being is it's going to 4:49:38 distribute more spend to the highest 4:49:40 performing ads and the highest 4:49:41 performing adsets, which is good, right? 4:49:43 You're going to get the highest ROI or 4:49:44 return on ad spend in the account. The 4:49:46 disadvantage is that all the spend is 4:49:47 just going to go to the highest ads and 4:49:49 so you can launch new tech, new ads, and 4:49:51 all this money on creative production, 4:49:52 put it into the account, and it gets no 4:49:54 spend. So, if you just went and spent 4:49:55 $20,000 on a new campaign shoot, and 4:49:57 then you went put all that content into 4:49:59 the account, it's not spending. What do 4:50:00 you do? or let's say that you went and 4:50:02 just spent $20,000 on a very expensive 4:50:05 influencer to run a partnership ad for a 4:50:08 60-day period, put it in the account, 4:50:09 isn't getting spent. Now, you can put 4:50:12 minimum spend caps in CBOS, of course. 4:50:14 So, you can go and put minimum spend 4:50:16 limits. You can also put maximum spend 4:50:18 limits as well, but then you're 4:50:19 effectively just running an where 4:50:22 either you're just increasing the 4:50:23 complexity of management because running 4:50:25 min and max spend caps is just annoying. 4:50:27 It's just a harder management tool 4:50:28 within the account. Or you do this with 4:50:30 like let's say 60% of budget and then 4:50:32 you let 40% of budget get distributed by 4:50:35 meta how it want. Now that's actually in 4:50:36 my opinion the best structure to run. 4:50:38 Most of the structures that we run these 4:50:40 days is that structure which is we're 4:50:42 running CBO but we're using minimum 4:50:45 spend limits and maximum spend limits 4:50:47 but we're allowing about 40% of total 4:50:49 budget to just flow wherever it wants 4:50:51 and 60% we're tightly controlling and 4:50:54 saying you have to go here, you have to 4:50:55 go. The reason why I'm not a proponent 4:50:57 of this at like lower spend level 4:50:59 accounts is that this is just like a lot 4:51:01 of management and it requires a lot of 4:51:03 oversight, which is fine for us when 4:51:05 we're managing very large accounts with 4:51:07 large ad spend with large revenue. It's 4:51:08 worth your investment for someone like 4:51:10 us to be able to do this kind of degree 4:51:11 of management and budget segmentation. 4:51:13 But if you're managing like 10k a month 4:51:15 in ad spend, this is it's just kind of 4:51:16 overkill. Like just run an or just 4:51:18 run a CBO. Choose your risk versus 4:51:20 efficiency profile and run it. So that's 4:51:21 really the thought through of the 4:51:23 process. So if you prefer ABOS and once 4:51:24 again I have seen ABOS running with 4:51:26 majority of ad spend on ad accounts 4:51:28 spending a4 million a day and I've also 4:51:30 seen CBOS running on ad spends quart 4:51:32 million a day. So it really comes more 4:51:34 so down to personal preference as to how 4:51:37 you want the testing methodology rolled 4:51:39 out in their account. Do you want spend 4:51:41 forced through every new creative or do 4:51:43 you not? Do you trust that the algorithm 4:51:45 is only going to give spend to an ad if 4:51:46 it's going to do well? That all comes 4:51:47 down to your trust and the algorithm 4:51:49 what you have seen historically 4:51:50 subjectively within the account. Has the 4:51:51 algorithm not given spend to an ad 4:51:53 before? And then you went, "Oh, I 4:51:54 thought this is a good ad. Let's put a 4:51:55 minimum spend cap." You put a minimum 4:51:57 spend cap, it becomes a winner. 4:51:58 Immediately, you start losing your trust 4:52:01 that the algorithm knows what it's doing 4:52:02 cuz it wouldn't have spent on that ad if 4:52:04 you didn't force it to spend and now 4:52:05 it's the top performing ad in the 4:52:06 account. So, it's going to come down to 4:52:07 honestly your own bias and subjectivity 4:52:09 in relation to how much you trust the 4:52:11 platform in whether you're going to go 4:52:13 or CBO. And I honestly don't think 4:52:14 there's a right or a wrong answer. You 4:52:16 can run whatever. They both perform 4:52:17 well. The only caveat I will add is what 4:52:19 I said previously, which is that if you 4:52:21 have really high AOV and low conversion 4:52:24 volume, I wouldn't recommend CBO. And 4:52:26 the reason being is that the CBO will 4:52:28 prioritize soft higher intent metrics 4:52:30 like clickthrough rate and CPCs. And so 4:52:31 you'll just end up with spend getting 4:52:33 distributed to the best soft metric ads 4:52:36 rather than to the best ads that are 4:52:37 actually going to drive conversions 4:52:38 within the business. There is also the 4:52:40 complexity of we will move out and in of 4:52:44 ABOS's and CBOS depending on seasonality 4:52:47 which sounds kind of weird but because 4:52:49 if you put these names aside and you 4:52:52 just think high risk high ROI and then 4:52:54 you think low risk slightly lower ROI 4:52:57 like the ROI isn't that much lower here 4:52:58 but it's a little bit lower let's say 4:52:59 10%. Well I want to pick this during 4:53:02 Black Friday and November December and 4:53:04 peak periods where we're just trying to 4:53:05 spend as much as possible in a short 4:53:07 window. I don't care what creatives get 4:53:09 spend. I just want to maximize 4:53:11 efficiency and maximize spend. But then 4:53:12 in like Jan Feb in Q1, I kind of want 4:53:15 this, right? I want to set Q1 up for a 4:53:18 bunch of creative testing. I want to 4:53:20 test as much as possible. This is a 4:53:21 great environment for testing because 4:53:23 you aren't artificially inflating 4:53:24 conversion rates and making ads seem 4:53:26 like they're doing well when they're 4:53:26 actually not. We have a relatively low 4:53:28 risk profile during January. We don't 4:53:30 want to increase our risk profile during 4:53:32 this part of the year. And so what we 4:53:33 will often do is actually rotate into a 4:53:36 CBO in Q4 for some accounts and then 4:53:38 roll it back to an in Q1. Now 4:53:40 regardless of whether you run ABOS or 4:53:42 CBOS, the ad set composition remains the 4:53:45 same, which is that you always want a 4:53:46 minimum of three ads in an ad set 4:53:49 because it allows the adset to sequence 4:53:51 across ads. If you only put one ad in an 4:53:53 ad set, it can't sequence across 4:53:55 anything. And so it's just going to 4:53:56 serve that one ad to users, which 4:53:57 ultimately isn't going to get people to 4:53:58 buy because generally people need to see 4:54:00 an ad 2 3 4 5 six times before they 4:54:02 actually purchase. You want a minimum of 4:54:04 three ads. You want a maximum of really 4:54:06 infinity depending on spend. And so you 4:54:09 can go all the way up to the maxed ad 4:54:10 limit at an adset level. That's fine. We 4:54:12 have accounts where we run that. We 4:54:13 might have 200 creative in an adset. 4:54:15 Generally for most people watching this, 4:54:17 you don't have anywhere near enough 4:54:18 creative volume to support 200 ads in an 4:54:20 adset level. Like you may as well have a 4:54:22 degree of segmentation there. and 4:54:23 there's likely going to be different 4:54:24 concepts in those 200 ads you don't want 4:54:26 to consolidate anywhere. So, as a rule 4:54:28 of thumb, you can think about three to 4:54:29 15 ads per adset. Number two is that you 4:54:32 want all of your ads resonating with the 4:54:35 same audience. Now, this doesn't mean 4:54:37 that you need different stages of 4:54:38 awareness or different types of ads. So, 4:54:41 this is a really common question I get, 4:54:42 which is that if I launch an adset with 4:54:44 five ads in it and it's all under one 4:54:46 persona talking to moms 40 to 50 who 4:54:49 need a rain jacket, do I put every stage 4:54:51 of awareness in there? Do I put my super 4:54:53 topfunnel VSSL style ads that are trying 4:54:55 to sell this weird raincoat through a 4:54:57 story line and a 5-minute video, but 4:55:00 then also just put a static ad that just 4:55:01 says here's a RCO 50% off. Do I put them 4:55:03 in the same ads there? Because they're 4:55:04 at very different stages of awareness. 4:55:06 So, do I split that at the adset level? 4:55:07 And the answer is you can do either. 4:55:09 It's just dependent on how you want to 4:55:11 read the data in the platform. And so, 4:55:13 let me walk you through an exact 4:55:15 example. So, option one is you split it 4:55:17 out. You have one adset which is the 4:55:19 persona, the raincoat, but this is very 4:55:22 top offunnel creative. It's high up on 4:55:25 the stages of awareness. And then you 4:55:26 have adset two which is going to be very 4:55:28 bottom of funnel in the stages of 4:55:30 awareness in the way that this creative 4:55:31 is designed. That's adset two. Option 4:55:33 two is we just consolidate it. This has 4:55:34 our top of funnel, it has our middle of 4:55:36 funnel, it has our bottom of funnel ads 4:55:38 all consolidated under the one adset. 4:55:40 Which option is better? My understanding 4:55:41 of meta's machine learning and the way 4:55:43 that conversion data is split across the 4:55:45 account generally pushes me a little bit 4:55:47 more towards option. The reason being is 4:55:48 that we want these ads to be able to 4:55:50 easily sequence against each other, 4:55:51 bring people through the journey and 4:55:53 convert. However, I have seen option one 4:55:55 also work. The reason why option one is 4:55:57 an inferior option. The reason why I 4:55:59 would recommend most people don't do 4:56:01 this and they go for option two isn't to 4:56:03 do with the actual machine learning and 4:56:04 the way that it works. Instead, it is to 4:56:06 do with your own human bias. This adset 4:56:08 down the bottom here will have a 4x row. 4:56:10 This adset up here will have a two. This 4:56:12 will have a six. Now, if you go into 4:56:14 this account on option two and you look 4:56:16 at this adset and it's at a 4x, your 4:56:18 conclusion is this concept is doing 4:56:20 well. Let's do more like this. Let's 4:56:21 also scale budget. Incredible. Now, 4:56:23 let's say instead you go into this ad 4:56:25 account and you go, okay, we have two 4:56:26 adset. This one's not doing well. This 4:56:28 one's doing well. What do we do? Do we 4:56:30 blend the numbers and say that overall 4:56:32 the concept at a 4x therefore do more 4:56:34 top offunnel creative? Is the top 4:56:36 ofunnel creative really performing well? 4:56:38 Do we turn this off and just like leave 4:56:41 bottom ofunnel creative for this? Like 4:56:42 what do we do off the back of it? Now 4:56:44 the answer in terms of what you do is 4:56:45 you should just blend the numbers. 4:56:47 Assume that these two adsets are working 4:56:48 together to get the conversion and 4:56:50 therefore do more scale it etc. Exactly 4:56:51 what you would do here. But due to your 4:56:53 own human bias in the way that you're 4:56:55 going to read the data, you're 4:56:56 inherently not going to want to do that. 4:56:57 You're going to look at this 2x and 4:56:59 particularly if you're working with 4:57:00 clients that aren't inclined to think of 4:57:02 it like this, they're going to go, "No, 4:57:04 no way. Don't put more spend into a 2x. 4:57:06 That's not a good idea. Put more spend 4:57:07 here. Even though this is obviously 4:57:08 bottom of funnel and then this traffic 4:57:10 is driving into here which is converting 4:57:11 and this is getting the last click 4:57:12 attribution. And so this just adds a lot 4:57:14 more complexity into the human 4:57:16 decision-making process. And the human 4:57:18 decision-making process is where this 4:57:20 falls apart. And so I would recommend 4:57:22 that you just put your middle, bottom, 4:57:23 top of funnel ads all together under the 4:57:25 one adset. Now this is an underrated 4:57:27 topic when it comes to adset structure, 4:57:29 which is actually naming adsets 4:57:31 correctly. What you'll see in most ad 4:57:32 accounts is something like this. March 7 4:57:36 creatives. So, as batches of creatives 4:57:38 are made, they're just named, thrown in 4:57:40 the adset level, and right. The issue 4:57:42 with this is that there's no ability to 4:57:44 easily filter into what concept we're 4:57:46 actually testing here. This also infers 4:57:48 that we're just batching a bunch of 4:57:49 concepts together rather than carefully 4:57:51 structuring them so that we can get 4:57:52 clear data insight and iterate into the 4:57:54 creative team. And so, as a product of 4:57:55 that, when we pick up accounts like 4:57:57 this, it's very annoying and complex. 4:57:59 And often we'll actually go back and do 4:58:01 retrospective renaming of old ads set so 4:58:03 that we can at least get some kind of 4:58:05 clear data structure historically so we 4:58:07 can infer what best to double down on 4:58:09 moving forward because if you don't have 4:58:10 clear naming conventions in place of for 4:58:12 example you'd want to rename this into 4:58:14 the exact persona angle offer and then 4:58:17 content type if you're going to group a 4:58:18 content type and then date of launch so 4:58:20 that we can start to bundle and KPI 4:58:22 different concepts across a campaign. So 4:58:24 what about when it comes to retargeting 4:58:26 DPAs and existing customers? Here's 4:58:29 three principles you need to keep top of 4:58:30 mind. Retargeting is a capped supporting 4:58:33 layer. It's absolutely not a growth 4:58:36 driver. This is unfortunately a mistake 4:58:38 that I see in so many ad account audits 4:58:40 that we do, which is that you go in, you 4:58:42 do an audience segment breakdown, you 4:58:44 pull up from the bottom of the account, 4:58:45 there's a little button there, and you 4:58:47 can just see total spend distribution 4:58:48 across the account. And you just see 4:58:50 like 50% of spend going to existing 4:58:52 customers and engaged audiences. Now, 4:58:53 like 50% of your spend is going towards 4:58:55 support spend. that's not actually going 4:58:57 to drive incremental new customer growth 4:58:58 within the business. It's just 4:58:59 supporting existing spend and it's not 4:59:01 even that incremental. And so the key 4:59:03 premise to keep top of mind here is that 4:59:05 you want to be at all times minimizing 4:59:08 retargeting spend as much as possible. 4:59:10 This is the different thought process 4:59:12 between cold and retargeting which is on 4:59:14 cold on new customer acquisition. The 4:59:16 constant question that we're asking 4:59:18 ourselves internally is where can we 4:59:20 spend more money? What platform? What 4:59:21 channel? What campaign? What ad? Where 4:59:23 can we spend more? We need to spend more 4:59:25 to get more incremental lift as long as 4:59:27 it's profitable. As long as we can make 4:59:28 a profitable exchange, we can put $1 4:59:30 into this ad and get four out. Or we can 4:59:32 put $1 into Tik Tok and we'll get four 4:59:33 out. We want to keep putting more money 4:59:35 there to grow new customers and grow the 4:59:37 business. On existing customers and 4:59:39 retargeting, it is the opposite frame. 4:59:41 We want to be thinking, how can we spend 4:59:42 the least amount of money? Where can I 4:59:44 take money out of? Are we spending too 4:59:45 much retargeting on TikTok? Take it out. 4:59:47 Take it out of Pinterest. Take it out of 4:59:48 Meta. Because often the top end of spend 4:59:50 on retargeting is actually not 4:59:52 incremental. We're almost always 4:59:53 overspending and so we want to be in 4:59:55 almost a scarcity mindset on retargeting 4:59:58 and then an abundance mindset on cold 5:00:00 targeting and new customer growth. 5:00:01 Number two on retargeting is 5:00:03 consolidation of audiences. Uh what is 5:00:06 very like 2019 and you should not be 5:00:08 doing is having a bunch of different 5:00:09 adsets segmented by like retargeting 5:00:12 different product categories based on 5:00:13 what landing page they landed on or 5:00:15 retargeting add to cart separate from 5:00:16 initiate checkout separate from website 5:00:18 visitors. that hyper segmentation in 5:00:20 retargeting. All it does is increase 5:00:23 CPMs. It just makes it more expensive 5:00:25 for you to serve retargeting ads and it 5:00:26 has almost no direct upside. If someone 5:00:28 can give me a really strong reason as to 5:00:30 why their ad account has six different 5:00:31 adsets targeting six different degrees 5:00:33 of separation of warmth in the funnel, 5:00:36 like I don't know what you're doing. The 5:00:37 CPM increase is never outweighed by the 5:00:40 conversion rate increase. And then 5:00:41 number three is really your core KPI 5:00:43 other than obviously incremental revenue 5:00:45 on retargeting campaigns should be 5:00:47 frequency. You should just be looking at 5:00:49 frequency very tightly to understand how 5:00:51 many times are we retargeting existing 5:00:53 customers, engaged audiences every 5:00:54 single month, every single week and is 5:00:56 this intuitively or subjectively too 5:00:58 high or too low? Should we be 5:00:59 retargeting existing customers 30 times 5:01:01 a month, but most people probably not. 5:01:02 And so intuitively, you know, you're 5:01:03 probably overspent. So you can probably 5:01:04 pull that down. Now, a big part of 5:01:06 retargeting is DPAs, which is dynamic 5:01:09 product ads. Uh, DPAs are a pretty 5:01:11 critical component of most ad accounts, 5:01:14 particularly if we're talking about 5:01:15 fashion, particularly if we're talking 5:01:16 about large cataloges or SKs within a 5:01:18 product portfolio. DPAs are great 5:01:20 because they're going to dynamically 5:01:22 retarget a user with a catalog ad that 5:01:24 is going to rep prioritize products 5:01:26 based on the products that they looked 5:01:28 at on the website if everything's 5:01:29 working correctly and if the pixel 5:01:30 worked and actually fired the correct 5:01:32 data back. Amazing. Now, the bad thing 5:01:34 about DPAs is that they will pretty much 5:01:36 only place right before the moment of 5:01:39 purchase as like a bottom offunnel most 5:01:41 aware retargeting ad. And so, because of 5:01:43 that, they get way more credit than they 5:01:45 should actually get. The rorowaz on 5:01:47 these things looks way better than it 5:01:49 actually is. And so, people end up 5:01:51 naively overspending on them 5:01:54 substantially. And so, there's a few 5:01:55 rules when it comes to DPAs. Number one 5:01:57 is make sure that when you're assessing 5:01:59 return on ad spend, you're assessing 5:02:01 based on 7-day click or based on 5:02:04 incremental attribution. If people don't 5:02:06 know what I'm talking about with these 5:02:07 attribution settings, just watch our 5:02:09 video that was posted on the 18th of May 5:02:11 called if your ad account looks good but 5:02:13 profit isn't, watch this video. It's an 5:02:15 hour and 2 minutes long. It goes all 5:02:17 into attribution models and data 5:02:18 integrity so you can better understand 5:02:20 that. Number two is that your DPA 5:02:22 performance is really just a direct 5:02:24 reflection of your topfunnel investment. 5:02:26 If you increase budget on top of funnel, 5:02:28 if you go harder on top of funnel, DPA 5:02:29 rorowes goes up because you just capture 5:02:31 more revenue at the bottom. If you 5:02:32 decrease top of funnel, your DPA 5:02:34 performance will go down. We actually 5:02:36 made an entire video on this like 6 5:02:37 months ago, which was called the DPA 5:02:39 death spiral of fashion brand. The whole 5:02:41 idea was that I had seen six fashion 5:02:42 brands in a row on six audits I did in 5:02:44 the course of like 2 weeks where all of 5:02:46 them were spending 80% of their spend on 5:02:48 DPAs and the account was just declining 5:02:50 over the past 6 months. And the reason 5:02:52 being is that they saw the performance 5:02:53 on DPAs. the agency just kept bumping 5:02:55 budgets because it was good and then the 5:02:58 brand as a function of the agency's 5:03:00 advice was like oh DPAs are doing so 5:03:02 well do we need to make as much creative 5:03:03 and they were like no you don't because 5:03:05 DPAs are holding up performance so they 5:03:06 just started decreasing creative 5:03:08 production and creative and topfunnel 5:03:10 spend allocation within the ad account 5:03:12 and then the whole business just falls 5:03:13 apart because if DPA sit at the bottom 5:03:15 of the funnel and then you just stop 5:03:16 spending on top obviously that is not 5:03:18 sustainable then when we think about 5:03:19 existing customer spend allocation what 5:03:22 happened which was really bad for the 5:03:24 entire performance marketing and ecom 5:03:26 space in my opinion was that Meta 5:03:28 launched advantage plus campaigns like 2 5:03:31 years ago and the first setting that 5:03:33 there was on advantage plus campaigns 5:03:34 was this little box that you could tick 5:03:36 and it was called existing customer 5:03:38 spend cap and it was a percentage input. 5:03:41 So you would input your percentage that 5:03:42 you wanted going towards existing 5:03:44 customer. This framed everyone into the 5:03:47 mindset of not only thinking about 5:03:48 existing customer spend, which was good, 5:03:50 but it got them framed towards, yeah, 5:03:52 what percent should we put towards 5:03:54 existing customers? What should our 5:03:55 percentage be in terms of meta spend 5:03:57 allocation? It's a terrible way to look 5:03:58 at it. It's the wrong frame entirely 5:04:01 because percentage is irrelevant. Who 5:04:03 cares what the percent is towards 5:04:05 existing customer? I only really care 5:04:06 about two things. Number one, is it 5:04:08 profitable? If we spend another $10 a 5:04:10 day or $1,000 a day on existing 5:04:12 customers, do we make enough return 5:04:14 revenue to be able to pay for that spend 5:04:15 and make profit? Number two, which is 5:04:17 kind of a softer metric that gives us 5:04:19 more realtime visibility, is what is our 5:04:22 frequency? Yes, sure, we can have a 20% 5:04:24 existing customer spend, but if that 5:04:25 means that our frequency on existing 5:04:27 customers is a 50 every 30 days, what 5:04:29 are we doing? Why are we targeting 5:04:31 people 50 times a month? Vice versa, if 5:04:33 our spend is 20%, but we only have a 5:04:35 frequency on existing customers of one, 5:04:37 we're probably substantially 5:04:38 underspending. And so your existing 5:04:40 customer spend as a percentage is a bad 5:04:42 metric because percentages change based 5:04:44 on how big the existing customer pool is 5:04:47 and how much budget allocation you 5:04:48 currently have to new customer 5:04:50 acquisition. Those two variables will 5:04:52 just substantially change the spend 5:04:53 allocation as a percentage to where it's 5:04:55 useless. So how do you think about it 5:04:57 instead? You just think about it as a 5:04:58 dollar value. So, how much dollar spend 5:05:00 should we have per month to existing 5:05:02 customers? And that you can calculate by 5:05:04 taking your amount of customers, you 5:05:06 times by your CPMs, and then you times 5:05:09 by the amount of frequency that you want 5:05:11 per month on those customers. And this 5:05:13 tells you exactly what your monthly 5:05:15 budget should be towards an existing 5:05:17 customer audience. So, here's some 5:05:18 settings that you should be turning off 5:05:19 and not using. And number one is 5:05:21 flexible ads. I explained briefly why 5:05:25 before, which is the fact that you can 5:05:26 load a bunch of creative into one single 5:05:28 ad, but the issue is that that ad 5:05:29 performs well or if it doesn't perform 5:05:31 well, we don't actually know what 5:05:32 creative was driving performance or not. 5:05:34 And so the issue here becomes that it 5:05:36 doesn't follow the number one foundation 5:05:38 that we had for account structure, which 5:05:40 is data integrity, which allows for 5:05:42 feedback loops. So we can read the data 5:05:44 and then we can make decision. On 5:05:45 flexible ads, you can't read the data. 5:05:47 So it's fundamentally a bad ad type. 5:05:48 Now, the only reason why you would do 5:05:50 this is to try to get around the page 5:05:52 limit within the account. There's much 5:05:54 better ways to get around the page 5:05:55 limit. I already gave them to you. If 5:05:56 you still can't do that, if you're going 5:05:58 to do flexible ads, you need to put a 5:05:59 bunch of creative that's very similar. 5:06:01 And so that if it does perform well, we 5:06:03 can kind of say that this type of ad 5:06:04 does well because it's all similar. If 5:06:06 you put any kind of large variation of 5:06:09 asset types under a flexible ad, you 5:06:10 then no longer know what's actually 5:06:12 working and you're just introducing 5:06:13 stuff into the account that doesn't 5:06:15 facilitate a decision-making feedback 5:06:17 loop. Why we as an agency really don't 5:06:18 like flexible ads on top of that is that 5:06:20 you don't have control over cropping the 5:06:23 images and so it can serve the one by 5:06:25 ones and story placements. It'll 5:06:27 randomly crop the story placements into 5:06:29 feed placements. It's generally terrible 5:06:31 for the actual quality of how the ad 5:06:33 presents on the feed. What Meta can also 5:06:35 do sometimes is if you have a bunch of 5:06:37 flexible ads, it will just string them 5:06:39 together and create a carousel out of 5:06:40 them, which also isn't ideal when a 5:06:42 client hasn't approved a carousel. 5:06:44 That's a string of random images that 5:06:46 have been put in the account. So, 5:06:47 generally not a fan. Next is cost caps 5:06:50 or bid caps. I just really wouldn't be 5:06:53 concerned with worrying about bidding 5:06:54 strategies if you're spending under 5:06:56 $200,000 a month in an ad account. It's 5:06:58 just not the right bottleneck to be 5:07:01 solving for within the account. The 5:07:02 question that you should ask yourself 5:07:03 is, will the business double if we just 5:07:05 really focus on bidding strategies 5:07:07 within Meta? And the answer normally is 5:07:09 no, because it's a low leverage 5:07:10 opportunity. Now, is that to say that 5:07:12 you shouldn't run cost cap? Not at all. 5:07:13 Like if you know what you're doing, you 5:07:14 can run cost caps on any spend level. 5:07:16 More so, I'm saying focusing on changing 5:07:19 and testing bidding strategies at lower 5:07:21 spend levels generally isn't the biggest 5:07:23 lever to be able to unlock growth. 5:07:25 Number three is turn off one day view 5:07:28 attribution or at least don't use it 5:07:30 within the decision-m of how you're 5:07:32 flowing budgets at an adset level and 5:07:34 what's working and what's not because it 5:07:35 will substantially overattribute based 5:07:37 on existing customer revenue. Number 5:07:38 four is you want to check for all of the 5:07:40 advantage plus creative optimizations. 5:07:43 Um, particularly you want to go into 5:07:45 your advertiser settings and go into 5:07:48 creative sub dropown and then there'll 5:07:50 be a bunch of automated rules that are 5:07:53 turned on which will effectively flip 5:07:55 your creative optimization settings back 5:07:57 on over time if you don't turn it off at 5:07:59 the ad account advertiser setting level. 5:08:01 So, not only do you need to do this on 5:08:02 every single ad that you launch, which 5:08:04 is incredibly annoying, but you also 5:08:05 need to do it at the advertiser setting 5:08:07 level. And then number five is you 5:08:08 really want to make sure that your 5:08:10 audience segments are set up correctly. 5:08:12 What has happened a lot and that we keep 5:08:14 an eye on is that the sync of the 5:08:17 audience segment from Clavio to Meta 5:08:20 will break periodically. And when that 5:08:22 happens, your audience segment stops 5:08:24 flowing through existing customers 5:08:25 correctly and so you end up spending on 5:08:27 existing customers thinking they're new 5:08:28 and obviously that starts to erode data 5:08:30 integrity within the account and you 5:08:31 start spending in places that you don't 5:08:33 want to. When it comes to testing 5:08:34 discipline in the account, when you're 5:08:36 setting up any kind of test, you want to 5:08:38 answer two questions that are very 5:08:39 similar to what we talked about before, 5:08:40 which is what's the stat threshold? So, 5:08:43 how many conversions do we need to say 5:08:46 that it's statistically significant? Is 5:08:47 it five? Is it 10? What is that 5:08:50 threshold? And then number two is how 5:08:51 quickly do we want to actually learn? 5:08:53 So, do we want to spend to that 5:08:56 statistical significance in a 3-day 5:08:58 period, in a 7-day period, in a 30-day 5:09:00 period? How are we spending to the 5:09:02 threshold to then be able to move out of 5:09:04 the test or say that the test was a su 5:09:06 success and scale it? The way that you 5:09:07 actually calculate it using these two 5:09:09 questions is that your daily testing 5:09:11 budget on a test is your target 5:09:13 conversions times by your expected CPA. 5:09:16 So what is your expected cost per 5:09:18 acquisition divided by the test duration 5:09:20 in days. For example, if our target 5:09:23 conversions is 20, that's the threshold 5:09:25 in which we'll call the test 5:09:26 statistically significant. Our expected 5:09:28 cost per acquisition is $100. We want 5:09:31 the test to go for 14 days. Then the 5:09:33 daily budget on this test is $143 per 5:09:37 day. Now, this is also why the common 5:09:40 feedback from Meta is wrong, which is 5:09:42 that Meta says that you need 50 5:09:44 conversions every 7 days to be able to 5:09:47 exit the learning phase and achieve 5:09:49 statistical significance. The issue is 5:09:50 is that when you try to back propagate 5:09:52 this into daily test budgets, it just 5:09:54 blows out and becomes ridiculous for 5:09:56 like 99% of avatars. So if you take this 5:09:58 exact test case as an example, if we 5:10:00 need to target 50 conversions and CPA is 5:10:03 obviously the same and we need to do it 5:10:05 in 7 days, that means daily budget here 5:10:08 needs to be $714 5:10:10 per day. And that's just for one concept 5:10:13 adset test, which is just ridiculous cuz 5:10:16 if we want to put like five tests in or 5:10:18 10 tests in, suddenly you're spending 5:10:20 $7,000 a day just on the testing 5:10:22 campaign. If I was to compress this 5:10:24 whole training into just four questions 5:10:26 that you should write down and walk away 5:10:28 from when you're looking to build an 5:10:30 account structure is number one, does 5:10:32 the structure produce data that I can 5:10:34 read to make decisions? Because 5:10:36 ultimately the whole point in the 5:10:37 account structure is that we're 5:10:38 producing readable data so that we can 5:10:40 continue to iterate and make decisions 5:10:42 within the business. Number two is does 5:10:44 the structure match how the business 5:10:46 actually makes? So is the structure 5:10:48 commercially aligned to the objectives 5:10:50 of the business and how the business is 5:10:52 producing profit. Number three is the 5:10:54 structure appropriate for the spend tier 5:10:56 and creative output. Creative throughput 5:10:58 as well as the amount that you're 5:10:59 spending is going to substantially 5:11:01 change the structure that can be allowed 5:11:03 within the account. And then lastly, are 5:11:05 the support layers, so DPA, retargeting, 5:11:08 website visitors, existing customers, is 5:11:10 that aligned with the actual retention 5:11:12 in the business? Do we need this amount 5:11:14 of spend? How much spend do we need? How 5:11:16 are we structuring DPAs and retargeting? 5:11:18 How much spend distribution is going 5:11:19 here? This is all going to be once again 5:11:21 context dependent on the actual business 5:11:23 and the retention within the business. 5:11:25 If the answer to any of these four 5:11:28 questions is no, then the account 5:11:30 structure is wrong and you need to 5:11:31 restructure accordingly. So, that's the 5:11:34 end of the video. Two things. Number 5:11:35 one, if you're a performance marketer, 5:11:36 we're always hiring. Please reach out to 5:11:38 us at hiringblensedigital.com.au. 5:11:41 If you're an e-commerce brand or retail 5:11:42 business doing over $5 million a year in 5:11:45 online revenue, click the link below. 5:11:46 You can get a free audit from us where 5:11:48 we'll break down how all of this 5:11:49 strategy actually translates into your 5:11:51 account. We'll give you the nuances of 5:11:52 account structure. And if you're in 5:11:54 neither of those two buckets, subscribe. 5:11:56 In 2026 and 2027, creative is the lever 5:11:59 that sits within meta ads. Media buying 5:12:01 fundamentally is not your advantage 5:12:03 right now. You need to become incredibly 5:12:05 good at creative strategy, at creative 5:12:07 design, production, then moving that 5:12:09 into the ad account and analyzing the 5:12:11 creative to create this flywheel effect. 5:12:13 By the end of this video, you'll know 5:12:14 everything you need to know about 5:12:16 creative for 2026. We're going to be 5:12:18 running through why creative is the 5:12:20 growth lever, how Meta's algorithm 5:12:22 actually works in 2026, concept 5:12:24 architecture, hook strategies that 5:12:26 actually scale, all the formats you can 5:12:28 use and test, a testing framework to 5:12:31 take that creative and actually put it 5:12:32 into the account to get statistically 5:12:34 relevant results. We're then going to 5:12:36 move into creative volume, creative 5:12:37 diversity, and a financial model that 5:12:39 ties it all together. We're going to 5:12:41 talk about fatigue, scaling winners, 5:12:42 portfolio management, creative 5:12:44 production. So, what did the teams, the 5:12:46 briefing, and the process look like? And 5:12:48 then lastly, we'll tie all of it 5:12:50 together and package it into the 2026 5:12:53 playbook. So, with that being said, 5:12:54 diving into section number one, we have 5:12:56 why creative is the growth lever. I've 5:12:58 been media buying on Facebook since 5:13:00 2018. So, I've been about eight coming 5:13:02 up on 9 years in the Meta Ads Manager. 5:13:05 And I remember back in 2018, 2019, 2020, 5:13:08 there was so much leverage in media 5:13:10 buying. You actually didn't need to be 5:13:11 very good at creative. And as long as 5:13:13 you had a massive edge on media buying, 5:13:15 you could beat a lot of people. And the 5:13:17 reason for that is that there was a 5:13:18 massive skill gap. In fact, I would 5:13:20 argue that any arbitrage opportunity 5:13:22 that exists within a market is a product 5:13:25 of the skill gap that exists at that 5:13:27 point in time. So back then, the 5:13:29 difference between someone that just 5:13:30 goes and opens up the ad manager and 5:13:33 someone that is a super advanced media 5:13:35 buyer that spends 12 hours a day doing 5:13:36 this and they're three years in, there 5:13:38 is such an enormous difference that that 5:13:41 difference allows for arbitrage and 5:13:43 allowed for the opportunity. Now over 5:13:45 time, Meta has slowly compressed media 5:13:48 buying. The targeting is now fully 5:13:50 automated. I would recommend 99% of 5:13:52 people are just running broadly, so 5:13:54 there's no skill within interest 5:13:56 targeting. Andromeda now optimizes for 5:13:58 the distribution of the creative. 5:13:59 Consolidation is always favored across 5:14:01 all of these digital platforms now 5:14:03 because conversion data consolidates 5:14:05 which allows for better machine learning 5:14:06 modeling. And all of the media buying 5:14:08 hacks that used to exist like 5:14:09 duplicating adsets or using bidding 5:14:12 models in a certain way. None of them 5:14:14 really work that well anymore. None of 5:14:16 them really give you a big arbitrage. 5:14:18 And so because of that this arbitrage 5:14:20 opportunity has compressed and instead 5:14:22 where the big arbitrage opportunity 5:14:24 exists right now is in creative. And it 5:14:27 is because most people fortunately for 5:14:30 you are terrible at creative. And so 5:14:32 because of that there is this huge 5:14:34 opportunity that if you simply become 5:14:36 good at creative strategy at 5:14:38 understanding what a good ad looks like 5:14:40 and then being able to build a system 5:14:42 that can churn out a ton of highquality 5:14:44 ads. This is where you now make all the 5:14:46 money on meta. And by the end of this 5:14:48 video, you should unlock this 5:14:51 opportunity. So, the first thing I want 5:14:52 you thinking through here is the 5:14:54 portfolio model. So, let's say that you 5:14:56 have $1,000 a day in ad spend and there 5:14:59 is two different scenarios here. 5:15:01 Scenario one is you have one single 5:15:04 creative that's holding the entire 5:15:06 $1,000 per day and you're at a 4x row 5:15:09 and then you have another 5:15:13 pathway that you could choose. You've 5:15:14 got three ads holding $1,000 a day 5:15:17 together. And this is also at a 4x 5:15:19 rorowaz blended across these three 5:15:21 creators. Now, I would rather be in this 5:15:24 situation every single day of the week. 5:15:26 And the reason being is that this 5:15:28 situation decreases risk enormously 5:15:31 within the business because any of these 5:15:34 three creative can fatigue and the other 5:15:36 ones will pick up the slack and you can 5:15:37 continue operating at this efficiency, 5:15:39 at this ad spend. But over here, if this 5:15:42 creative fatigues, you're done. And I 5:15:45 say this out of PTSD, working with 5:15:47 probably 8 to 10 clients now over the 5:15:50 course of the last 6 years who have come 5:15:52 to us 40% down yearon-year in new 5:15:55 customer acquisition, their business is 5:15:56 in a terrible spot, and they're going, 5:15:59 "Our business is declining. It's because 5:16:01 we built this eight figureure business 5:16:03 off three creatives that held a million 5:16:05 dollars in ad spend and now these three 5:16:07 creatives are all fatiguing and we're 5:16:09 trying to spin up variance and it's kind 5:16:10 of keeping us at some reasonable level 5:16:13 but we just can't get back to the level 5:16:14 and we keep decline. And my response 5:16:16 these days is it's too late. We're not 5:16:19 taking on that account because you 5:16:21 should have fixed that 6 to 8 months 5:16:23 ago. What people will do is they will 5:16:25 get one or two winning ads that can hold 5:16:27 a tremendous amount of ad spend and then 5:16:29 what they do is they just stop making 5:16:30 more ads and they go like why would we 5:16:32 spend money on creative production? 5:16:34 That's useless. Let's just double down 5:16:35 and put all our ad spend here and then 5:16:37 focus on other areas of the business. 5:16:38 Yeah, amazing. Until these creatives 5:16:40 fatigue and the only reality of 5:16:43 creatives in meta ads is that every 5:16:45 single creative will fatigue. Every 5:16:47 single creative has a maximum spend 5:16:49 threshold that it will reach. And so you 5:16:51 want to be constantly thinking through 5:16:53 creative production, ad account 5:16:54 management, and creative through the 5:16:56 lens of this portfolio model, which is 5:16:58 that we want to get as many ads as 5:17:00 possible holding as much spend as 5:17:02 possible. We do not want to overlever 5:17:04 into one or two ads because this is 5:17:06 ultimately what ends up destroying a lot 5:17:07 of businesses. Let me give you three 5:17:09 common mistakes that most teams are 5:17:11 making when it comes to creative. Number 5:17:12 one is a lot of people confuse activity 5:17:14 for strategy. I'll give you an example 5:17:16 of this, which is recently in the last 5:17:18 few months, I've seen two ad accounts 5:17:20 that are launching 2,000 plus new ads a 5:17:23 month, and they're barely able to crack 5:17:25 a4 million in ad spend per month. Now, 5:17:27 for those that work on large ad 5:17:28 accounts, you would know that that is 5:17:30 insane. You do not need 2 to 3,000 new 5:17:34 ads a month on a4 million budget. That's 5:17:36 so much volume. And they weren't 5:17:38 profitable. And so you're looking at the 5:17:40 fact that they're launching so much 5:17:41 volume, they're not profitable, they're 5:17:43 doing so much activity, but there's 5:17:46 clearly a gap in strategy because if you 5:17:49 were putting 2,000 great creative into 5:17:51 the account with a good offer, with a 5:17:53 good persona, with a good angle, then 5:17:55 you shouldn't need anywhere near 2,000 5:17:57 ads a month. And so, yes, I think 5:17:58 volume's I think you should be launching 5:18:00 as many ads as humanly possible, but 5:18:03 with a standard of quality and with a 5:18:06 strategy behind the creative that is 5:18:08 actually going into the account. So, do 5:18:09 not mistake activity for strategy. This 5:18:12 is really common right now because 5:18:13 people think they can just launch more 5:18:15 ads and their brand will scale. In fact, 5:18:16 I hear this all the time, which is 5:18:18 people are like, "Okay, I need to double 5:18:19 my business. Do I just launch 40 more 5:18:20 ads a month?" I was like, "If it's that 5:18:22 simple, everyone would do it." There's a 5:18:24 little bit more to just increasing 5:18:26 activity. The second one is 5:18:28 misdiagnosing bottlenecks. So with 5:18:31 increative, there is a ton of different 5:18:33 bottlenecks that exist that could be 5:18:34 holding you back. It could be the 5:18:36 quality of the creative. It could be the 5:18:38 format diversification. It could be the 5:18:40 quantity that's sitting at the different 5:18:42 stages of awareness or the stages of the 5:18:44 funnel. So you have nowhere near enough 5:18:46 top offunnel assets to produce the scale 5:18:48 that you actually want. This might be a 5:18:50 bottleneck on not actually introducing 5:18:52 partnership ads or employing different 5:18:54 strategies into the ad account to unlock 5:18:56 more spend. There's so many different 5:18:57 bottlenecks that exist just within 5:18:59 creative that holds a business back and 5:19:01 most people don't understand all the 5:19:03 bottlenecks that exist. So they 5:19:04 misdiagnose it. They don't realize what 5:19:06 the actual bottleneck is. And so then 5:19:07 they focus on activity that doesn't 5:19:09 actually move the needle. For example, 5:19:11 those two brands that I spoke about 5:19:12 before, they were not only mistaking 5:19:14 activity for strategy, but they were 5:19:16 also misdiagnosing the bottleneck in the 5:19:18 business, which is that they thought 5:19:20 just increasing sheer volume would 5:19:22 suddenly unlock growth, but it's not 5:19:23 doing anything for them. And it's 5:19:24 because they're trying to fix something 5:19:26 that actually doesn't need to be fixed. 5:19:27 And then the last one, which I hope 5:19:29 people are getting over by now, but I 5:19:30 still see it from time to time, which is 5:19:34 that creative is seen as a cost center 5:19:38 rather than a revenue driver. And this 5:19:41 is fundamentally because of the shift 5:19:43 away from media buying to creative, 5:19:45 which is that back in the day when media 5:19:47 buying had a lot of arbitrage behind it, 5:19:48 you didn't need a lot of creative and a 5:19:50 lot of volume of creative. And so 5:19:51 because of that, business owners would 5:19:53 often see the ad spend in the platform, 5:19:55 the spend on Google, the spend on Meta, 5:19:57 the spend on Tik Tok, Pinterest, 5:19:58 Microsoft as revenue driving, but the 5:20:01 actual creative that's used to fuel that 5:20:04 spend as a cost center. We want to 5:20:06 minimize that cost as much as possible 5:20:08 because we just want the platforms to 5:20:09 find us customers. But that's the 5:20:11 complete wrong way around. The creative 5:20:14 is actually what is driving the revenue 5:20:16 and Meta and these other platforms are 5:20:17 just driving the distribution. Now, the 5:20:19 fact that you can actually just post 5:20:20 this stuff on organic and get free 5:20:22 distribution as well means that creative 5:20:24 is also the revenue driver organically. 5:20:26 And so, if you're thinking that creative 5:20:27 is a cost center, you've got the 5:20:29 completely wrong approach. We're going 5:20:30 to be diving into later in this video 5:20:32 exactly how much spend you should have 5:20:35 on creative production respective to 5:20:37 your ad spend level. So, you can know if 5:20:39 you're currently under or overinvesting 5:20:41 in creative. Before diving into creative 5:20:43 strategy, you first need to understand 5:20:45 how Meta's algorithm works because this 5:20:47 is the distribution tool for the assets 5:20:49 that you're putting in it. Now, 5:20:50 everyone's probably heard the word 5:20:51 Andromeda. It's a massive buzz word. 5:20:53 I've talked about it in a lot of 5:20:54 content. And to put it simply, it's just 5:20:56 an algorithm change that Meta made 5:20:58 because they had a lot of GPUs for 5:21:00 training AI for their metaverse that 5:21:02 kind of fell flat. And so, they 5:21:03 repurposed all of those training chips 5:21:05 towards a better ad serving platform. 5:21:08 with that the mechanical shift that you 5:21:11 need to be across is that before 5:21:13 andrometer and to be honest we weren't 5:21:16 even doing this pre-andrometer but it's 5:21:17 an easy way to simplify the concept and 5:21:19 then I'll go a little bit more nuanced 5:21:20 which is that pre-andrometer you would 5:21:22 choose the audience so you would say hey 5:21:24 here is our interest 5:21:26 please target this interest and you 5:21:28 would then go and load all the creatives 5:21:30 in and so you put creative one creative 5:21:33 two maybe you're putting three creatives 5:21:34 under this ad set and you're saying 5:21:36 here's the interest here's the ads go 5:21:38 and match them. Now, post Andromeda, 5:21:40 what happened is you actually do the 5:21:42 opposite. So, you put the ads into the 5:21:45 account and you don't select any 5:21:47 targeting and then Meta goes and looks 5:21:50 at the creatives and it goes, "Okay, I 5:21:52 think that these are going to resonate 5:21:53 with these people based on the 5:21:54 transcript, based on the words in the 5:21:56 ads, based on the copy, based on my 5:21:58 understanding of historical conversion 5:22:00 data." And then it will go and find the 5:22:03 people to target. Now, this is 5:22:05 fundamentally a way better system 5:22:07 because interest targeting has a lot of 5:22:09 downside. Okay? When you're doing 5:22:11 interest targeting, you're effectively 5:22:12 labeling people with an interest, let's 5:22:14 say pets, and then just arbitrarily 5:22:16 serving all your creative to people that 5:22:18 are interested in pets. But the reality 5:22:19 is is that it's not just people 5:22:21 interested in pets that want your 5:22:23 product. It's probably people that are 5:22:24 interested in pets within a specific age 5:22:26 range that also have other variables or 5:22:29 psychographic data points that are 5:22:31 similar between them. And so there's so 5:22:33 much additional nuance that needs to be 5:22:34 captured and this serving mechanism ends 5:22:37 up capturing that nuance better than the 5:22:39 old one. Now the reality is is that meta 5:22:42 was kind of already doing this pre 5:22:43 Andromeda. We haven't used interest 5:22:45 targeting in most accounts for like 2 5:22:47 years. But what Andromeda really did is 5:22:49 popularize this new approach to 5:22:51 structuring an ad account. And so there 5:22:53 were still so many agencies and so many 5:22:55 people running interest targeting it was 5:22:57 crazy. Even though we have a video that 5:22:59 was put out two years ago about this 5:23:01 exact topic on the meta algorithm and 5:23:03 this was before Andromeda. So I was 5:23:04 explaining this concept before this 5:23:06 whole update even rolled out and why we 5:23:08 don't use interest targeting. It's 5:23:10 called how machine learning works in 5:23:12 2025 on meta ads. But this has 5:23:15 popularized this approach and now this 5:23:17 is really how you want to be thinking 5:23:18 through the asset. Now why this becomes 5:23:20 important is because the creative that 5:23:21 you loads in that you load in is the 5:23:23 targeting. So, if you load a bunch of 5:23:25 creative that's all the same, you end up 5:23:27 just targeting the same person. You 5:23:28 don't reach new people. If you end up 5:23:30 loading out a bunch of creative that 5:23:32 isn't clearly resonating with a specific 5:23:34 target demographic, then it's going to 5:23:36 get confused. For example, if you're 5:23:37 calling out everyone in your ads, if 5:23:39 you're like, "Hey, everyone, look at 5:23:41 this product." Meta is going to be very 5:23:42 confused cuz it's going to go, "What? 5:23:43 Should we target the entire population? 5:23:45 Is there a subset? Who should we 5:23:46 actually go after?" You also need to 5:23:48 think through this targeting mechanism 5:23:50 at an adset level in terms of 5:23:52 structuring because if you just put a 5:23:53 bunch of ads together that are all 5:23:54 targeting different people under the one 5:23:56 adset, the adset will also get confused. 5:23:58 Who are we targeting? So there's a lot 5:24:00 of additional nuance that we'll break 5:24:01 down here when we get into the structure 5:24:03 and testing section of this video. But 5:24:04 this is fundamentally the key concept 5:24:06 that you need to understand here. 5:24:07 There's also the creative similarity 5:24:09 score that rolled out in meta recently. 5:24:11 Now, most people still can't access this 5:24:12 in their ad accounts, but if you reach 5:24:13 out to your dedicated meta rep, they can 5:24:16 get you this score. So, you can have an 5:24:18 idea of what your similarity score is in 5:24:20 the business. And as it currently works, 5:24:22 the higher the number, the worse it is 5:24:25 because the more similar your creatives. 5:24:27 Now, why this matters so much is because 5:24:29 of Andromeda bundling. And so, if you 5:24:32 have, as I said before, multiple 5:24:34 different creatives here, let's say 1 2 5:24:37 3, but they're all slight variants. 5:24:39 Let's say they're the same campaign 5:24:41 shoot, but the model's just standing in 5:24:42 a slightly different position. Or let's 5:24:44 say it's UGC, but maybe there's a slight 5:24:47 variation in the script halfway through 5:24:49 the video. Or let's say these are 5:24:50 actually very different videos in terms 5:24:52 of formatting, but they're all speaking 5:24:54 to the exact same audience. Well, what 5:24:56 ends up happening is Andromeda will 5:24:58 bundle them together and serve them all 5:25:01 to the same pool of audience. Now, 5:25:03 because of this, because they're similar 5:25:04 creatives going to the same audience, 5:25:06 what happens in the ad account is 5:25:08 frequency goes up. So frequency is how 5:25:10 many times users are seeing your ads. 5:25:12 Obviously, these ads are just going to 5:25:13 serve to the same people over and over 5:25:15 again. So frequency goes up. Your reach 5:25:17 will go down. So you'll stop reaching 5:25:19 net new users. And as a function of 5:25:21 this, over time, your return on ad spend 5:25:24 will go down. And so you'll start seeing 5:25:26 diminishing returns in the platform. So 5:25:28 instead, the position that you actually 5:25:30 want to be in here is launching three 5:25:32 creatives into the account, but these 5:25:34 three creatives talk to different people 5:25:37 and therefore 5:25:38 reach their own audiences. 5:25:41 And with that frequency declines because 5:25:43 you're reaching different people with 5:25:44 these ads, it allows you to unlock more 5:25:46 spend because as these scale, you can 5:25:48 almost think about this circle expanding 5:25:49 as it goes to colder and colder 5:25:51 audiences. But because these are talking 5:25:53 to different people, you can see there's 5:25:54 not much audience overlap. And as a 5:25:57 function of that, you're going to be 5:25:58 able to scale to new audiences 5:26:00 effectively without just increasing 5:26:01 frequency and seeing efficiency decline. 5:26:04 Another incredibly important concept 5:26:06 here about how the platform algorithm 5:26:08 actually works is in sequencing of 5:26:10 creative and the reliability of return 5:26:12 on ad spend as you decline through the 5:26:14 account structure. So when it comes to 5:26:16 sequencing creative, the reality is and 5:26:19 I think everyone watching this will 5:26:20 agree with this which is that if you 5:26:22 just serve one ad to a user once, will 5:26:25 they buy? And the answer 99.99% 5:26:27 of the time is no. Generally speaking, 5:26:30 you need to see an ad more than once to 5:26:33 be confident in purchasing. Now, yes, 5:26:34 there are some people out there that are 5:26:36 getting one hit by ads and suddenly 5:26:38 buying. All right, that does happen. In 5:26:39 fact, it's probably happened to me at 5:26:41 one point in my life, but it's very, 5:26:43 very rare and you definitely cannot 5:26:44 build an efficient business off of those 5:26:46 people. And so, because of that, you 5:26:48 need to serve them generally more than 5:26:50 one ad. Now, you could serve them this 5:26:52 ad again a second time. However, Meta 5:26:55 put out a article ages ago. It was like 5:26:58 four or five years ago which said that 5:27:00 after two impressions of an identical 5:27:02 creative probability of purchase 5:27:04 declines precipitously. And so if you 5:27:07 look at the probability of someone 5:27:09 buying, let's say this is a percentage 5:27:12 and then you look at the amount of ads 5:27:14 that they are seeing. So this is 5:27:15 frequency of an identical ad. So we're 5:27:17 talking about the same ad here. You're 5:27:18 not showing them different ads, just the 5:27:19 same ad. When you show them the ad once, 5:27:23 there might be a certain percentage uh 5:27:25 chance of them buying. Then when you 5:27:26 show them the ad twice, it might be the 5:27:28 same. It actually might even be slightly 5:27:29 higher because on second viewing now 5:27:32 they're warmer and they're more likely 5:27:33 to buy. But then on third viewing it 5:27:36 declines massively and then on fourth 5:27:39 massively and then it pretty much 5:27:40 declines to zero. And so there is very 5:27:43 little purpose at all in showing a user 5:27:45 the same identical ad more than twice. 5:27:48 And the way to actually stress test this 5:27:50 idea that I'm telling you right now in 5:27:52 your own ad account is you can pull your 5:27:53 ad account up. You can go down to the ad 5:27:55 level, look at the last 30 days, 60 5:27:57 days, and look at frequency at an ad 5:27:59 level on cold campaigns. And what you 5:28:02 will see is that cold ads almost never 5:28:05 have a frequency above a two. And if 5:28:07 they do, you have a creative fatigue 5:28:09 issue. You do not have enough 5:28:10 diversification and volume in your 5:28:12 account. And that is a red flag. any 5:28:14 good account that we're running, which 5:28:15 is all of them, um, no one has a 5:28:18 frequency above a two on a cold out just 5:28:20 doesn't happen because if we see this 5:28:21 immediate red flag, we need to fix it. 5:28:23 We need more diversity. We need more 5:28:24 volume. Now, the big kicker here is that 5:28:28 what meta are published in that 5:28:30 statement, and if I can go and find that 5:28:32 article, I'll put it in the description 5:28:33 below, is that if you then go and 5:28:35 introduce a novel creative, the 5:28:38 probability of purchase stays high. And 5:28:41 so after seeing the same ad once and 5:28:43 then twice, if we go and serve a 5:28:45 different ad in as the third impression, 5:28:48 probability stays up here. And then if 5:28:51 we go and serve a different ad as the 5:28:52 fourth impression, probability stays up 5:28:55 here. And so we can actually keep 5:28:57 probability of purchase up by rotating 5:29:00 in new novel creative. Now the question 5:29:02 becomes, how does the algorithm do this? 5:29:04 And how does it decide the sequencing of 5:29:06 creatives to these users to try to get 5:29:08 them to convert? And this is where 5:29:10 sequencing comes into the ad account 5:29:12 structure. When meta is optimizing 5:29:14 across serving users ads, they're 5:29:16 optimizing across understanding what 5:29:19 sequence of creative actually gets the 5:29:21 purchase. So is showing users ad one 5:29:24 then add two then add three going to get 5:29:26 the click and purchase or is it showing 5:29:28 users ad three as the opening creative 5:29:31 then add two then add one going to get 5:29:33 the purchase or is it some other kind of 5:29:34 combination? Are we coming in here and 5:29:36 then going back here? There's all of 5:29:38 these different combinations of serving 5:29:39 these ads in a particular order that 5:29:41 might yield a click and might yield a 5:29:43 conversion. Now, the kicker here is that 5:29:46 meta only attributes based on last click 5:29:49 attribution. So, if we assume that the 5:29:52 first sequencing example was what is 5:29:54 actually occurring, they're going 1 2 3. 5:29:56 If the click occurs here, all of the 5:29:59 return on ad spend, all of the revenue 5:30:01 gets attributed to this ad and it looks 5:30:03 really good. Let's say it's at a 6x row. 5:30:06 Add ad two. Maybe some people buy at 5:30:08 this stage in the sequence, but not a 5:30:09 lot. And two at a 2x. And then on this 5:30:12 top ad, we're down at like a 1.4x. When 5:30:14 you look at this and you look at spend 5:30:16 distribution, let's say spend is equal 5:30:19 across all three creative. Naturally, 5:30:20 you go in here and you go, "Oh, this is 5:30:23 obvious. Turn ad one, turn ad two off, 5:30:25 cuz ad three is getting all the 5:30:26 performance." But that would be a bad 5:30:28 idea. And this is the concept in a lot 5:30:30 of our content about do not turn a 5:30:32 creative off when you are hitting KPI. 5:30:34 If this is hitting your performance goal 5:30:36 overall as an adset, don't touch 5:30:38 anything because Meta is sequencing 5:30:40 between these creatives on purpose and 5:30:42 it is creating an outcome. Meta's 5:30:44 optimization is not based on last click. 5:30:47 It's really important concept to 5:30:49 understand. Meta's optimization is 5:30:51 actually based on multiclick and view. 5:30:53 And so Meta sees that people are viewing 5:30:55 this ad, viewing this ad, then clicking 5:30:57 on this ad and buying. And therefore, it 5:30:59 distributes spend here knowing that it 5:31:01 played a role in the purchasing journey, 5:31:03 but you don't see that. Unfortunately, 5:31:05 we don't get any visibility into this at 5:31:07 all within the ads manager, which is 5:31:08 actually really annoying. And so, as a 5:31:10 product of that, we just need to trust 5:31:11 the system. We also need to think about 5:31:14 the creatives that we're putting into 5:31:15 the account. How are they working 5:31:17 through this purchasing journey? 5:31:18 Normally, if you're good at creative, if 5:31:20 you're good at creative strategy, you 5:31:22 can look at these three creatives and 5:31:24 you can go, "Oh, it's pretty obvious to 5:31:25 me that this is actually the opening ad. 5:31:27 This is then probably the second ad." 5:31:29 And then this is the converting ad 5:31:30 because this is a super unaware topunnel 5:31:33 creative. This is then a product aware 5:31:35 creative. And then this is a bottom of 5:31:37 funnel very aware creative. And so it's 5:31:39 super obvious to me just looking at the 5:31:41 three ads that that's what the sequence 5:31:42 looks like. But if you don't understand 5:31:44 creative strategy, you're going to look 5:31:46 at this and go, I don't understand 5:31:48 what's going on. Why are the rorowes 5:31:49 like this? Turn off, turn off, turn off, 5:31:51 destroy performance. We're going to go 5:31:52 into this in a lot more detail later on 5:31:54 once we start going through the testing 5:31:56 structure and actually KPIing and 5:31:58 understand what is a good ad, what is a 5:31:59 bad ad. But this is a really important 5:32:01 concept to be across. Second concept to 5:32:03 be across is that the reliability of 5:32:06 rorowaz declines as you go down in the 5:32:09 account structure. So everyone who's 5:32:11 opened up a meta ad account knows the 5:32:13 basic structure which is you have 5:32:14 campaigns you have adets then you have 5:32:16 ads right the campaign you set up the 5:32:19 budget typically and you set up the 5:32:21 overall optimization event so is it 5:32:23 optimizing for sales reach etc ads set 5:32:26 this is where all the targeting occurs 5:32:28 now people forget that these days 5:32:30 because people go oh targeting is just 5:32:32 broad so I don't even really know what's 5:32:34 happening at the adset level but it's 5:32:35 really important to reinforce targeting 5:32:37 is occurring at the ads level and then 5:32:39 down here the ads This is obviously the 5:32:40 creative that gets served to the end 5:32:42 user. Now, when we look down at the ad 5:32:44 level and we start assessing creative, 5:32:47 the issue here is everything I just went 5:32:49 through, we can look at the rorowaz 5:32:51 numbers, but they're kind of meaningless 5:32:53 because their last click attribution and 5:32:56 there's a multi-click or multi- view 5:32:58 through journey occurring here. And so, 5:32:59 I can go in and say, "Yeah, this has a 5:33:01 6x row as amazing. Let's like try to 5:33:03 scale it or make more ads like that." 5:33:05 But it's often a bad decision. So 5:33:07 instead when we move up to the ads set 5:33:09 level right here now we can look at all 5:33:12 of these returns as a group and we can 5:33:14 go let's not go into the individual ads 5:33:16 because the reliability of uh the 5:33:18 attribution here is terrible because 5:33:20 there's so much happening across these 5:33:22 different sequences. So let's just go up 5:33:23 to the ad set and look okay at the adset 5:33:25 level what's our rorowaz here and maybe 5:33:27 overall it's at a 3x. Now this is a lot 5:33:29 more reliable. It is worth noting that 5:33:31 cross serving of ads still occurs at the 5:33:34 adset level. So you can have uh multiple 5:33:38 different ad sets. So if I just throw 5:33:39 another ads set in up the top here, 5:33:42 we'll call this ads set two and let's 5:33:44 just put a couple creatives under here. 5:33:47 A sequence that can occur is that people 5:33:49 see this ad and adset one. They then see 5:33:52 this ad and then they get served another 5:33:54 adset. Now it's not prioritized by the 5:33:57 algorithm. The algorithm will always 5:33:58 prioritize serving adjacent ads under an 5:34:01 adset due to the way that targeting is 5:34:03 siloed. But there is cross adset 5:34:05 targeting. And so because of that, we 5:34:08 might have a 3x here. We might have a 4x 5:34:10 up here. This adset seems to be doing 5:34:12 better. But it's actually just because 5:34:15 this adset is picking up some bottom of 5:34:17 funnel conversions being generated from 5:34:18 this adset. That can be completely 5:34:20 avoided. As long as your structure is 5:34:22 good and you're diversifying concepts 5:34:24 across the adset level, you shouldn't 5:34:25 have much cross targeting. And you can 5:34:26 measure all this and we'll go into it 5:34:28 later on. But the whole idea is adset is 5:34:30 better. We can trust this number. we can 5:34:32 trust it more than the ad level. Then 5:34:34 once you go all the way up to the 5:34:36 campaign level, the campaign rorowaz 5:34:38 number, as long as it's on 7-day click 5:34:41 attribution is super reliable. This is 5:34:44 generally very congruent with the actual 5:34:46 P&L, as long as it's on 7-day click and 5:34:49 as long as existing customers are 5:34:50 excluded. And so because of that, when 5:34:52 we're making decisions within meta, when 5:34:54 we're thinking about structure, when 5:34:55 we're thinking about introducing 5:34:56 creative, we want to be thinking about 5:34:59 measuring at the adset or the campaign 5:35:01 level, not the ad level. And we want to 5:35:02 be thinking through this idea of 5:35:03 sequencing. So the practical 5:35:05 implications of Andrometer and the way 5:35:07 that the algorithm and ad platform 5:35:09 currently works is four-fold. Number 5:35:11 one, you need concept diversity. We're 5:35:14 obviously going to go into what a 5:35:15 concept is, how to break it down, how to 5:35:17 come up with concepts, what diversity 5:35:19 looks like later on. But number one is 5:35:20 you need diversity within the creative 5:35:22 going into the account. You al also 5:35:24 obviously need that introduced into the 5:35:26 structure so that you're having a 5:35:27 concept per adset so you don't have any 5:35:29 of this cross targeting going on. Number 5:35:31 two is you need format variety. Okay? 5:35:34 You can't unfortunately just have the 5:35:36 same format of all of your creative. And 5:35:38 the reason being is that meta will 5:35:40 prioritize formats that people resonate 5:35:42 with. I'll give you an example which is 5:35:44 that I never ever and I could scroll my 5:35:47 Instagram feed for the next 3 hours and 5:35:48 show you. I will never get a piece of 5:35:50 UGC under my profile. Just doesn't 5:35:53 happen. And it's because I never click 5:35:55 on them. I never resonate with them. And 5:35:56 so Meta inherently knows that lowfi UGC 5:36:00 content should not be served to me. And 5:36:02 so if your ad account is just loi UGC, 5:36:05 I'll never see it. Now, same thing 5:36:07 actually for DPA ads. Now DPA are very 5:36:10 bottom of funnel. This isn't really 5:36:12 cold. I'd have to be already very very 5:36:14 product aware on the website. But I 5:36:16 don't think I have ever been served a 5:36:19 DPA almost in the history of uh the ads 5:36:22 that I get served on my personal 5:36:23 profile. Now that is likely a product of 5:36:26 the fact that DPAs are not a creative 5:36:29 type that I resonate with. I never click 5:36:30 on them. And so if you just had UGC and 5:36:32 DPAs, it's very unlikely you'd ever 5:36:35 reach me. So you need to be making sure 5:36:36 you have as much diversity in format 5:36:38 type. And I'm going to be giving you all 5:36:39 the formats and how you should be mix 5:36:40 and matching them later on so that you 5:36:42 can reach as many people as possible. We 5:36:43 then have hook quality. So this is going 5:36:46 to be a big section of this video, but 5:36:48 with Andromeda, any kind of persona call 5:36:50 out that occurs within the actual 5:36:52 creative is going to go and prioritize 5:36:54 that persona. So if I make an ad and at 5:36:57 the start I say, if you're a business 5:36:58 owner doing over $5 million a year, 5:37:00 Andromeda is going to read the 5:37:02 transcript. It auto transcribes every 5:37:03 single ad and it's going to go, "Okay, 5:37:05 let's try to find a similar audience to 5:37:06 what he just said. What if my ad or my 5:37:08 content has nothing to do with that 5:37:10 audience? Well, then we've just push the 5:37:11 creative to an audience that's not 5:37:12 actually relevant. And people do this a 5:37:14 lot. People's hook actually isn't as 5:37:16 good as they think it is. People call 5:37:18 out a persona without problem agitating. 5:37:20 There's there's so many different issues 5:37:22 that actually lie in hooks, which is why 5:37:23 we're spending so much time on it cuz I 5:37:25 think people are terrible at hooks. Most 5:37:26 ads I see, they're just not good. And 5:37:28 then lastly is that with Andromeda, 5:37:30 volume is absolutely critical. The spend 5:37:34 capacity of most ads isn't very high. 5:37:36 Most ads that you enter the into the 5:37:37 account, if we look at mean ad spend per 5:37:40 ad unit, sits at about $700 to $1.5,000 5:37:44 depending on the ad account. And so 5:37:46 generally, you can't get much spend 5:37:47 through most ads. So you need volume to 5:37:49 compensate, but you also need strategy 5:37:51 at the same time because if you're just 5:37:53 uh putting sheer volume into the 5:37:54 account, which I've seen so many people 5:37:56 do with no actual strategy behind the 5:37:59 volume, it doesn't actually work. So 5:38:00 let's run through concept architecture. 5:38:02 This is how you need to be thinking 5:38:03 through the segmentation of different 5:38:05 creative. This is how your naming 5:38:06 conventions should be set up on meta as 5:38:08 well. This is how you should be 5:38:09 ideulating in a Google sheet or whatever 5:38:11 format you want to think through this. 5:38:12 And we'll also talk about the AI 5:38:14 implications into this ideation process. 5:38:16 So what actually is a concept? A concept 5:38:19 is a persona, angle, and offer. And so 5:38:21 to have diversity within the creative 5:38:23 that's going into the account, you need 5:38:25 one of these three variables to change. 5:38:27 If the offer changes, it will resonate 5:38:29 with a different audience, with a 5:38:30 different pool of people because it's 5:38:32 either a different product or if it's 5:38:34 it's a different value proposition or 5:38:35 packaging of the product. If the angle 5:38:37 changes, obviously, it's going to 5:38:38 resonate with with someone different 5:38:39 because you're problem agitating on 5:38:41 something else. And so, if you sold skin 5:38:43 care, you might start by problem 5:38:44 agitating on wrinkles, but then you 5:38:46 might have another ad that problem 5:38:48 agitates on that. They're going to reach 5:38:49 two very different audiences even though 5:38:51 it's the same product because the angle 5:38:53 changed. And then lastly, we have a 5:38:55 persona change. Now, the easiest way to 5:38:56 think through a persona change is 5:38:58 obviously who are we talking to, but 5:39:00 also who is actually in the ad. And so, 5:39:01 if there's a 60-year-old woman in the 5:39:03 ad, it's obviously going to resonate 5:39:05 with a very different target demographic 5:39:07 than if there is a 20-year-old woman in 5:39:09 the ad. And so, because of that, we can 5:39:11 adjust the persona and that will 5:39:12 inherently adjust who resonates with the 5:39:14 creative and who it will target on the 5:39:15 platform. And so one of the easiest ways 5:39:17 to ideulate here and come up with almost 5:39:19 infinite creative concepts is to start 5:39:22 ideulating on who are all the different 5:39:24 personas that we want to reach. What are 5:39:25 the core personas in the business? And 5:39:27 you can go through and you can probably 5:39:28 come up with two or three or four. And 5:39:30 then if you really drill deeper, you can 5:39:32 come up with tons more. So yeah, we 5:39:34 could go like females age between 40 to 5:39:36 45 etc. But then we can start digging 5:39:39 into like what are the actual pain 5:39:40 points? And so females aged 20 to 25 5:39:43 dealing with acne. Females aged between 5:39:45 30 to 35 dealing with acne which is a 5:39:47 slightly different problem because acne 5:39:48 shouldn't be occurring at that age. So 5:39:50 maybe it's hormonal acne. And then we 5:39:51 can go women aged between 40 to 45 with 5:39:55 acne as well. That's also a completely 5:39:57 different problem that should be fixed 5:39:58 by now. And so what are the different 5:40:00 underlying causes that we can then 5:40:02 resonate with in that persona? And so 5:40:03 you can build personas out in ridiculous 5:40:06 detail based on all of the problems that 5:40:09 the product solves. By the way, this is 5:40:11 the case in every industry as well as 5:40:13 fashion. So, in fashion, people really 5:40:15 struggle to get their head around 5:40:16 concept architecture, but with personas, 5:40:19 it's the same thing. Most large retail 5:40:21 fashion brands will break their target 5:40:23 demographic down into like four core 5:40:25 personas, and they'll give them names. 5:40:26 But the reality is, you can go way 5:40:27 deeper than this because a persona could 5:40:30 be a particular problem that a 5:40:33 particular person is facing. And so 5:40:35 there's this particular person who's a 5:40:37 mom aged between 30 to 35 with young 5:40:40 kids. And the problem that they face is 5:40:42 they need an affordable dress that can 5:40:44 go from a work dress into a night dress 5:40:46 on a Friday afternoon. That is a 5:40:47 persona. That's a persona that you could 5:40:48 make infinite creative for for that 5:40:51 exact particular painoint and 5:40:52 circumstance. And you could probably 5:40:54 think of 400 examples of that. When we 5:40:56 move into angle, it's the presentation 5:40:58 of how we're actually uh going after the 5:41:00 problem. So what is the angle? What is 5:41:02 the value proposition? Why would you 5:41:04 purchase the product? And then the offer 5:41:06 is obviously really straightforward, 5:41:07 which is what are we actually offering? 5:41:09 The key here is personas aren't 5:41:11 demographics. They're archetypes that 5:41:13 are defined by problems, desires, and 5:41:15 buying triggers. So, we want to be 5:41:17 thinking about who are our avatars. You 5:41:19 want to think hard on this. Spend 20 5:41:21 minutes on it. Then, what problems do we 5:41:24 actually solve for these people? Then 5:41:27 what are the differences in pain points, 5:41:30 language as to how we talk to this 5:41:32 person and the proof that is required 5:41:34 for this person. And once we understand 5:41:36 these three, we should be able to build 5:41:38 out a whole lot of personas which you 5:41:41 can then once that is done manually go 5:41:44 to an LLM like Claude or ChatBT or 5:41:46 Gemini. Give it the entire persona list 5:41:49 and tell it come up with 20 to 30 more 5:41:51 and give it as much context as possible 5:41:53 which you should already have loaded in 5:41:54 there in some capacity about the 5:41:56 business about the target demographic 5:41:57 etc. Claude, OpenAI, LLMs, they're only 5:42:01 going to give you averages. And so if 5:42:03 you just give it three or four personas 5:42:05 and you give it your website and you 5:42:06 say, "Hey, come up with some more 5:42:07 personas," you're going to get an 5:42:08 average outcome. And everyone's going to 5:42:09 be doing it and you're going to get the 5:42:10 same as them. And you're going to revert 5:42:11 to the main. So what you want to do is 5:42:13 if you are going to do this, you need to 5:42:14 put all the manual effort into actually 5:42:15 building out proper 30 to 40 personas 5:42:18 that are actually good. And then you 5:42:19 need to give it an unbelievably wide 5:42:21 amount of context as to the business and 5:42:22 the target demographic and everything 5:42:24 you can so that it can actually idiate 5:42:26 properly on another 20. And of the 20 5:42:28 that it produces, you should just be 5:42:29 putting that in there. You should be 5:42:31 refining down. Probably cut down to four 5:42:32 to five. Give that back in. Say these 5:42:34 four to five were good. Ideate on 5:42:35 another 10 to 20. And then that way you 5:42:38 can beef this out from 20 to 30 personas 5:42:40 or whatever you get to and double the 5:42:42 number. One really important note here 5:42:44 is that people die in persona work by 5:42:48 keeping the personas too broad. broad 5:42:50 personas end up producing forgettable 5:42:53 creative that is nowhere near specific 5:42:55 enough to the target demographic that 5:42:57 you want to resonate with. And so if you 5:42:59 come away with a persona while you're 5:43:00 building this out, which is, let's say, 5:43:02 women age between 25 to 45, this is a 5:43:07 terrible persona. The creative that 5:43:09 you're going to make for this audience 5:43:10 is nowhere near specific enough that 5:43:13 it's going to resonate with enough 5:43:14 people to be able to hold a good amount 5:43:16 of spend. as an example, instead of 5:43:17 going for the persona busy professional, 5:43:20 which I actually see so much when when I 5:43:21 tell people to do this concept buildout, 5:43:23 and instead you would want to delineate 5:43:25 this into a consultant who flies twice a 5:43:28 week and needs a carry-on wardrobe that 5:43:30 works for boardrooms and bars. Way more 5:43:32 specific. And now with that persona call 5:43:33 out, you're already thinking about how 5:43:35 to talk to this person. You can almost 5:43:36 visualize who this is, what their 5:43:38 problem is, what we need to solve, what 5:43:39 the creative should look like. It gives 5:43:41 so much more specificity into the actual 5:43:43 briefing and production process. And 5:43:45 then guess what? thought we actually 5:43:46 reach this audience properly. But when 5:43:48 we say busy professional, what is a busy 5:43:50 professional? That is so wide. There are 5:43:52 so like the age is all over the place. 5:43:54 Their gender is all over the place. What 5:43:55 do they actually do? What do they get up 5:43:57 to? Why do they even need this product? 5:43:58 This is so broad that you will you will 5:44:01 end up with mediocre creative. You are 5:44:03 better off making five ads for this 5:44:05 target demographic than five for this. 5:44:07 Now, five for this will scale harder if 5:44:09 you can actually do a proper unaware ad 5:44:12 that is going to resonate with this 5:44:13 entire pool of people. Yes, it's a 5:44:14 bigger total addressable market. So, 5:44:16 yes, you'll be able to hold more spend. 5:44:18 But the reality of you making a good ad 5:44:19 for an audience this broad is actually 5:44:21 very unlikely. And so, you are way 5:44:23 better off making more ads for very 5:44:25 specific personas. And then as you start 5:44:28 to become a eight mid 8 figure brand, 5:44:30 you can start to play around in trying 5:44:32 to get ads to work for very broad 5:44:34 audiences. But the skill level required 5:44:36 here is so high and it is just not 5:44:38 required for most ad accounts. most ad 5:44:41 accounts you can just go hyper specific 5:44:42 and get way better results than needing 5:44:44 to go and try to target the entire of a 5:44:46 population. Another example here is 5:44:48 instead of doing, let's say, 5:44:49 healthconscious parent, you could target 5:44:52 a mom who's tried six different 5:44:54 supplements because her toddler won't 5:44:56 eat vegetables. Once again, way more 5:44:58 specific. Now, you could go away and you 5:44:59 could write a script for that. You know 5:45:00 exactly what that ad looks like. You 5:45:01 know, who should be in it. You know who 5:45:03 you're talking to. You know what the 5:45:04 proof needs to be for that particular 5:45:06 target demographic. you know what the 5:45:07 language needs to be and you know the 5:45:09 pain points of that particular persona. 5:45:11 Now an angle is the specific argument or 5:45:13 perspective that you take when you're 5:45:16 speaking to a persona. This is where a 5:45:18 lot of creative strategy actually ends 5:45:19 up failing because brands default to 5:45:22 product features rather than crafting 5:45:24 genuine angles that resonate with the 5:45:26 persona. So I'm going to give you a lot 5:45:28 of tactical application here. Number one 5:45:30 is you want to problem agitate in some 5:45:32 capacity. You don't need to do this in 5:45:33 all of your creatives. In fact, 5:45:35 especially if people aren't even problem 5:45:36 aware, then you can't even problem 5:45:38 agitate, right? But you want to be 5:45:39 leading with pain as much as possible. 5:45:41 And you want to make the viewer feel the 5:45:43 pain, feel like they have been heard 5:45:45 before you go and introduce the 5:45:46 solution. Another great angle here is 5:45:48 approaching with a contrarian truth. So 5:45:50 you want to challenge a common belief 5:45:52 that people have which allows them to 5:45:54 reframe. An example of this, 5:45:56 particularly just as a hook, is 5:45:57 everything you've been told about X is 5:45:59 wrong. Everything you've been told about 5:46:01 the protein industry is wrong. And then 5:46:03 you go into obviously an educational 5:46:05 piece that then might problem agitate 5:46:07 that then obviously loops into a 5:46:08 solution. You want to typically always 5:46:09 have social proofing in the angle. You 5:46:11 want to build some kind of authority or 5:46:13 expertise in the domain. So this is more 5:46:16 specifically domain expertise. So you 5:46:18 don't need for example a doctor and a 5:46:20 lab coach showing up in every single ads 5:46:22 for every industry. Right? If you're 5:46:24 selling a running product, you want a 5:46:25 marathon runner showing up. You want 5:46:27 someone with authority in the domain 5:46:28 that you're trying to sell in. Curiosity 5:46:30 is great and is commonly in most angles 5:46:32 that do well. And this is specifically 5:46:34 creating what's called a curiosity gap. 5:46:36 And so when you open up a video asset, 5:46:38 and you can actually do this in image 5:46:39 assets, too, but when you open up the 5:46:41 video, you open up and create a 5:46:43 curiosity gap or an information gap. As 5:46:46 an example, you'd say, "Here's what the 5:46:48 top 1% of millionaires know about saving 5:46:52 money that they don't want you to know 5:46:53 about." And then you don't answer it 5:46:55 straight away. And you go into a story 5:46:56 and you go into something else. You can 5:46:57 go into any of these other aspects. You 5:46:59 problem agitate, you have a contrarian 5:47:00 truth, you go into some degree of social 5:47:02 proofing, although that's probably 5:47:03 stacked towards the back of the ad. And 5:47:05 then eventually the ad goes on and on 5:47:07 and on. You pull people through this 5:47:08 journey, you educate them on everything, 5:47:09 and then you end up actually answering 5:47:11 the curiosity gap. And so you didn't 5:47:14 have any information. You want to get to 5:47:16 the information and eventually the 5:47:17 information is provided to you. And so 5:47:18 that's how you hold people through an 5:47:20 ad. All good angles also have some 5:47:22 degree of comparison or objection 5:47:25 handling. And so often people who are 5:47:27 watching your ad have tried to solve 5:47:29 their problem through other solutions. 5:47:31 And so you want to tell them why your 5:47:33 solution will work versus all of the 5:47:35 competitors. You've probably tried this. 5:47:36 You probably tried this. You probably 5:47:37 tried this. If it's a piece of user 5:47:39 generated content, you'll commonly see 5:47:40 the actual person in the script going, 5:47:42 "I tried this just like you would have 5:47:44 and it didn't work for me. I then tried 5:47:45 this. I thought it was overhyped. It 5:47:46 didn't work. And then finally, I was 5:47:48 like, okay, I'm going to give one more 5:47:49 thing a shot." And it actually worked. 5:47:50 And then this one is the least 5:47:52 compliant, but it is the best performing 5:47:55 component to add into an angle. So, I'm 5:47:57 going to throw it in, which is a 5:47:58 transformation, aka a before verse 5:48:02 after. Uh, before verse afters, Meta 5:48:04 does not like them. You have to sneak 5:48:06 these into creatives in a way that the 5:48:08 algorithm doesn't flag it. But these are 5:48:10 some of the best performing creatives 5:48:12 ever because people want to see visually 5:48:13 a before and after. I think this is 5:48:15 basic human psychology. Before and 5:48:17 afters will always crush for as long as 5:48:19 they exist. And because of that, Meta 5:48:20 doesn't like you running them. So, when 5:48:23 we're thinking about the angle, we're 5:48:24 starting at the persona and then we're 5:48:26 thinking about what angle and approach 5:48:28 do we need to be able to resonate with 5:48:30 this target persona. And what you will 5:48:31 notice is I've given you a lot of 5:48:33 tactical application that can be 5:48:35 included or built together into a script 5:48:38 that creates the angle. So, the real key 5:48:41 with the angle is this. You can simplify 5:48:42 it into a one sentence, which is if our 5:48:44 persona is women with kids aged 35 to 40 5:48:47 that are having energy issues, we're 5:48:50 going to then sell them creatine powder 5:48:52 and the angle is going to be that it 5:48:55 solves energy issues based on particular 5:48:58 studies and then we can continue 5:49:00 chaining it out. The thing is the angle 5:49:02 realistically if you're going to write 5:49:03 this out properly, we're just talking 5:49:04 about the script. Okay? And so you just 5:49:06 want to write the whole script. If it's 5:49:08 a static, this is obviously going to be 5:49:09 a bit tighter and you could probably 5:49:10 encapsulate this in a few sentences. But 5:49:12 the idea is we want to figure out what 5:49:14 angle are we approaching selling this 5:49:16 person on with a myriad of different 5:49:18 strategies that's going to resonate with 5:49:20 them. And so if we take for example that 5:49:23 woman age 40 to 45 with creatine and 5:49:26 energy issues, well, we're going to want 5:49:28 social proofing, but we want to we're 5:49:29 going to want social proofing for that 5:49:31 particular persona. So we're going to 5:49:32 want either case studies around persons 5:49:34 people like that or the clinical studies 5:49:36 that we're going to reference which is 5:49:37 going to give us expertise and authority 5:49:39 needs to reference their particular age 5:49:41 and gender. We want to think about 5:49:43 problem agitation on that particular 5:49:45 persona. So how are we going to agitate 5:49:47 to that problem? Are we going to pull in 5:49:49 some kind of contrarian truth or 5:49:50 curiosity gap that's going to hold them 5:49:52 into the asset? Can we do any kind of 5:49:54 comparisons to stuff that they've tried 5:49:56 in the past? Has this persona tried 5:49:58 things or is this very new to them? And 5:50:00 then can we pull in any kind of 5:50:02 transformation at the end as a piece of 5:50:04 social proof to back into the creative? 5:50:06 And so we choose the elements that we 5:50:08 want. We think about how we're going to 5:50:09 slice and dice these into a script and 5:50:11 that becomes the angle. And we're going 5:50:12 to build out some actual examples of 5:50:14 this in a moment. But before we do, I 5:50:17 want to talk about the testing priority 5:50:19 order. Which of these three components 5:50:21 has the largest impact? What should you 5:50:23 be thinking about testing the most of or 5:50:26 the least of? Number one at the top is 5:50:27 angles. So this is exactly what we're 5:50:29 just discussing. This is the message or 5:50:31 the argument. The reason why this is 5:50:33 such a high priority is because we can 5:50:35 sell to any persona that is relevant to 5:50:37 the product. You should be able to 5:50:38 effectively look at who has bought from 5:50:39 you in the past. What persona do they 5:50:41 sit in? Why did they buy from you? And 5:50:43 then we just need to craft an angle to 5:50:45 that particular persona. If that persona 5:50:46 isn't purchasing from us, we know that 5:50:49 it's actually really an angle or an 5:50:50 offer issue. It's not an issue with the 5:50:52 persona cuz we've already proven that 5:50:53 these people buy from us. Okay? And so 5:50:55 when something's not working, we 5:50:56 typically don't want to look at the 5:50:58 persona as long as you did it well. Like 5:51:00 let's not do broad 25 to 45year-old 5:51:03 females. Let's get hyper specific into a 5:51:05 very specific person that has purchased 5:51:07 from us before and let's target them. 5:51:09 But then ultimately it's going to come 5:51:10 down to an angle or offer issue. Which 5:51:12 is why angle is the first thing you 5:51:14 should test. If things aren't working, 5:51:15 you need to rotate out and improve the 5:51:17 angle. And then the second is the offer. 5:51:19 Now fundamentally the offer actually 5:51:21 might be the number one most important 5:51:22 thing out of all of this. But the issue 5:51:24 is is that you actually don't have much 5:51:26 flexibility in changing the offer. What 5:51:28 do I mean by that? The offer is two 5:51:29 things. The offer is the product and 5:51:31 then the offer is the value proposition 5:51:34 or the way that we're framing value. 5:51:36 Now, when it comes to the product, if 5:51:37 you have a single product and ads aren't 5:51:40 working well, I can't really tell you to 5:51:41 change the product. Right? Now, you 5:51:42 could you could change the whole 5:51:43 business, but the reality is is that the 5:51:46 product is the product, we need to 5:51:47 change the angle or the persona, right? 5:51:49 So even though this is the most 5:51:50 important thing, the product ultimately 5:51:52 is king out of anything, we don't have 5:51:53 much flexibility here. Now if you have a 5:51:55 very large skew count and maybe this is 5:51:57 a new product that's being launched into 5:51:58 an existing 20 product portfolio, then 5:52:01 sure, then we can go and say, is this an 5:52:03 offer issue? Is this an issue with the 5:52:05 product? And we can start to 5:52:06 troubleshoot accordingly. But for most 5:52:07 people, you're pretty stuck on the 5:52:09 product. Then when it comes to the value 5:52:10 proposition, the value proposition, 5:52:12 we've got a couple options, right? We 5:52:14 can go and try testing bundles. We can 5:52:16 go and try testing like a gift with 5:52:18 purchase. We can go and try all of these 5:52:20 different repackaging of the offer, 5:52:24 right? We can do like tid as well, so 5:52:25 you get percentage off when you buy two 5:52:26 or three. We can go and throw in like a 5:52:28 subscription offer. So, there's a lot of 5:52:30 testing that we can do here. And we do 5:52:31 this with a lot of clients is we'll come 5:52:32 up with a bunch of different offers and 5:52:34 how we should reframe value to try and 5:52:36 get the funnel working more effectively. 5:52:37 But ultimately, you are limited here. 5:52:39 There's only so many offers you can 5:52:41 test. And we run into this issue which 5:52:42 is like we've tested like seven 5:52:44 variations of how we can spin these two 5:52:46 to three products and now we're 5:52:48 relatively stuck and the only way that 5:52:49 we can increase the value perception on 5:52:51 the offer is to just do harsh deep 5:52:53 discount and we don't want to do that. 5:52:55 We don't want to go and just eat 40% 5:52:57 into margin and build a brand off a 5:52:59 heavy discount reliance. And so although 5:53:01 the offer is unbelievably important, 5:53:03 this is a huge variable that we do a lot 5:53:05 of testing on and we do a lot of offer 5:53:06 testing for clients. It's something that 5:53:08 eventually you're going to hit a wall 5:53:10 on. Like there's only so much offer 5:53:11 testing you can actually do and once 5:53:13 you've found the best performing offer, 5:53:14 you're stuck on it. And so that is where 5:53:16 then we want to be doing a lot of switch 5:53:18 ups in personas. And personas is 5:53:20 ultimately what's actually going to give 5:53:21 us breadth and volume in the account. 5:53:23 And then I have number four in here. 5:53:24 There is actually a fourth component of 5:53:26 concept design. Now this didn't really 5:53:29 used to exist pre-andromeda, but it's an 5:53:32 additional component that we've added in 5:53:34 recently. And it's because of the way 5:53:36 that Andromeda will take different 5:53:37 formats and serve it to different 5:53:39 people. So, as I said before, I don't 5:53:40 get UGC content. I don't get dynamic 5:53:42 product ads. And so, you could do all 5:53:44 the mix and matching that you want of 5:53:46 personas, of angles, of offers to try to 5:53:48 hit me. But, if the only ads in the 5:53:50 account is UGC and DPAs and so there is 5:53:53 a fourth component to be able to reach 5:53:54 more people, which is the format. You 5:53:56 can have the same persona, same angle, 5:53:58 same offer, and you can change the 5:53:59 format and you will reach a different 5:54:01 audience. However, this is nowhere near 5:54:03 as effective as these other components. 5:54:05 If you do this and then you change the 5:54:07 format, the reality is you're probably 5:54:09 going to hit a lot of the same people 5:54:10 cuz you are talking to the same people. 5:54:12 You are going with the same angle and 5:54:13 you are offering them them the same 5:54:14 thing. You'll get a bit of novel reach. 5:54:16 You'll obviously reach people that don't 5:54:17 resonate with the current formats, but 5:54:19 this isn't like a huge unlock great 5:54:21 variable in the account. And that's why 5:54:22 it is down at number four in the testing 5:54:24 priority. Now, the big issue here and 5:54:26 the reason why I spend so much time 5:54:27 going through the testing priority order 5:54:29 is everyone gets this completely the 5:54:30 wrong way round. People spend all of 5:54:32 their time testing formats. They go, 5:54:35 "Oh, let's test some UGC cuz UGC will 5:54:37 work." UGC as a concept doesn't work. It 5:54:40 is the script. It is the persona. It is 5:54:42 the angle. It is the offer that is 5:54:44 contained within this type of content 5:54:46 that is what works. Just like a 5:54:48 professional photo shoot doesn't work. 5:54:50 The format isn't what works. It's what's 5:54:52 contained within the format. Then people 5:54:54 move up to the persona and they go we 5:54:56 need to reach different people. Okay, we 5:54:57 understand that Andromeda we need to 5:54:59 talk to different people. So let's talk 5:55:00 to different people. This is the second 5:55:01 priority. It is not because if your ads 5:55:04 aren't working fundamentally you have 5:55:06 bad messaging and you have bad arguments 5:55:08 for why they should buy. The creative is 5:55:10 just bad. And creative being bad is 5:55:12 normally a function of the angle and the 5:55:14 scripting. And so this is what you 5:55:15 should be fixing. Changing up the 5:55:16 persona is just you trying to target 5:55:18 different people with terrible ads. Then 5:55:19 they go to the offer and they say, 5:55:20 "Well, all the big influencers, Hormosi 5:55:23 says offer is king. That's the thing 5:55:24 that matters the most. We need to go to 5:55:25 the offer." Sure, but you're going to 5:55:27 once again fatigue offer testing pretty 5:55:29 quickly. There's only so much you can do 5:55:30 with your one or two product. And then 5:55:32 they end up at the angle, which is their 5:55:33 lowest priority, but it should be the 5:55:35 highest. So, this is how you need to be 5:55:36 thinking through testing priority. Now 5:55:38 the practical application of thinking 5:55:40 through this format type actually 5:55:41 translates directly into the account 5:55:44 structure which is that when you have a 5:55:46 campaign you want your adsets to be 5:55:48 segmented out based on concept. And the 5:55:51 reason being is that the targeting 5:55:53 exists at the adset level. And so if you 5:55:56 go and put in as I said before a bunch 5:55:58 of different ads that are a bunch of 5:56:00 different concepts resonating with 5:56:01 different people well then the ad set 5:56:04 that's trying to figure out who to 5:56:05 target is going to get confused. It's 5:56:07 going to go, "Wait a second. This is 5:56:08 resonating with men, old men. This is 5:56:10 resonating with young women. Like, why 5:56:12 are there all these different angles in 5:56:13 here? Who do we actually target?" Now, 5:56:15 yes, there is some siloing of target 5:56:17 demographic targeting at an individual 5:56:18 ad level, but a lot of it still sits at 5:56:20 the adset level. And so, you're going to 5:56:22 get worse performance when you go and 5:56:23 bundle all different concepts together. 5:56:24 Not only that, because you can't trust 5:56:26 the reliability of rorowaz at an ad 5:56:29 level very well. When you go and put all 5:56:30 these together and you don't hit KPI, 5:56:32 you do hit KPI. Let's say maybe you got 5:56:34 a 3x rorowaz here. Well, then you're 5:56:36 going to look at these ads and not 5:56:39 actually understand what concept works 5:56:41 because there's three concepts here. So, 5:56:43 what is the learning from this adset? 5:56:44 Well, if it was all under one concept, 5:56:46 one persona, one angle, one offer, we 5:56:47 could go, "This concept works really 5:56:49 well. Let's go and make more like this. 5:56:51 Let's double down in this direction. 5:56:52 Let's not overlever. Let's think about 5:56:54 portfolio management, but we should 5:56:56 definitely put a little bit more 5:56:57 resourcing here cuz there's a squeeze 5:56:59 available." If instead you have three 5:57:00 different concepts in here and it's 5:57:01 hitting KPI, what what's the conclusion? 5:57:03 is is all the concepts working, but this 5:57:05 one doesn't have a good rorowaz. But we 5:57:06 know that the rorowaz at the ad level 5:57:07 isn't reliable. So like what is it? How 5:57:09 do we draw conclusions out of a campaign 5:57:11 structure with mixed up concepts at the 5:57:13 adset level? You can't. It's very 5:57:14 difficult to do. You just got to hope 5:57:15 your ads are good and that you get a 5:57:17 good return. But if you want to actually 5:57:18 have any kind of learning feedback loop, 5:57:20 you want to be structuring the account 5:57:23 based on concepts. All right. So, here's 5:57:25 a full buildout that I've just put 5:57:26 together of a persona angle and offer 5:57:29 that I think would actually crush the 5:57:31 formats we can add as we go along here. 5:57:34 So, and you're going to have to excuse 5:57:36 my uh writing. Persona, casual runner 5:57:39 training for a sub 4-hour marathon. It's 5:57:41 a male. They're a gym goer. So, they're 5:57:44 not just a runner, but they've gone to 5:57:45 the gym. They follow XY Z influencers. 5:57:48 So, there's particular influences that 5:57:49 we've called out, and we know that 5:57:50 they're probably following them in some 5:57:52 capacity. and they make about 70k a year 5:57:55 perom. This is important in terms of 5:57:57 framing value around what we're actually 5:57:59 going to try to sell to them. So where 5:58:00 do they actually sit in terms of income 5:58:02 level? In terms of the angle, we're 5:58:04 going to come in with science backing 5:58:06 because we know that this person 5:58:08 probably follows someone like Andrew 5:58:09 Hubman. So we can come in with at least 5:58:11 an angle around him. We're going to go 5:58:13 through how creatine improves recovery, 5:58:15 but it doesn't improve recovery in 5:58:17 general because we want to talk to this 5:58:18 persona. So it improves recovery for the 5:58:21 long run. Now people are only going to 5:58:23 be doing a long run if they're training 5:58:25 for a marathon. That enables a fiveinut 5:58:27 improvement on the marathon time. So 5:58:30 we're directly connecting the entire 5:58:31 angle to the persona and we're giving 5:58:34 them an expected outcome. Then the offer 5:58:35 is relatively straightforward. Creatine 5:58:37 gummies buy two get one free. Now this 5:58:40 offer was selected by myself just due to 5:58:42 the income level here. So, we probably 5:58:43 want increased perceived value cuz 5:58:45 creatine gummies are generally quite 5:58:46 expensive and it's going to be hard to 5:58:48 push it on to someone and to reframe 5:58:50 them against creatine for running the 5:58:53 actual format here. This is where we can 5:58:55 chop and change this and we can have 5:58:56 like 10 different ads, right? We could 5:58:57 go into a VSSL format. Probably not 5:58:59 going to work as well on this age 5:59:02 demographic, but it is a potential 5:59:04 option. We could go into UGC, which I 5:59:06 think is going to be great. We could go 5:59:07 into founder talking head. We could go 5:59:09 into a more educational hi-fi styled 5:59:12 piece. Uh you could also actually 5:59:14 translate this into a carousel. Formats 5:59:16 go on and on, but they're probably the 5:59:18 starting ones that I'd go for. VSSL I 5:59:20 might just leave out for now because the 5:59:21 age is too low for a VSSL to work on. 5:59:24 And so that is one concept where now we 5:59:26 can come into here and we can start 5:59:28 changing one of these variables to be 5:59:30 able to mix and match it to different 5:59:31 target demographics. And so the reality 5:59:33 is that this angle will also work for 5:59:37 women that are training for a marathon. 5:59:39 So we can change over that persona and 5:59:40 we have now a whole different uh concept 5:59:43 that we can start making ads for. Now we 5:59:45 can tweak the angle a little bit and 5:59:46 rather than making it for long days, we 5:59:49 could make it for people that are trying 5:59:50 to get up to a 5k run. And if you're 5:59:53 just started running and you're trying 5:59:54 to get to a 5k run, creatine is a great 5:59:56 way to build up initial improvements in 5:59:58 recovery and therefore we can now 5:59:59 resonate with a different persona. So 6:00:01 you can start to mix and match this 6:00:02 accordingly to reach way more people. 6:00:04 Now on hook strategy, everyone thinks 6:00:06 that the hook of an ad is a trick to try 6:00:08 to get someone to watch the video. But 6:00:10 the hook of an ad is a play towards 6:00:12 relevancy. What you were doing is 6:00:14 promising the user that what they're 6:00:16 about to watch is relevant and 6:00:17 contextual to them. And that distinction 6:00:20 matters enormously because a lot of 6:00:22 people think that they should just be 6:00:23 baiting people into watching. They 6:00:25 should be trying to hook them through 6:00:26 some kind of meme at the start. And yes, 6:00:28 that'll give you really high hook rates 6:00:29 if that's what you want to optimize for, 6:00:31 but it's not actually going to give you 6:00:33 conversions. Now, there's actually three 6:00:34 different layers to the hook. You have 6:00:36 the visual. This is what's actually on 6:00:38 the screen. You then have audio. This is 6:00:41 obviously the underlying audio. And then 6:00:43 you actually have the copy that overlays 6:00:46 the visual. This could also be the 6:00:47 primary text that sits above the ad. 6:00:49 Now, the reason why I'm even putting 6:00:51 this as a section in the video is 6:00:53 because the hook is so important in 6:00:55 video assets because 80% of people never 6:00:58 watch past the hook. And so, if you're 6:01:00 going and serving ads into the 6:01:01 marketplace and 80% of people never even 6:01:04 watch your ad cuz they don't like the 6:01:05 hook, so they just skip straight past 6:01:07 it, well, it's pretty critical then for 6:01:09 us to try to optimize the hook. Okay? If 6:01:11 most people are never going to see past 6:01:12 this point in the video, then let's make 6:01:14 sure this video is relevant, providing 6:01:16 value in some capacity or trying to pull 6:01:17 people through the video that are 6:01:19 relevant and need to watch. An example 6:01:20 of this in terms of resource allocation 6:01:22 is that you can spend an hour going and 6:01:24 making two hooks and then five different 6:01:27 bodies for this ad. Or you could go and 6:01:30 spend the same amount of time but just 6:01:32 make two bodies and then 15 hooks. And 6:01:35 you will probably always get better 6:01:37 performance here given that the actual 6:01:38 scripting is good on the bodies. And the 6:01:40 reason being is that the delta available 6:01:42 in performance increase is normally 6:01:44 higher in the hook than it is in the 6:01:46 body. Once again, assuming scripting is 6:01:48 good. If you're not good at scripting, 6:01:49 well then like one of these bodies might 6:01:51 actually be a good script and then it 6:01:52 will hit and you'll get a 6:01:53 misrepresentation of the actual impact 6:01:54 here of the different components of the 6:01:56 ad. Now, in terms of rating whether a 6:01:58 hook is good, cuz I do this a lot. I do 6:02:00 a lot of reaction videos on Instagram 6:02:02 saying this is a bad hook, this is a 6:02:03 good hook, and I'm rating how well 6:02:05 people have done in putting a hook 6:02:07 together. Well, there is actually a 6:02:09 grading system that you can think 6:02:10 through as to how good your hook 6:02:12 actually is. Number one is clarity. Can 6:02:15 a stranger understand what you are 6:02:18 actually talking about within the first 6:02:19 3 seconds or are you just trying to 6:02:21 confuse them as a way to hook them in? 6:02:22 Number two is relevance. Now, people 6:02:24 mistake this for a persona call out. 6:02:26 They mistake this for me taking the 6:02:28 persona that I was talking about before 6:02:30 and just calling them out at the start. 6:02:32 Hey, if you're a mom that has kids and 6:02:34 you're age 30 to 35 and you have this 6:02:36 problem, well, listen up. Persona call 6:02:38 outs are super overrated and it's 6:02:40 because no one wants to hear themselves 6:02:42 called out. People want to hear the 6:02:44 problem that they're actually dealing 6:02:45 with. And so you'll always see much 6:02:47 better performance agitating on a 6:02:49 problem than you will on any kind of 6:02:51 persona call out. And I have this from 6:02:53 personal experience. Anytime I get an ad 6:02:54 saying agency owners who need XY Z, 6:02:57 listen up. I never watch the ad. I move 6:02:59 on. But if they agitate on a specific 6:03:01 problem, it feels much more genuine. 6:03:03 actually wants to I actually want to 6:03:04 watch through and find out what the 6:03:05 solution is to the problem that I'm 6:03:07 actually currently facing. So you want 6:03:09 it to be relevant, but persona call outs 6:03:12 are nowhere near as effective as just 6:03:14 making the problem agitation relevant to 6:03:16 the person. We then have novelty. This 6:03:18 isn't a requirement, but this is a 6:03:19 really good way to create white space 6:03:22 within the ad that you're making. And so 6:03:24 this concept of whites space is also 6:03:26 this concept of purple ocean theory, 6:03:28 which is that you have a blue ocean over 6:03:31 here. You then have a red ocean over 6:03:33 here. Now, this is actually one of my 6:03:34 favorite books that I've ever read. I 6:03:35 actually reference it in a lot of 6:03:37 content, blue ocean theory. This is that 6:03:38 you want to reposition your product into 6:03:41 a market that differentiates you from 6:03:42 everyone else so that you're not 6:03:44 competing on pricing. You're not even 6:03:45 getting compared and you're seen as a 6:03:46 completely novel product or service 6:03:48 within the market. Red ocean means that 6:03:50 you're selling the same stuff as 6:03:51 everyone else and then you're just 6:03:52 getting priced down to zero. These are 6:03:54 terrible businesses. These are 6:03:55 incredible businesses. The issue with 6:03:56 blue oceans is they don't really exist 6:03:59 because if you're going truly blue, 6:04:01 you're selling something that's never 6:04:02 been sold before and that's kind of 6:04:04 risky because there's no actual market 6:04:06 knowledge. You need to do tons of 6:04:07 education and there's no proven product 6:04:10 market fit either. So yeah, you could 6:04:11 reposition into something that you think 6:04:12 people might want, but they might not 6:04:14 even want it. And so you realistically 6:04:15 want to sit like somewhere in the middle 6:04:17 here where you're repositioning it into 6:04:19 something novel, but it still has proven 6:04:21 product market fit. It's still proven 6:04:22 that it is going to work. It's just not 6:04:24 what everyone else is doing. it's 6:04:25 slightly adjacent. And so we do this a 6:04:27 lot in our own content in terms of 6:04:29 format testing. So we'll rotate in 6:04:31 different formats because we want to 6:04:32 test new things. We'll take ideas from 6:04:33 other industries, from other people. 6:04:35 We'll come up with our own ideas to try 6:04:37 to cut through and have novelty that 6:04:39 people haven't seen before. I would say 6:04:40 this YouTube video is actually an 6:04:42 example of that, which is that there 6:04:43 isn't many creative courses on YouTube 6:04:45 for meta ads that are 2 plus hours long. 6:04:48 So you want to be thinking through that 6:04:49 concept too in the hook. Now, once 6:04:50 again, it's not a requirement, but it 6:04:52 will help in standing out without having 6:04:55 some kind of gimmicky tactic to hold 6:04:57 people through the ad that's not 6:04:59 actually unique. The question I would be 6:05:00 asking yourself here is, has this hook 6:05:03 been done a million times? And if the 6:05:05 hook's been done a million times, 6:05:06 probably don't do it. The next is 6:05:07 specificity. So, is there a way, and 6:05:09 you're not going to be able to do this 6:05:10 in a lot of hooks, but in our specific 6:05:12 hook, is there a way we can use numbers, 6:05:15 maybe quantify names, or maybe even 6:05:17 quantify outcomes right out of the 6:05:19 gates, that's going to build authority 6:05:23 into the hook. Anytime you can start a 6:05:25 hook by name dropping someone that's 6:05:27 super famous, that's relevant to the 6:05:28 product in some capacity, it always does 6:05:30 better. If you can name drop some kind 6:05:32 of number, 67% of people got this 6:05:34 result, then once again, it's going to 6:05:36 build authority. If you can speak 6:05:38 straight to specific outcomes, it's 6:05:39 going to build authority and it's going 6:05:40 to be more specific to your target 6:05:42 demographic than you want to resonate 6:05:43 with. And then lastly, we have 6:05:44 credibility, which is fairly obvious. 6:05:45 Now, with credibility, you're usually 6:05:47 not going to be able to do this through 6:05:49 the copy or through the audio. It's 6:05:51 typically going to have to be done 6:05:52 through the visual. And so, you are 6:05:53 going to need either like someone famous 6:05:55 that pops up at the start. You're going 6:05:57 to need someone that is uh contextually 6:06:00 relevant. So, if you're selling 6:06:01 supplements, you have a doctor who pops 6:06:03 up. If you're selling running 6:06:04 supplements, it's obviously like a 6:06:06 runner that pops up. You want to be 6:06:07 thinking about how you can build 6:06:08 credibility. Now, you don't need 10 out 6:06:10 of 10 scores on all of these to be able 6:06:11 to have a good hook. You really only 6:06:13 need a couple of them done well. This is 6:06:14 how you should be thinking through hook 6:06:16 creation. Now, let me give you a bunch 6:06:17 of hooks that you can go and start 6:06:20 using. Number one is problem agitation. 6:06:22 So, you want to lead with the pain that 6:06:23 the viewer is already feeling and you 6:06:25 want to make them feel seen. An example 6:06:27 of this is, are you struggling with back 6:06:30 pain from desk work? This will only 6:06:32 resonate with problem aware audiences. 6:06:34 So you can get really good performance, 6:06:35 but it isn't going to be scalable above 6:06:37 that stage of awareness. Number two is a 6:06:39 contrarian truth. So you challenge a 6:06:41 common assumption. You create cognitive 6:06:43 friction that demands a resolution. An 6:06:45 example here is everything you've been 6:06:46 told about protein timing is wrong. And 6:06:48 this works across all levels of 6:06:50 awareness. And it works strong and 6:06:52 organic and paid. Number three is 6:06:53 specific proof. So you want to lead with 6:06:55 a measurable outcome, numbers, time 6:06:57 frames, specificity. An example of this 6:06:59 is I lost 12 kilos in 90 days without 6:07:02 giving up pasta. So you build 6:07:04 credibility throughout the hook. The 6:07:06 next one is a curiosity gap. We spoke 6:07:08 about this before which is that you 6:07:10 create an information gap that can only 6:07:12 be closed by watching. An example is 6:07:14 what the top 1% of brands know about 6:07:16 meta ads that you don't. And a funny 6:07:17 side note is that my average view time 6:07:19 when I use that hook or some variation 6:07:21 of that hook is over 45 seconds. And so 6:07:24 that hook actually works incredibly well 6:07:26 because it creates an information gap in 6:07:28 my own organic content. Truth bomb. 6:07:29 There's actually one specific client 6:07:30 where this is done unbelievably well 6:07:32 for. So that's why I threw it in here 6:07:34 because I don't see many people doing 6:07:35 it, which is that you lead with an 6:07:36 uncomfortable truth about the product's 6:07:38 price, ingredients, or industry. The 6:07:40 honesty itself becomes the hook. And so 6:07:42 the hook in this instance is something 6:07:43 along the lines of this product costs 6:07:46 $120. Yeah, it's super expensive, but 6:07:49 it's our best seller and everyone buys 6:07:51 it. Here's why. It ends up working 6:07:53 exceptionally well for premium or higher 6:07:55 average order value products where the 6:07:56 price ends up being the main objection 6:07:58 that people have. So you confront it way 6:08:00 up front and it actually works really 6:08:02 well at cutting through on cold 6:08:03 audiences cuz even if I've never seen 6:08:05 the product before, if someone's opening 6:08:06 with this thing is expensive, I'm like, 6:08:08 "Okay, what is that thing? Teach me 6:08:10 more." And then it can go into a very 6:08:11 very educational piece. Obviously that's 6:08:13 only going to be relevant to very 6:08:16 specific industries. The next is 6:08:18 psychological confrontation. And so this 6:08:20 is being direct and patent interrupting 6:08:22 with a statement that feels like a 6:08:24 friend calling you out. An example here 6:08:26 would be, you know, that drawer full of 6:08:28 products you never use, that's the 6:08:30 problem. Or stop buying things that make 6:08:32 you feel guilty for not using them. It's 6:08:33 not clickbait, but it's playing on an 6:08:35 emotional pain point, and so the viewer 6:08:37 ends up feeling personally addressed, 6:08:39 which ends up acting as a scroll 6:08:41 stopper. Once again, all of these things 6:08:42 also apply into organic, which is where 6:08:44 I think getting good at organic content 6:08:46 is actually a very critical skill for 6:08:48 getting good at creative strategy 6:08:49 because it just translates into paid. 6:08:50 We'll keep going here, which is you can 6:08:53 also do sensory I'm going to call this a 6:08:56 texture hook as well. This does 6:08:57 unbelievably well in fashion right now. 6:09:00 This is just an ASMR hook. So, you don't 6:09:01 need any words. 2 to 3 seconds close-up 6:09:04 of the fabric being handled, a product 6:09:05 being squeezed, liquid being poured, 6:09:08 package being opened, something ASMR 6:09:10 related. These do incredibly well. Uh we 6:09:12 have a lot of ads right now that are 6:09:13 high performers. On this hook, you've 6:09:15 then got a founders letter. You can go 6:09:17 and apply all of these into most brands. 6:09:19 So you want to open with the founder and 6:09:21 their origin story. So we started this 6:09:23 because dot dot dot or here is why I 6:09:25 left my career to go and build this. The 6:09:27 more risk you can build within the hook 6:09:29 as well, the better. And so you want to 6:09:30 build tension. And this is once again a 6:09:32 big organic strategy that you just want 6:09:34 to translate into paid, which is that 6:09:36 you want to have high stakes within the 6:09:38 hook. if you're going to have a founders 6:09:39 letter style hook so that people have a 6:09:42 reason to continue watching. I 6:09:44 reorggaged my house and put $300,000 6:09:46 down to be able to order my first order 6:09:49 of this product and here's what's 6:09:51 happened since. Like that hook I could 6:09:53 almost guarantee would do well. Now 6:09:55 obviously that has to be a real 6:09:56 circumstance which is why that hook does 6:09:58 well, right? It's things that you can't 6:10:00 fake. If you actually did put so much 6:10:02 risk on the line, it's a great angle to 6:10:05 be able to talk about that risk. The 6:10:06 last one here is social proof. So you 6:10:09 want to lead with some kind of 6:10:10 credibility marker. 90% of customers saw 6:10:12 results in 30 days. Now this is nine 6:10:15 different hook types that you can go and 6:10:16 take. You can probably apply at least 6:10:18 seven of these into your particular 6:10:19 brand right now. There are infinite hook 6:10:21 strategies, right? I could write hooks 6:10:23 here all day long. I could fill the 6:10:24 whole board with 50 different hook 6:10:25 types. You could go to Claude or JBT 6:10:27 right now and say, "Give me all the 6:10:28 different hook types that I can run." 6:10:30 The idea is that you want to take hook 6:10:32 strategies and you want to translate it 6:10:34 into your brand, but the scripting is 6:10:36 going to be critical. And where you're 6:10:37 going to really see uh a standout from 6:10:39 competitors is over here in this novelty 6:10:42 piece on whitespace. So you can become 6:10:43 an 8 figure brand by using these hook 6:10:45 strategies and having relatively good 6:10:47 scores over here. But if you want an ad 6:10:49 that's going to hold half a million 6:10:50 dollars in spend, you do probably want 6:10:52 to generate something novel at the 6:10:54 beginning of the creative that's going 6:10:56 to stand out from competitors. And then 6:10:57 what's going to happen is all your 6:10:58 competitors are just going to go and 6:10:59 copy you and so you will set the trend. 6:11:01 Now I've spoken about the hook a lot but 6:11:03 an an important subcomponent of the hook 6:11:05 is the bridge which is where you move 6:11:07 between specific components within the 6:11:09 video. Now the two big bridges that 6:11:11 often occur is when you go from hook to 6:11:13 body and then when you go from body to 6:11:15 CTA. These two bridges often don't end 6:11:18 up being done well. And so what ends up 6:11:20 happening is if you think about product 6:11:22 awareness on the y- axis and then you 6:11:25 think about time going through the video 6:11:26 asset on the x- axis, what ends up 6:11:28 happening is people will hook you in and 6:11:30 you'll become aware that it's an ad and 6:11:32 then suddenly as you bridge to the body, 6:11:34 some people will just instantly 6:11:36 introduce the product. So they'll hook 6:11:37 you with something like, are you trying 6:11:39 to train for a marathon and you want 6:11:41 better recovery? Well, creatine from 6:11:44 Blue Sense Digital is your solution. 6:11:46 That ad won't do well. Hook was good. 6:11:48 bridge ruin the whole thing. The whole 6:11:50 body could be good. Like the body 6:11:52 scripting could be perfect, hook 6:11:53 scripting could be perfect, call to 6:11:54 action could be great, but the bridge 6:11:56 ruins the whole asset. And so you don't 6:11:59 want your asset to look like this where 6:12:01 there's this massive step change in 6:12:03 product awareness as a user goes through 6:12:05 the creative. Instead, you want to blend 6:12:07 these out so that people are slowly 6:12:09 becoming aware that they're being solved 6:12:11 to. And so instead, you're looking like 6:12:13 this. And so I don't really even need to 6:12:15 give you that much information on how to 6:12:17 do this. To me, it's pretty intuitive, 6:12:19 which is just blend in the introduction 6:12:21 of the product. Don't just suddenly 6:12:23 introduce it. Or else people go quickly 6:12:25 from, oh, I was watching something that 6:12:26 was educational or that was teaching me 6:12:28 to, oh, I'm actually being sold a 6:12:30 product. And people don't intuitively 6:12:31 like that feeling. This also, this jump 6:12:34 in bridge will typically ruin the flow 6:12:36 of scripting in most very good top 6:12:39 offunnel unaware ads. And so when it 6:12:41 comes to really good topfunnel unaware 6:12:43 ads, what they do is they move people 6:12:45 through the stages of awareness as you 6:12:48 progress through the ad. This is 6:12:49 actually something that VS Cells video 6:12:51 salelet letter ads do unbelievably well 6:12:53 because they get the capacity to teach 6:12:55 through the asset. And so you start with 6:12:57 unaware. So you're hooking people in 6:12:59 with something that they might be 6:13:00 interested in, usually some kind of 6:13:02 curiosity gap. Then you're making them 6:13:04 problem aware. Okay, now I understand 6:13:05 that there's a problem with drinking 6:13:07 water. If I just drink water every day, 6:13:09 not only does it not have electrolytes, 6:13:12 which apparently I need. I didn't even 6:13:13 realize this was a problem, but also tap 6:13:15 water is incredibly bad for me. Okay, 6:13:17 that's interesting. What's the solution? 6:13:18 Oh, the solution is that I need this 6:13:20 amount of salt in this kind of ratio, 6:13:22 and I need to stop using tap water. 6:13:24 Okay. Oh, there's actually a product 6:13:26 that solves this problem. It's canned 6:13:28 mineral water with electrolytes in it. 6:13:30 Oh, okay. I was unaware of anything. I 6:13:32 became problem aware. I became solution 6:13:34 aware. I'm now product aware. Okay, this 6:13:36 is going to fix all of these problems 6:13:38 that I didn't realize I had. And now I 6:13:40 move down into the aware stage of 6:13:41 awareness where maybe I convert on a 6:13:43 retargeting ad. Now, this is generally 6:13:44 how you want to be thinking through 6:13:45 scripting on really good topunnel ads. 6:13:47 But what most people do is that they'll 6:13:49 start really unaware. They'll then go 6:13:51 problem aware and then like in here 6:13:53 they'll just introduce the product. 6:13:54 You're like, what are we doing? We 6:13:55 haven't even moved them to the next 6:13:56 stage of awareness. We can't introduce 6:13:58 the product yet. And people do this cuz 6:14:00 they think in retention curves. And so 6:14:02 they go, well, if we look at the 6:14:03 retention curve on an ad, which is how 6:14:05 many people are still watching on 6:14:07 average as we move through a creative, 6:14:08 the retention curve will look like this. 6:14:10 And so intuitively, you go, well, 80% of 6:14:14 people are going to drop off by the 90 6:14:16 secondond mark. So we need to introduce 6:14:17 our product before the 90cond mark cuz 6:14:20 we don't want most of the users to not 6:14:22 know they're being sold our product. 6:14:23 It's the wrong way to think through it, 6:14:25 okay? You actually want to introduce 6:14:26 your product as late as possible in the 6:14:28 creative because you want people to be 6:14:30 problem aware and s solution aware and 6:14:32 really preframe and engaged ready to buy 6:14:35 once the product is actually introduced. 6:14:37 You don't want to introduce the product 6:14:38 too early before they're actually aware 6:14:40 of what the problem and solution is. And 6:14:42 so you often end up getting way better 6:14:43 conversion rates out of an asset by 6:14:45 moving the product introduction and call 6:14:47 to action way later because these people 6:14:50 are now way more qualified to actually 6:14:52 buy and are way more higher intent. And 6:14:54 so, yes, you can introduce it sooner, 6:14:56 but they're not high intent enough that 6:14:58 you will actually have worse conversion 6:15:00 rates. And even though it's more total 6:15:01 people, you will ultimately get worse 6:15:04 total purchase volume out of the asset. 6:15:06 And so, that is something to be really 6:15:07 careful of when you're thinking about 6:15:08 bridging. Now, to wrap this section up, 6:15:10 one final question that I get a lot is, 6:15:13 does hook testing still work post 6:15:16 Andromeda? Because isn't the whole point 6:15:17 in Andromeda that we need diversity? And 6:15:20 if I have a 92 Facebook ad, me just 6:15:23 changing the first 3 seconds at the 6:15:25 start, that's not diversity, right? 6:15:28 That's the same ad. We just changed the 6:15:29 first 3 seconds. Meta will still 6:15:32 recognize a different hook as a 6:15:34 different creative. It will get its own 6:15:36 creative ID. It will not get pulled 6:15:39 together with this asset. It will go and 6:15:40 reach a novel unique audience. Now 6:15:43 obviously there is going to be some 6:15:44 degree of overlap because 87 seconds of 6:15:47 this asset is going to be talking to the 6:15:49 same person with the same problem with 6:15:51 the same offer. So if we actually look 6:15:52 at the concept framework it is the same 6:15:55 concept. It's just the hook change. So 6:15:57 the question becomes is hook changes 6:15:58 worth it? Yes. Why is it worth it? 6:16:01 Because it is so low effort. Rotating in 6:16:03 new hooks is almost free. Okay. When you 6:16:06 record the ad, you just record the hooks 6:16:08 six times and it takes an extra few 6:16:10 minutes and now you've got six different 6:16:12 hook variations. Or you can take 6:16:13 existing high performers and you throw 6:16:15 new hooks on them. We have had ads that 6:16:17 have spent $50, $100,000 fatigued. Then 6:16:20 we just rotate 50 new hooks on and we 6:16:23 get another $100,000 to spend at the 6:16:25 same efficiency. And those hooks took 6:16:27 like one hour to shoot. And so you can 6:16:29 very much so keep winners alive through 6:16:32 changing hooks. You can also take ads 6:16:34 that didn't work and then suddenly make 6:16:35 them work by fixing the hook if it had a 6:16:38 really bad hook rate. So yes, absolutely 6:16:40 hook testing still works post Andromeda. 6:16:42 You should be doing it. It is still the 6:16:43 highest leverage point in videos, 6:16:46 organic ads, YouTube, everything. All 6:16:49 right. So when it comes to formats, 6:16:50 there's five questions that you need to 6:16:52 be asking yourself when you're thinking 6:16:53 on a format. Number one, what is the 6:16:54 concept? We need to define the concept 6:16:56 first before we even think about 6:16:57 formats. You want to go out build your 6:16:59 persona angle offer then come back to 6:17:01 thinking about the format. You don't 6:17:03 start with the format and work 6:17:04 backwards. This is the last step. 6:17:06 Secondly, does this concept require 6:17:09 education? Now, if it requires 6:17:10 education, you likely need to go into 6:17:12 some longer form asset that's going to 6:17:14 educate. Number three, can the value 6:17:16 prop be communicated in a single frame? 6:17:18 This is obviously going to lean yourself 6:17:20 towards images, carousels, something 6:17:22 that's easier to produce, can obviously 6:17:24 still be in a video, but you do have an 6:17:26 advantage with images that they're very 6:17:28 easy and cheap to produce and you can 6:17:30 produce a lot of them. So, anytime you 6:17:32 can produce statics over videos, you 6:17:34 generally want to do it. Now, there's 6:17:36 two different approaches to statics if 6:17:37 we just leave these out for the moment, 6:17:38 which is number one, you can use statics 6:17:41 to quickly test different angles and 6:17:43 different concepts, get fast feedback, 6:17:45 and then develop that into a video 6:17:47 format where you can go a lot deeper and 6:17:48 create a more scalable asset. And then 6:17:50 number two is that images are very cheap 6:17:53 to make, and so they often end up 6:17:55 outweighing videos in the cost of 6:17:57 production. So if you look at the 6:17:58 average spend of an image versus a video 6:18:00 in your account, it might actually make 6:18:01 a lot of sense to just go way harder on 6:18:03 images because the cost of production is 6:18:05 super low and the average revenue 6:18:06 generation is decent in proportion. And 6:18:08 so this is what is called cost of 6:18:11 production against the average revenue 6:18:14 per ad. This is something that we 6:18:16 calculate for everyone during the order 6:18:18 process because we want to understand 6:18:19 what is the average revenue per ad that 6:18:21 you put into the account. For most 6:18:23 people, it's about like $2,000ish. And 6:18:26 then what is your cost of production 6:18:28 against this delineated down to 6:18:29 contribution margin? And so we minus off 6:18:32 gross profit. We minus off the cost of 6:18:34 serving the ad. We might find out that 6:18:36 average contribution margin per ad unit 6:18:38 is, let's say, $350. How much is it 6:18:41 costing us to make an ad? It might be 6:18:43 costing us $400. Okay, this equation 6:18:46 doesn't work. We're currently paying 6:18:47 more to produce an ad than we 6:18:49 contribution margin out of each ad. We 6:18:51 can then go and delineate this down to 6:18:53 images specifically. We might found out 6:18:55 images cost us about $50, but they're 6:18:58 producing us $300. Videos are costing us 6:19:02 $500 and we're only getting $400 here. 6:19:04 And so we actually find out even though 6:19:06 videos do perform better, the actual 6:19:08 cost of production versus expected 6:19:10 contribution margin is worse than image 6:19:12 assets. And therefore, we should 6:19:13 actually just double down on images. And 6:19:14 I know people doing uh $200,000 a day in 6:19:18 revenue on just AI images. And so 6:19:20 there's absolutely no reason as to why 6:19:22 you can't do hundreds of thousands of 6:19:23 dollars a day on images only. And if you 6:19:26 can't, it just ends up being a skill 6:19:27 issue. Number four is, does the product 6:19:29 need a demo? This is just going to be a 6:19:31 general question that applies to the 6:19:32 business as a whole, which is how much 6:19:34 education do we need to force people 6:19:36 through? Because obviously you're going 6:19:38 to be relatively restricted with images 6:19:40 then unless you get really fancy and you 6:19:42 have a good designer. And then lastly, 6:19:43 which ties into both of these points, is 6:19:45 what stage of awareness are we sitting 6:19:46 in? Okay, if we're sitting in a very 6:19:48 high stage of awareness, we need a 6:19:50 creative that is going to be a longer 6:19:52 format that can educate, typically a 6:19:54 video. If we are sitting at a very low 6:19:56 stage of awareness, like the product 6:19:57 aware or aware stages, uh we can just go 6:20:00 for DPAs and images and get away with 6:20:02 very bottom ofunnel ads. I want to 6:20:04 reiterate that the belief that image ads 6:20:06 are bottom of funnel and video ads are 6:20:08 top of funnel is fundamentally a skill 6:20:11 issue. Once again, I have seen multiple 6:20:13 ad accounts. I've talked to multiple 6:20:14 people that do hundreds of thousands of 6:20:16 dollars per day in revenue off just 6:20:19 static. And so if you're saying, "Oh, 6:20:21 statics don't work. They're bottom of 6:20:22 funnel assets." It's because, yeah, the 6:20:23 way that you design them and the way 6:20:25 that you approach them and concept them 6:20:26 out is bottom of funnel. But you can 6:20:28 make top of funnel statics. We're now 6:20:30 going to cut out to three ad breakdowns 6:20:32 that I have done on static images. We're 6:20:35 going to go through an organic tweet 6:20:36 format, which is an IM8 health ad. We're 6:20:39 going to go through a creatine gummies 6:20:40 ad. And then we're going to go through a 6:20:42 GLP1 hair loss ad. And you will see 6:20:44 three statics that are holding tons of 6:20:46 spend on top of funnel. This is an 6:20:48 incredible static ad. 4.7,000 likes down 6:20:52 here. It's by IM8. And let's break down 6:20:54 exactly why it's so good. Number one, 6:20:55 it's a new format. They're actually 6:20:57 using a tweet here and they're reposting 6:20:59 it so it feels organic to the platform, 6:21:01 but it's actually an ad. Now, further to 6:21:03 this, they're reinforcing it by using 6:21:06 not a partnership ad. So they're not 6:21:07 collaborating with IMA and Dr. James, 6:21:10 but it's just Dr. James. So, they're 6:21:12 running it as effectively a whitelisting 6:21:14 ad through this profile. So, you don't 6:21:16 know that this is an ad. In fact, you 6:21:18 would look at this for a very long time, 6:21:20 and it's only until you start getting 6:21:21 through the copy that sits under the 6:21:23 creative over here that you actually 6:21:25 realize that you're getting sold 6:21:26 something. So, what's so good about the 6:21:27 static over here other than it being a 6:21:29 common format and that you not really 6:21:30 knowing that it's an ad, so it feels 6:21:31 organic? Well, symptoms of vitamin 6:21:33 deficiencies. So right away we're 6:21:35 problem agitating into okay here's some 6:21:38 education around how vitamin 6:21:40 deficiencies could be impacting you then 6:21:42 this is an unbelievably wide set of 6:21:46 symptoms. So if you have dry eyes, if 6:21:48 you have poor vision, if you have low 6:21:49 energy, if you have bad skin, if you 6:21:51 have cracks on your mouth, if you have 6:21:52 fatigue, if you have bleeding gums, if 6:21:54 you have painful muscles, if you have 6:21:56 excessive bruising, all of these things, 6:21:58 very, very wide, total addressable 6:22:00 market, well then it's an issue with a 6:22:02 vitamin deficiency, and it tells you 6:22:04 exactly what the vitamin deficiency is. 6:22:06 So you get education around, okay, 6:22:07 here's my problem, here's the solution. 6:22:10 Then you start probably writing a copy 6:22:12 and you go the best multivitamin equals 6:22:14 IMA daily ultimate 10% off. Here's the 6:22:17 thing. Then you get more educational 6:22:19 pieces here about how the product 6:22:21 actually works. And obviously because 6:22:22 this is an ad, a big call to action will 6:22:24 appear at the bottom here that it will 6:22:26 allow you to click straight off the 6:22:27 site. So a problem agitates to a very 6:22:29 wide target demographic. It then 6:22:32 educates about exactly what the solution 6:22:34 is in terms of vitamin to solve your 6:22:36 particular problem. It does it in a 6:22:37 format that's super organic. It then 6:22:40 maintains that organic feeling by not 6:22:42 having the brand and company name 6:22:44 attached with the post. And then this is 6:22:46 obviously driving off to cold audiences, 6:22:47 probably driving off to this guy's 6:22:49 audience, which is even better because 6:22:51 we're obviously piggybacking off his 6:22:52 organic. This is an avitorial static ad. 6:22:55 These are really good. They perform 6:22:56 incredibly well. They only perform well 6:22:58 though if you go to an educationbased 6:23:00 landing page or you use the primary text 6:23:02 to educate heavily. So, let's break it 6:23:04 down. The best hair growth products of 6:23:06 2025 and the ones to avoid. Here's what 6:23:08 the big brands are not telling you. And 6:23:10 then we have a bunch of different 6:23:12 products here. So, we know or we at 6:23:13 least assume, okay, this is some kind of 6:23:15 blog. This is some kind of article 6:23:16 that's going to educate us around what 6:23:18 hair growth products we should actually 6:23:20 be buying. Now, even better, it's 6:23:21 running through a doctor's page. Okay, 6:23:23 so there's a bit of authority here. As a 6:23:25 dermatologist, I set out on a mission to 6:23:27 find the most effective hair growth 6:23:28 products. Not very long copy, so they're 6:23:30 not selling us on anything here. Okay, 6:23:31 so we want to find out more. So, what do 6:23:33 we do? We go and we go and click learn 6:23:35 more. Now, from here, we end up on a 6:23:36 website called Dermatologist Reviews. 6:23:39 Now, this website is owned by the brand 6:23:41 that's trying to sell you on something. 6:23:43 This is a really good funnel. So, hi, 6:23:45 who is the person? Why do they have 6:23:46 authority? This is the blog posted. I 6:23:48 set out on a mission. And here are just 6:23:50 the products ranked. And then it goes 6:23:52 into all of the reasons as to why. And 6:23:55 it goes into selling that product. And 6:23:56 then it does go into the other ones, but 6:23:58 it goes into why, yeah, I probably 6:23:59 wouldn't choose these. They're not the 6:24:01 best. Now, why is this a giveaway that 6:24:03 this is actually a brand pushing? Well, 6:24:05 because if we look at the rankings here, 6:24:06 you can't click on any of these, right? 6:24:08 But you can click on this one. And if 6:24:10 you click on this one, goes to a custom 6:24:11 landing page that then takes you through 6:24:13 a quizunnel. You go through a quizunnel 6:24:15 and then you get pushed through to the 6:24:16 conversion. So, this is a really good 6:24:18 funnel that they're running here. I 6:24:20 really like it. And almost to make sure 6:24:21 they're a little bit compliant, I 6:24:23 imagine. As you scroll down, this takes 6:24:24 you to the same funnel, but as we 6:24:26 continue going down, look at all the 6:24:27 green ticks here. So many green ticks, 6:24:29 one red. Only available online. And then 6:24:31 him and hers. Oh, widely available 6:24:33 online. Oh, that's the best thing about 6:24:35 it. All of these negatives, but you can 6:24:37 still click here and it takes you to an 6:24:39 Amazon affiliate link. And so, they're 6:24:40 like, h, if people are not clicking on 6:24:42 us, then we may as well still redirect 6:24:44 them somewhere. And then obviously, you 6:24:45 can see they put the effort into 6:24:46 reviewing all of these and all the 6:24:48 reasons as to why they wouldn't buy 6:24:49 them. This is a super effective funnel. 6:24:51 I imagine this is doing quite well for 6:24:52 them. If your target demographic is 50 6:24:54 plus, you can serve on Facebook feeds. 6:24:56 This ad type crushes. It's static image 6:24:58 ads with a long form copy. And the 6:25:00 reason why they do so well is they go 6:25:01 through a massive story that speaks to 6:25:04 an unaware audience and slowly pulls 6:25:06 them into being problem aware, pulls 6:25:08 them into being solution aware, then 6:25:09 pulls them into being product aware, and 6:25:11 then finally they're all the way at the 6:25:13 stage in which they can convert. These 6:25:15 ads will crush on cold audiences, but it 6:25:18 takes an enormous amount of skill to 6:25:20 actually be able to write this copy 6:25:21 properly. Like yes, you can be AI 6:25:23 assisted, but if you don't know 6:25:24 fundamentals of how to emotionally 6:25:26 appeal to a specific target demographic, 6:25:28 then it's not going to work. And this is 6:25:29 ultimately where getting good at 6:25:30 copywriting really matters. If you can 6:25:32 get good at copywriting and you can get 6:25:33 good at pulling people through the 6:25:35 stages of customer awareness, then that 6:25:37 is ultimately how you can create these 6:25:39 static image ads with long form copy 6:25:41 that are going to absolutely crush. Now, 6:25:43 in addition to that, to keep it 6:25:44 contextual to 2026 and the time in which 6:25:47 this video is being made, a type of 6:25:48 video asset that does incredibly well 6:25:50 right now on older target demographics 6:25:52 is a VSSL, a video sales letter. The 6:25:54 reason why these types of ads do so well 6:25:57 is because they pull people through all 6:25:58 of the stages of awareness and they're 6:26:00 highly educational. You can also really 6:26:02 leverage AI within them. We make a lot 6:26:04 of just AI VSSLs end to end and they do 6:26:07 quite well. And then some of them do 6:26:08 unbelievably well and hold hundreds of 6:26:10 thousands of dollars in ad spend. So 6:26:12 what we're going to do now is I'm going 6:26:13 to pop up a few examples of VSSLs and me 6:26:16 breaking them down so you can see an 6:26:18 example of what this ad type looks like. 6:26:20 >> You have IBS, leaky gut, or bloating? 6:26:22 You're probably missing this one thing. 6:26:25 You're This is a really good VSSL. First 6:26:27 off, it starts with a persona call out. 6:26:29 If you have these issues, keep watching. 6:26:30 >> Doctor probably told you to just manage 6:26:32 it with a course of pills. That's 6:26:34 because then it's going into objection 6:26:36 handling. The common way to fix it. 6:26:38 They're treating the symptom, not the 6:26:39 cause. But here, then a problem agitates 6:26:42 and opens a curiosity gap. Okay. I 6:26:44 thought it was actually going to cause 6:26:46 us to fix the problem, but this ad is 6:26:48 saying it's not. All right, I'm going to 6:26:49 keep watching. 6:26:50 >> Here's what they're not telling you. 6:26:51 Your gut isn't broken, it's starving. 6:26:53 Your gut is an eco. 6:26:54 >> I don't love that because it's an AI 6:26:56 script. So that's called contrast 6:26:58 negation, which is where you say it's 6:26:59 not X, it's Y. You see a lot of this in 6:27:01 a lot of scripting these days. I 6:27:03 probably would take that out. 6:27:04 >> An ecosystem of trillions of bacteria. 6:27:06 When the good bacteria die and bad 6:27:08 bacteria take over, everything 6:27:10 collapses. That's when you get 6:27:11 inflammation, a leaky gut, and your body 6:27:14 stops absorbing nutrients. This really 6:27:16 good educational piece. 6:27:17 >> This is dispiosis, and it's not rare. 6:27:19 40% of people have it. Most of them 6:27:21 don't even know. You might be among 6:27:23 them. But here's the problem with really 6:27:25 good way to enforce that. No, you do 6:27:28 have this problem. Because when you 6:27:29 prescribe a problem to someone, 6:27:32 particularly when it's like a disease or 6:27:34 it's a physical medical issue, the 6:27:35 immediate reaction is, well, I probably 6:27:37 don't have this. I imagine this is rare. 6:27:39 So, instantly increasing the tamb by 40% 6:27:41 of people actually have this problem is 6:27:43 a really good way to keep people 6:27:44 watching to go, okay, maybe I do have 6:27:45 the problem. Tell me a little bit more 6:27:46 about it. With every gut supplement 6:27:48 you've tried. They just throw probiotics 6:27:50 at it. That's like sending a single 6:27:52 soldier into a war zone. One probiotic 6:27:54 can't survive in a toxic gut. It dies. 6:27:57 It doesn't colonize. It doesn't help. 6:27:59 IM8 Daily Ultimate Essentials doesn't 6:28:02 just add probiotics. First, digestive 6:28:04 enzymes clear. I don't love the bridge. 6:28:06 If the ad team is watching this, I would 6:28:08 extend for another 15 seconds and 6:28:10 educate around how the IM8 ingredients 6:28:13 actually fix the issue and then 6:28:15 introduce IM8 as the solution that has 6:28:17 those ingredients and undigested food. 6:28:19 Second, prebiotics create a safe 6:28:21 environment for good bacteria. Third, 10 6:28:24 billion CFU of probiotics flood in and 6:28:26 colonize. Fourth, postbiotics strengthen 6:28:28 your gut lining and stop the leaks. So 6:28:31 the 10 to 15 seconds would be saying uh 6:28:33 the way that you get probiotics to work 6:28:35 is that you have prebiotics and 6:28:36 postbiotics around it and that's 6:28:38 actually what's creates the system to 6:28:39 work. Fortunately, IM8 has put all of 6:28:42 those three solutions in one sachet. So 6:28:43 you can get them all at once and it can 6:28:45 fix uh the issues that we agitated at 6:28:47 the start. It's a complete military 6:28:49 operation, not a single soldier. That's 6:28:51 an AI script once again. Contrast 6:28:52 negation. I try to avoid that. And it 6:28:54 works in a 12week clinical trial. 85% of 6:28:57 And then it goes into social proofing. 6:28:59 It keeps going on. Overall, really good 6:29:00 ad. Just some small points of 6:29:01 improvement. This ad has 25 million 6:29:04 views. It's been running for 3 to 4 6:29:06 years now. It's absolutely crushing. And 6:29:08 I'll show you what it does so well. It 6:29:09 takes a really common format of VSSLs 6:29:12 and then it adds objection handling UGC 6:29:15 to the front. So, let's click play and 6:29:16 start to watch. 6:29:17 >> I want to talk to you about something 6:29:18 that's been making waves on social media 6:29:20 lately. You probably seen this guy all 6:29:22 over your feed claiming that cutting 6:29:24 carbs is so they're using the split 6:29:26 screen for engagement and to add more 6:29:27 context to the video. Is it necessary 6:29:29 for weight loss? No need for long hours 6:29:31 on the treadmill either. Now, I know it 6:29:33 sounds too good to be true, but let me 6:29:34 tell you, there's more to this than 6:29:36 meets the eye. I decided when you're 6:29:37 watching this, you think that this guy 6:29:38 is obviously going to disagree with the 6:29:40 cutting carbs because he's coming from a 6:29:42 news authorative perspective. And it 6:29:44 kind of feels like a call out video. 6:29:46 Like, this guy is talking about how you 6:29:47 don't need to eat carbs. It's crazy. 6:29:48 Like, what is he talking about? I 6:29:50 decided to delve into this topic and do 6:29:52 some research. Turns out this guy is a 6:29:53 celebrity trainer who's gained a massive 6:29:55 following for challenging mainstream 6:29:57 advice. He claims that what we've been 6:29:58 told about weight loss is wrong. 6:30:00 Completely wrong. And you know what 6:30:01 makes a lot of sense when you listen to 6:30:03 what he has to say. So that's the first 6:30:05 time in which he validates the person 6:30:07 he's talking about. But everything 6:30:09 through that first 28 seconds was 6:30:10 contrarian hooks. Contrarian takes over 6:30:13 and over again. Celebrities don't cut 6:30:15 carbs. That ain't how you lose weight. I 6:30:18 don't do that. None of my clients do 6:30:20 that. Nobody that has ever asked me how 6:30:22 to lose weight have I ever told them to 6:30:24 do that because it simply just doesn't 6:30:26 work. For anyone who doesn't know their 6:30:28 metabolic type, listen up. The fastest 6:30:30 way to burn fat and get shredded. It's 6:30:33 not keto. It's not paleo. It's not 6:30:35 carnivore. It's not vegan. 6:30:37 >> Note that this creative is just going 6:30:40 very very hard on contrarian takes. 6:30:43 That's pretty much the entire strategy 6:30:45 of this whole ad is like how do we put 6:30:46 as many contrarian takes in here as 6:30:48 possible to continue to string people 6:30:50 along as we kind of educate and we kind 6:30:52 of hint towards that we have our own 6:30:53 system and then eventually we obviously 6:30:55 bridge into that system. So they're 6:30:56 creating a curiosity gap through this 6:30:58 whole thing. Okay, if no one needs to 6:31:00 cut carbs then what is the actual 6:31:02 solution? And it's definitely not super 6:31:05 intense exercises like this because 6:31:07 listen, if you're trying to get in 6:31:08 shape, and I don't care what it's for, a 6:31:10 wedding, vacation, summer, don't care. 6:31:12 If you're trying to get in shape fast, 6:31:14 you need to follow a plan for your body 6:31:16 type. And there is a very simple 6:31:18 breakdown of body types. There's three 6:31:20 body types. Uh you have the skinny 6:31:22 person who's trying to put on muscle. 6:31:24 Then you have someone who's kind of in 6:31:25 the middle, maybe has a little bit too 6:31:27 much fat. 6:31:27 >> Turn's going to go into the educational 6:31:29 piece. is effectively going to sell his 6:31:31 system and the way that he views weight 6:31:33 loss and body types and programming, 6:31:35 etc. This is such a good ad. In the last 6:31:37 5 days, it's got 2,000 likes. Let's 6:31:39 break down exactly why. 6:31:41 >> She hop on chess is opens with a 6:31:43 contrarian hook and obviously the 6:31:45 visuals pull you in. What I would say if 6:31:47 I was a caveman that wasn't aware of all 6:31:49 the way 6:31:50 >> builds contrast and then goes into an 6:31:52 educational pace that you can boost it 6:31:54 naturally. Being low tea can be caused 6:31:56 by many. switches over to a different 6:31:58 format which is the faceless format 6:32:00 which is better inducive to education 6:32:02 particularly on organic and we're trying 6:32:03 to make this ad feel organic. High 6:32:05 stress, vitamin deficiencies, not enough 6:32:08 fat in your diet, not enough good sleep, 6:32:10 microlastics, high estrogen, high 6:32:12 inflammation. It's become a serious 6:32:14 problem for young men who often feel 6:32:15 like there's only one solution after. So 6:32:18 notice they'll problem agitate and then 6:32:19 they'll educate. Problem agitate then 6:32:21 educate. So there's all the problems. 6:32:23 Here's the solution. This isn't the 6:32:25 solution for you. goes back into 6:32:27 problems. 6:32:27 >> After all, global testosterone levels 6:32:29 have dropped over 30% in the last 50 6:32:31 years. Young, fit, healthy. 6:32:33 >> So, there's the education and then 6:32:35 here's the problem agitation. 6:32:36 >> Guys, we'll have the testosterone of a 6:32:38 70-year-old that has never seen the 6:32:40 inside of a gym in their life. It's a 6:32:41 grim side effect of our modern society. 6:32:43 With low tests taking away the feeling 6:32:45 of being a real man, a consequence of 6:32:47 this is a wave of young men, even 6:32:49 teenagers, hopping on TRT out of 6:32:52 desperation and because they don't want 6:32:54 to fall behind. But injecting test is a 6:32:56 minefield of side effects. Rapid hair 6:32:58 loss, high estrogen causing gyno mood 6:33:00 swing, objection handling around the 6:33:02 core premise of the hook, which was 6:33:04 injecting test. And so now he's 6:33:05 explaining why that's not a good idea. 6:33:07 Infertility and acne, not to mention a 6:33:09 big ass needle injected multiple times a 6:33:11 week. In amongst this chaos, how do I, a 6:33:14 lifetime natural, have more testosterone 6:33:16 than most guys running cycles? Whilst 6:33:18 I'm doled with my sleep, training, 6:33:20 recovery, and diet, so are many guys who 6:33:22 are low tea. However, my secret weapon 6:33:24 is a Natty Plus Ultimate. And there you 6:33:26 go. There's the product. So, the 6:33:27 product's been introduced after about a 6:33:29 minute and a half into the creative, 6:33:31 which is why it's done so well. It's why 6:33:32 it's scaling so well. It's on an 6:33:33 influencer profile. People know this 6:33:35 influencer. It's likely prioritizing 6:33:37 getting served to his own audience. And 6:33:38 so, when people are watching this, 6:33:39 they're not knowing it's an ad. They're 6:33:40 thinking it's educational. And honestly, 6:33:42 it's a great educational piece. It goes 6:33:44 into everything you need to know about 6:33:46 testosterone and why you would want to 6:33:47 get behind and buy this product. And it 6:33:49 does it in a way that's very 6:33:50 entertaining. It's fast cut. It's 6:33:52 constantly changing formats. It's built 6:33:54 native for the platform, which is why 6:33:56 it's scaling so well. One important PSA 6:33:58 here is DPAs. Okay, DPAs, dynamic 6:34:02 product ads. Those are the carousel ads 6:34:04 that will dynamically pull in the 6:34:05 product that someone visited on the 6:34:07 website as long as your pixel is set up 6:34:08 correctly, pushing back in data on which 6:34:10 pages were viewed. DPAs cause a death 6:34:14 spiral in a lot of brands. I see this in 6:34:16 fashion particularly, uh, which is that 6:34:18 DPAs have really good attributed 6:34:20 rorowaz. the rorowaz number is very 6:34:21 high. Now the rorowaz actually isn't 6:34:24 very high once you delineate out 6:34:26 existing customers. So if you remove 6:34:28 existing customers it drops 6:34:30 precipitously. And then you also want to 6:34:32 delineate the attribution model down to 6:34:35 incremental attribution. And so you can 6:34:38 do this by clicking on the breakdown 6:34:41 button in the top actually sorry you 6:34:43 click on the columns button compare 6:34:45 attribution settings incremental 6:34:47 attribution apply and then you will get 6:34:50 a split out of what the incremental 6:34:51 rorowaz is on the ad. A quick lesson on 6:34:54 how incremental attribution works is 6:34:56 that every single day meta is targeting 6:34:58 let's say 10,000 people in this square. 6:35:00 It will hold out 10% of users and it 6:35:02 will look at these people that didn't 6:35:04 see your ads. How many of them 6:35:05 converted? Let's say 1% of them 6:35:07 converted. Over here, the people that do 6:35:09 see your ads, 2% of them convert. Okay, 6:35:12 great. Or we can do what's called in 6:35:13 data science is a difference in 6:35:15 difference calculation, which is simply 6:35:16 minusing the control versus the 6:35:18 treatment. This equals 1%. So your ads 6:35:21 are actually having a 1% impact, not a 6:35:23 2%. And so when we see a five rorowaz on 6:35:26 that DPA, let's say as an example, we 6:35:28 actually need to adjust that down by 50% 6:35:30 to 2.5. So that's what incremental 6:35:32 attribution is doing. That's what it's 6:35:33 going to show you on the rorowaz here. 6:35:35 and the rorowaz on your DPA is going to 6:35:37 be way lower than you think it is, 6:35:38 particularly on new audiences. I am 6:35:40 still yet to see an ad account where 6:35:41 this is a great scalable ad type on cold 6:35:43 audiences. Now, is it an ad type that 6:35:45 can work on cold audiences? For sure. 6:35:48 Can you spend $1,000 a day on it? For 6:35:50 sure. We have some ad accounts that do. 6:35:53 Can you spend $50,000 a day on it? 6:35:55 Absolutely not. Like, you just can't 6:35:57 scale DPAs that well on cold audiences. 6:35:59 Most people are way overspending because 6:36:01 the rorowaz number looks good. to just 6:36:02 be super careful cuz what ends up 6:36:04 happening there is because DPAs sit at 6:36:06 the bottom of the funnel down here is 6:36:08 that people start reallocating their 6:36:10 budgets from these topfunnel efforts 6:36:12 that have bad return on ad spends and 6:36:14 they just start moving budget down here. 6:36:16 And when you start moving budget down 6:36:18 here, you start losing the quantity of 6:36:20 users that are coming through from the 6:36:21 top of funnel and you start scaling 6:36:23 something that isn't scalable and the 6:36:25 whole funnel topples over and return on 6:36:26 ad spend declines and everything goes 6:36:28 poorly. So in fact when you're thinking 6:36:29 about funnel budget allocation you want 6:36:32 to be thinking about 70 to 90% of budget 6:36:34 going towards top of funnel you want 6:36:36 20%ish of budget going to middle of 6:36:39 funnel and then you want really like 5 6:36:41 to 10% of budget going to bottom of 6:36:42 funnel. There's actually a better way to 6:36:44 calculate bottom of funnel budgeting 6:36:45 which I'm going to show you later in the 6:36:47 video. But this is rough split on how it 6:36:49 should look. Now what's interesting 6:36:50 these days is that you used to do this 6:36:52 through campaign structure. So you would 6:36:54 have your top ofunnel ads and you would 6:36:55 just set your budget here. You'd have 6:36:56 your middle of funnel. You'd set your 6:36:58 budget here and then you'd have your 6:36:58 bottom bottom of funnel. Uh the issue 6:37:00 right now is there actually is a lack of 6:37:02 campaign structure. You shouldn't be 6:37:03 structuring like this. You should just 6:37:04 consolidate these down and you might be 6:37:06 able to have a bottom of funnel that 6:37:07 sits separate. And so because of that, 6:37:08 the actual budget allocation is coming 6:37:11 from the creative production volume. And 6:37:13 so you want 70 to 90% of the volume of 6:37:15 assets you make to be at the high stages 6:37:18 of awareness. And you want 20% of them 6:37:20 to be at the middle to bottom of funnel 6:37:22 stages of awareness. So this is more so 6:37:24 a percentage allocation of creative 6:37:26 production than it is a percentage 6:37:28 allocation of budgets across a campaign 6:37:30 structure. I would say one of the most 6:37:31 underutilized formats in 2026 is 6:37:34 partnership ads. Now I think people are 6:37:36 catching on to this actually and the 6:37:37 arbitrage is probably coming towards its 6:37:39 end. So by the time you watch this video 6:37:40 if you're way into the future this might 6:37:42 not be the case anymore. You might have 6:37:43 missed it. But partnership ads what they 6:37:45 are is they run through the creators 6:37:46 profile. So you'll have your profile 6:37:48 here. to your company name on the top of 6:37:50 the post and then you'll have and and 6:37:53 then the influencers profile right here. 6:37:55 Now, these days there's actually a 6:37:56 setting that allows them to change the 6:37:58 combinations and so sometimes you'll 6:38:00 just see the company, sometimes you'll 6:38:01 just see the influencer, sometimes 6:38:02 you'll see both and it will split test 6:38:04 and decide on whatever the best mix is. 6:38:05 But the reason why these ads do so well 6:38:07 is that the targeting uses the audience 6:38:11 on both handles. And so if this is an 6:38:13 influencer that's actually somewhat 6:38:14 contextually relevant to your target 6:38:16 demographic, which I hope it is, this ad 6:38:18 will go and reach their audience and it 6:38:19 will use their audience data to find new 6:38:21 people. And so you're effectively 6:38:23 leveraging the influencers's audience to 6:38:26 help in your targeting on cold, not even 6:38:28 just targeting a warm. And that's why 6:38:30 they end up doing so well. They also 6:38:31 feel super authentic. When you get an ad 6:38:33 from a uh influencer profile like this, 6:38:35 it doesn't feel as much so like an ad. 6:38:37 So it doesn't feel like you're getting 6:38:38 sold to. So when you're getting an 6:38:39 organic piece of content, it resonates 6:38:40 better. In some accounts that we're 6:38:42 working on, we have ad spend up to 40% 6:38:44 of total account spend running through 6:38:46 partnership or collaborator ads. And so 6:38:48 this is a really important thing that 6:38:49 you should have in your account. If you 6:38:51 are not running them, you were just at a 6:38:52 disadvantage to competitors. All right, 6:38:54 so going into the testing framework and 6:38:55 how this then translates into the ad 6:38:57 account. This is what the testing cycle 6:39:00 looks like, which is that you build a 6:39:02 hypothesis, which is you define what 6:39:04 you're actually testing. So when we're 6:39:05 introducing new creative, generally the 6:39:07 test is the concept and particularly one 6:39:09 of those four variables that we're 6:39:11 changing in the con. Then we want to 6:39:12 build a bunch of ads under that concept. 6:39:14 One big mistake that a lot of people 6:39:16 make is that they come up with a 6:39:17 concept, they build it all out, it looks 6:39:19 amazing, and then they make one ad or 6:39:20 they make two ads or they make three 6:39:22 ads. Like really, if you want to test a 6:39:23 concept properly, you want to at least 6:39:25 10 plus ads considering that most 6:39:27 people's hit rates in their account are 6:39:28 around about 5%. And so technically 6:39:30 speaking, if you have a new concept 6:39:32 coming in, you would need to do if you 6:39:33 had a 5% hit rate, you'd need to do at 6:39:35 least 20 ads to expect a winner. And so 6:39:37 thinking about your hit rate in the 6:39:39 account as a function of how much volume 6:39:40 you need to do to be able to validate 6:39:42 the hypothesis is critical. Then you 6:39:44 want to go and launch and I'm going to 6:39:46 show you with the exact structure that 6:39:47 you should launch into. You then want to 6:39:48 go and analyze particularly against the 6:39:50 hypothesis. So what did we expect? What 6:39:52 how much spend did we want to go 6:39:53 through? How many conversions did we 6:39:55 expect? What was the ROI? And then how 6:39:56 did that compare? If it worked, we want 6:39:58 to be looking at how we can employ 6:40:00 strategies to scale. And then we want to 6:40:01 go down and iterate. And this iteration 6:40:03 process gets faster and faster the 6:40:06 bigger your ad spend. So now when we 6:40:07 think about account structure and 6:40:08 introducing tests into the account, this 6:40:10 is where there becomes a lot of 6:40:12 flexibility. And this is where we prefer 6:40:14 not to actually make broad sweeping 6:40:16 account structure recommendations 6:40:18 because the fundamentals of account 6:40:20 structure is that the complexity of the 6:40:21 ad account should be a product of the 6:40:24 complexity of the business. And so if 6:40:25 you run a startup that's doing $10,000 a 6:40:28 month in revenue and you have one 6:40:30 product and you have one persona that 6:40:31 you're trying to crack, the account 6:40:33 structure should be unbelievably simple. 6:40:35 You don't need a 100 different 6:40:36 campaigns. On the other hand, if you're 6:40:38 an $800 million fashion retailer with 50 6:40:41 new SKs coming out per week and you have 6:40:44 all of these different business units 6:40:45 and regions within the business, well, 6:40:47 the ad account's probably going to be 6:40:48 pretty complex because it needs to match 6:40:50 that complexity in some degree so that 6:40:52 we can align the commercial goals of the 6:40:54 business with the structure that is 6:40:55 being run in the ad account. And so the 6:40:56 structure I'm about to give you is a 6:40:58 very broad sweeping this will work if 6:41:01 you're a small to midsize business, but 6:41:03 it's probably not exactly what you 6:41:05 should run right now or moving into the 6:41:07 future as it will likely evolve as the 6:41:09 business dynamics evolve. But to keep it 6:41:11 simple, you generally want a testing 6:41:14 campaign. Now, whether this is an or 6:41:16 a CBO is honestly going to come down to 6:41:18 personal preference, and I'll tell you 6:41:20 how to think about this problem in a 6:41:21 moment. But then underlying this, you 6:41:23 have your concepts at the adset level. 6:41:25 So this is concept one, this is concept 6:41:28 two, this is concept three, etc., etc. 6:41:33 Now, you're either setting budgets 6:41:34 across these manually using an or 6:41:36 the CBO is just going to distribute 6:41:37 budget. Under here, you're going to have 6:41:39 all your ads. A common question I get 6:41:41 is, well, okay, we go and put four ads 6:41:43 or five ads under here. How many ads 6:41:45 should we launch? And when we come up 6:41:46 with new ads for this concept, do we 6:41:47 launch it in here or do we launch a new 6:41:49 ad set? Which is both really good 6:41:50 questions. In terms of ad volume, this 6:41:53 is really open based on budgets, but you 6:41:55 generally want a minimum of three ads 6:41:58 and then a maximum of probably 50 to 6:42:00 100, but honestly, the limit is kind of 6:42:02 the sky based on your ad spend. And so, 6:42:04 I wouldn't really be thinking about 6:42:06 upper funnel uh upper limits. I would 6:42:07 just be thinking about lower limits, 6:42:08 which is a minimum of three. When you 6:42:10 then want to go and introduce more ads, 6:42:13 this is based on whether you're hitting 6:42:14 KPI. So what is going to happen when you 6:42:17 introduce more ads under this ads set is 6:42:19 you are going to reset the learning 6:42:21 phase to some degree because there is 6:42:23 sequencing that is occurring across all 6:42:24 of these different creative and then 6:42:26 when you go and launch more ads in here 6:42:28 you disrupt the sequencing and it 6:42:30 relearns how to distribute spend. Now 6:42:32 that is not good if the adset's 6:42:34 performing well. If this adset is 6:42:36 crushing it and you're really happy with 6:42:37 it and you're scaling it up, don't go 6:42:39 and mess with the learning phase. Don't 6:42:41 go and inject a bunch of new ads. you 6:42:44 might mess the performance up. And I've 6:42:46 actually personally done this years ago. 6:42:47 And so that's why I'm very big on not 6:42:49 touching stuff that's working. If it's 6:42:50 not hitting KPI, and this is way 6:42:52 underperforming, 6:42:54 go for your life. Do whatever you want. 6:42:55 You can turn ads off. It doesn't matter 6:42:57 cuz yes, you're going to break some 6:42:58 sequencing, but the sequencing isn't 6:43:00 working. So, who cares? Go and launch as 6:43:02 many ads as you want in here, right? Do 6:43:03 whatever you want because it's not a 6:43:05 KPI. So, you need to change something. 6:43:07 But if things are working, definitely 6:43:08 don't go and launch more ads in here. If 6:43:10 you go and make more ads for this 6:43:11 concept, let's say the concept's 6:43:13 crashing it, so you go and make 20 more 6:43:14 ads, go and launch another ad set, 6:43:16 right? And call it concept one and then 6:43:19 like shoot two or obviously whatever 6:43:21 naming convention you want to use to be 6:43:23 able to clarify the difference between 6:43:25 these, but then also be able to filter 6:43:26 that they are the same concept. Now the 6:43:28 question becomes, well, how do you 6:43:29 scale? Let's say concept one is 6:43:31 performing well. What do we do? Well, 6:43:33 number one, increase budgets. Never turn 6:43:35 this off. A big mistake that people make 6:43:36 is they take something that's winning 6:43:38 and they turn it off and they go and 6:43:39 launch it in like a scaling campaign. 6:43:41 It's such a rookie mistake just 6:43:42 fundamentally speaking because if 6:43:44 something works, don't touch it. Leave 6:43:46 it. Increase the budget. Now, in 6:43:48 addition to that, sure, you can take 6:43:50 these post IDs and you can wrap them 6:43:52 around into a scaling campaign. That's 6:43:54 fine. And that will work. It works in a 6:43:56 lot of accounts. We run scaling 6:43:58 campaigns in a lot of accounts. A lot of 6:43:59 accounts, we don't run scaling 6:44:00 campaigns. Once again, it's dependent on 6:44:01 the actual ad account in the business. 6:44:03 You can run a scaling campaign. You can 6:44:04 take these, you can post them over here. 6:44:06 If it doesn't work, don't do it. Just 6:44:07 keep it in the testing. And then next to 6:44:08 these two, the only third campaign that 6:44:10 I would usually recommend is to have a 6:44:12 retargeting campaign that is generally 6:44:15 on existing customers. So it gives you 6:44:17 the ability to exclude existing 6:44:20 customers from here and exclude existing 6:44:22 customers from here. So these can purely 6:44:23 be cold and that allows you to control 6:44:25 budgets. Depending on how long the 6:44:26 customer journey is as well, you might 6:44:28 want an adset here on 90-day website 6:44:30 visitors as a midfunnel. But this is 6:44:32 only if you think that you need to 6:44:33 increase frequency on engaged audiences. 6:44:35 Now, what are the metrics that matter 6:44:37 for being able to measure the 6:44:38 performance of creative? Let's list them 6:44:40 out in order of reliability. Number one 6:44:43 is amount spent. Now, this is 6:44:44 counterintuitive. A lot of people look 6:44:46 at this and go, why would budget 6:44:47 distribution to an ad be the most 6:44:49 important proxy for performance over 6:44:51 extended time periods? And the reason 6:44:53 being is due to the sequencing that we 6:44:55 talked about before, right? So, if you 6:44:57 have a bunch of different ads, 6:44:58 attribution is only going to go to the 6:44:59 last ad. So, if we use rorowaz as the 6:45:01 primary KPI, we're going to miss the 6:45:04 fact that meta is actually putting all 6:45:05 your budget here. Why? When this only 6:45:06 has a 1.7 and this has a 2.2. Why would 6:45:09 all the budget go here? Probably because 6:45:10 this ad is causing purchases elsewhere 6:45:13 in the account. And so, this is actually 6:45:15 a great topfunnel asset that likely has 6:45:17 a low frequency that's cutting through 6:45:18 on new audiences. And so, if you're 6:45:20 going to look at these three ads, I 6:45:22 would say that this is the highest 6:45:23 performer if it has double the spend of 6:45:24 the other ones. Even though this is a 6:45:26 higher row, sure, but if this could 6:45:27 scale, it would have more spend in it. 6:45:29 and it doesn't. And so amount spent 6:45:30 becomes the best proxy for performance 6:45:33 when looking at creative. Number two, 6:45:35 sure, is rorowaz or CPA. You just want 6:45:38 to be making sure you're looking at 6:45:39 7-day click or incremental attribution. 6:45:41 Number three is going to be CPC and 6:45:45 CTRs. This is a leading indicator for 6:45:48 the effectiveness of the ad. Obviously, 6:45:50 you can have good CTRs and good CPCs, 6:45:52 but the conversion rate of the ad can 6:45:54 suck and it's not a good ad. And so this 6:45:56 obviously isn't a primary indicator of 6:45:59 the quality of a creative. That is where 6:46:01 you have to use multiple different 6:46:02 metrics in conjunction to be able to 6:46:04 assess quality. Um but it is a good 6:46:06 leading indicator particularly on the 6:46:08 fringes. So like if we look at CPCs as 6:46:11 an example and let's say this is what 6:46:12 your CPCs on average look like across 6:46:14 all of your ads. If above this point is, 6:46:17 let's say, a $3 CPC, and none of these 6:46:19 ads have ever been profitable, well, we 6:46:21 can use this as a leading indicator to 6:46:23 know that these ads at very early spend 6:46:25 will actually never work cuz it's too 6:46:26 expensive to drive traffic. The next two 6:46:28 are hook rates and hold rates. These 6:46:29 only apply to video assets, obviously, 6:46:31 because the hook rate is how many people 6:46:33 make it past 3 seconds in the video. And 6:46:35 the hold rate, there's different 6:46:36 definitions of hold rate. We actually 6:46:38 look at a lot of different definitions 6:46:39 in our reporting, but as a 6:46:41 generalization, this is how many people 6:46:42 make it past 15 seconds. Obviously, hook 6:46:45 rates and hold rates are not correlated 6:46:48 to conversion performance, but they are 6:46:50 a diagnostic tool that allows us to 6:46:52 iterate on creative based on what we 6:46:54 think is letting us down. And so, we may 6:46:57 have a creative that we thought was 6:46:59 going to do really well, but the hook 6:47:00 rate is terrible against our average. 6:47:02 And so, we try to switch the hook out to 6:47:04 try to fix that to see if the body 6:47:06 actually performs. Same thing applies to 6:47:08 hold rate. People might not be holding 6:47:09 through the bridge. Turns out the bridge 6:47:11 was actually done poorly. So we redo 6:47:13 that part of the video and then we can 6:47:14 iterate accordingly. So before I said I 6:47:16 would explain verse CBO and the risk 6:47:18 profile here. So the reason why you 6:47:20 would choose one over the other is 6:47:22 actually based on risk tolerance. So 6:47:24 what happens when you use CBO is that 6:47:25 it's going to play in to Prito's 6:47:28 principle. This applies literally 6:47:30 everywhere I've ever seen it which is 6:47:32 that 80% of something will drive will be 6:47:35 driven from 20%. So this applies into 6:47:38 companies where 20% of people will drive 6:47:40 80% of revenue generation. This applies 6:47:42 into the ad account which is that 20% of 6:47:44 ads will drive 80% of revenue. 20% of 6:47:47 ads will hold 80% of revenue. Crazy 6:47:49 thing about Parto's principle is that it 6:47:51 compounds in on itself. And you see this 6:47:53 within every single data set I have done 6:47:55 on ad accounts breaking down their ad 6:47:57 profile which is that of this 80% 80% of 6:48:02 this which comes out to 64% of total is 6:48:05 driven by 20% of this which this comes 6:48:07 out to 4%. So 4% of ads will hold about 6:48:11 64% of total spend and total revenue 6:48:14 generation within the business. Now CBOS 6:48:17 play into Pareto's principle. And what 6:48:19 that means is that if you just allow the 6:48:21 campaign to distribute budgets however 6:48:23 it wants, it will naturally end up 6:48:25 distributing budgets like this where 6:48:27 majority of all of your budget in the 6:48:29 account will go to 4% of your ads. Now 6:48:31 that's fine cuz that's going to yield 6:48:33 the best efficiency and performance. But 6:48:35 it is super risky because if these ads 6:48:37 fatigue, you are in a very bad position. 6:48:40 You don't have backups. If we go back to 6:48:42 the start of the video, which is 6:48:43 portfolio management, CBOS are terrible 6:48:46 for portfolio management, okay? Because 6:48:48 you're going to overlever into a few 6:48:51 ads. This also becomes an issue when you 6:48:53 want to control sell through rates 6:48:55 across multiple different products. And 6:48:57 so you might have a large product 6:48:58 portfolio that all sit under the one CBO 6:49:01 and then all of the spend just goes to a 6:49:03 few particular products because Parto's 6:49:05 principle applies into a product level 6:49:07 as well. 20% of products will drive 80% 6:49:09 of revenue. And so your entire business 6:49:11 will end up skewing and increasing the 6:49:13 risk profile which you might be fine 6:49:14 with but that's the trade-off that you 6:49:16 have to make. So when you think about 6:49:17 these two CBO is increased risk but 6:49:22 likely increased ROI. This is decreased 6:49:25 risk but likely slightly lower ROI. And 6:49:28 it depends on which position you want to 6:49:30 be in. I personally, if I'm running the 6:49:32 business, I actually choose this every 6:49:35 day of the week dependent on like what 6:49:37 season and where you are in the 6:49:39 business. And so there are periods in 6:49:40 time in which I would actually switch 6:49:42 over to CBO if I need a quick spike in 6:49:45 efficiency. Efficiency is isn't where we 6:49:46 need it to be. I don't care about 6:49:47 testing and de-risking right now. We 6:49:49 just need ROI to come through. But then 6:49:52 ideally you want to be in a steady state 6:49:53 over here where the business is derised 6:49:56 across more ads across more products and 6:49:58 a lot more stuff is working so that if 6:50:00 something stops working you aren't in a 6:50:02 terrible position. Uh this next point 6:50:03 I'm particularly calling out because we 6:50:05 work with a lot of nine figure large 6:50:06 retailers here in Australia and if any 6:50:08 of them are watching then they'll know 6:50:10 that this problem exists which is that 6:50:12 in a lot of large retailers who a lot of 6:50:14 my audits are spent on there is this 6:50:16 9010 problem with desire versus 6:50:19 painpointled content and this 6:50:21 fundamentally comes from old style 6:50:22 marketing which is campaign shoots. And 6:50:25 so a lot of these large uh retailers in 6:50:28 fashion or whatever industry it is, they 6:50:30 were built around campaign shoots. And 6:50:32 campaign shoots do one thing and that is 6:50:34 that they build desire. And so they have 6:50:37 models, they have shots that make you 6:50:38 want to desire having the product, but 6:50:41 they don't play to pain points at all. 6:50:43 And so you end up with the ad account 6:50:44 being significantly overweighted towards 6:50:46 desire rather than towards pain. And at 6:50:49 the end of the day, the best performance 6:50:51 marketing assets are those that sit on 6:50:53 pain. And so I would strongly recommend 6:50:55 looking at your current creative mix 6:50:57 through this lens and thinking about how 6:50:59 do we actually start to shift this ratio 6:51:02 over to 60/40. Now I personally think 6:51:05 that the best brands are probably closer 6:51:07 to 7030 here. And you can almost think 6:51:09 of this in a way of like this is 6:51:11 performance marketing driven assets. 6:51:13 This is branding driven assets. Now one 6:51:15 last thing in the testing framework is 6:51:17 assets for retargeting. So how should 6:51:19 you be thinking about retargeting ads on 6:51:22 bottom of funnel? This is something I've 6:51:23 been saying for a very long time. This 6:51:25 is probably one of my first YouTube 6:51:26 videos that we put out, which is that on 6:51:28 your bottom of funnel retargeting 6:51:30 campaigns, you want to think about 6:51:32 objection handling. This is the primary 6:51:34 purpose of a product ad. And you can 6:51:37 think about this mathematically, which 6:51:38 is that let's say you have a 5% 6:51:40 conversion rate on the website. That 6:51:42 means that 95% of people did not buy. 6:51:45 Now the question is why didn't these 6:51:48 people buy? What were their objections? 6:51:50 Maybe 20% of them had a price objection. 6:51:54 It was just priced too high. They never 6:51:56 will buy at that particular price. Now, 6:51:58 they might buy in the future when you're 6:52:00 on sale, but right now they won't buy. 6:52:01 20% of people don't trust the brand. 6:52:04 There wasn't enough trust, credibility 6:52:06 markers on the website or the ads that 6:52:08 they particularly saw to reinforce them 6:52:10 to actually purchase the product. 20% 6:52:12 don't like the quality. They think the 6:52:15 quality is probably going to be bad. 6:52:16 Now, this is arguably a trust issue, but 6:52:18 these are really two different things, 6:52:19 and they should be come at in two 6:52:20 different ways. And then maybe another 6:52:22 20% of people are just not ready to buy 6:52:24 yet. Okay? So, it's too soon. Was the 6:52:27 first ad they've seen, it was the first 6:52:29 ad they've clicked on, they just need a 6:52:31 little bit more time to purchase. So, we 6:52:33 want to understand this mix. We want to 6:52:35 understand why people aren't buying. And 6:52:36 then we want to handle all of these 6:52:38 objections in our bottom of funnel 6:52:40 created. And so, on the price objection, 6:52:42 we want to be showing discounts. We want 6:52:43 to be value anchoring. We want to even 6:52:45 going for a creative that's like a cost 6:52:47 per use angle. These actually crush on 6:52:50 bottom of funnel. So if you're selling 6:52:52 like a supplement, you compare it to a 6:52:53 coffee. It's like half the price of a 6:52:55 coffee with the same caffeine, etc., 6:52:57 etc. If they don't trust the brand, we 6:52:59 want to come in with social proofing. 6:53:01 How many customers do we have? How many 6:53:02 fivestar reviews do we have? Do we have 6:53:04 any user generated content testimonials 6:53:05 that we could use here? If they're 6:53:07 doubting the quality of the product, we 6:53:09 want to show UGC of the product actually 6:53:11 in use. We want before and afters. We 6:53:13 want demonstrations and if it's simply 6:53:15 too soon for them to buy yet, we want to 6:53:17 be staying present and top of mind with 6:53:19 educational or lifestyle content. And 6:53:21 then this is what the mix should be in 6:53:23 all of our bottom of funnel assets. And 6:53:24 this is how you ultimately yield strong 6:53:26 conversion rates once people have been 6:53:28 moved through the funnel. The question 6:53:29 is, does diversity or volume matter more 6:53:31 when it comes to creative? So you want 6:53:32 to be thinking through volume through 6:53:34 this hierarchy, which is that volume 6:53:36 matters the most, then concept quality, 6:53:39 then hook strength, then editing 6:53:41 quality, then UGC talent, then 6:53:42 storytelling all the way down the 6:53:44 bottom. Now, it's not to say that any of 6:53:45 these components aren't important, but 6:53:48 it's the fact that volume ends up 6:53:49 creating quality and hit rates are 6:53:52 limited within an ad account. And so a 6:53:53 way to conceptualize through this idea 6:53:55 is that if you take the best creative 6:53:58 strategist, you can pull 10 of them out 6:53:59 of 10 different businesses and you can 6:54:01 show them five ads that are all 6:54:03 objectively good creative and you tell 6:54:06 them which of these five was the winner. 6:54:08 Which one produced the most spend or the 6:54:11 most return? None of them will be able 6:54:12 to guess it. It will just be a complete 6:54:14 guessing game. They'll assign reasons as 6:54:17 to why they believe this creative is 6:54:18 better than the other ones. might have 6:54:20 been due to the concept quality, might 6:54:21 have been due to the hook. But the 6:54:23 reality is is that people are not very 6:54:25 good at guessing the top end level of 6:54:28 creative performance. And so instead, 6:54:30 you want to understand where the bar of 6:54:32 quality is and just ensure that ads are 6:54:35 reaching that standard. And so if all of 6:54:37 these five ads sit above this bar of 6:54:39 quality, and we could objectively say 6:54:41 that this is a seven out of 10. So if an 6:54:43 ad is better than a seven out of 10, it 6:54:47 meets that bar and it is 6:54:48 indistinguishable from any other 6:54:50 creative in being able to try to guess 6:54:52 its performance. Where you end up dying 6:54:54 in creative strategy is trying to win 6:54:58 the game of understanding which ads will 6:55:00 perform well at a high level. No one is 6:55:02 able to objectively do this well. I've 6:55:04 never seen anyone be able to pick 6:55:05 winners. And so you instead want to 6:55:07 become very good at understanding what 6:55:09 is the minimum bar of quality. What does 6:55:10 a good ad look like? And then how do we 6:55:13 reach that level of quality and then 6:55:14 maximize volume above it where people 6:55:16 get volume completely wrong and the 6:55:18 example that I gave way back at the 6:55:19 start which was people were 6:55:20 misunderstanding activity for creative 6:55:22 strategy is that they were putting tons 6:55:25 of ads into the account but they didn't 6:55:26 have an understanding internally of what 6:55:28 the bar of 7 out of 10 actually looks 6:55:30 like. And so when you go and do an ad 6:55:32 breakdown of all the creatives that they 6:55:34 were putting into the account which was 6:55:35 thousands a month the ads just weren't 6:55:37 very good. They didn't have any of the 6:55:39 elements of a good concept. They didn't 6:55:41 have any of the elements of a good hook. 6:55:43 They didn't have any of the elements of 6:55:45 formats that were relevant to the stage 6:55:47 of awareness of the creative that they 6:55:48 were trying to produce. And so because 6:55:50 of that, yes, they put tons of volume in 6:55:52 the account. Yes, they followed this 6:55:54 hierarchy, the most important thing on 6:55:56 it. But it was just junk volume. It was 6:55:58 just a bunch of ads that aren't even 6:55:59 good. And so instead, you need to focus 6:56:02 your efforts on how do we understand 6:56:03 where the bar is, then how do we 6:56:04 maximize volume, and then how do we 6:56:06 start steping down through all of the 6:56:08 different priorities to create a good 6:56:10 ad. The question off the back of this 6:56:11 then becomes, okay, well, how much 6:56:13 volume do we actually need then? And 6:56:15 there's a really easy, simple answer 6:56:17 which I'll give to you. And then there's 6:56:18 a more complicated answer that you'd 6:56:19 have to go and build a financial model 6:56:21 out for. And if you're an e-commerce 6:56:22 brand, uh, we'll give it to you 6:56:24 completely for free. We'll put the link 6:56:25 in the YouTube description below. So, 6:56:27 the simple answer is for every $1,000 in 6:56:30 monthly spend, you want one new ad. And 6:56:34 so, if you're spending 30K a month, 30 6:56:36 new ads, 50K a month, 50 new ads. Now 6:56:38 this is in AUD also translate this into 6:56:41 USD roughly that ad volume is fine. Now 6:56:43 the better way to do this is to back 6:56:45 propagate based on hit rates and 6:56:47 expected spend per ad. So what you can 6:56:49 do is you can calculate the average 6:56:51 spend of an ad. So if you know that 6:56:53 across the last year of ads you've 6:56:55 launched let's say 500 ads and you've 6:56:58 spent $500,000. Well then each ad here 6:57:01 on average spent $1,000. 6:57:04 Now, if you know the average return of 6:57:06 an ad, which you could look at either 6:57:08 your acquisition me, which is new 6:57:10 customer revenue divided by ad spend in 6:57:11 the business, or you could just look at 6:57:13 rorowaz, but if you're going to look at 6:57:14 this, look at it on a 7-day click basis. 6:57:17 And let's say your average rorowaz is a 6:57:19 three and you've got $1,000 in mean 6:57:23 spend. Well, then we know that the 6:57:25 average return or you could call this 6:57:27 the expected value if we're using 6:57:30 probabilistic math is then $3,000 6:57:34 in revenue. So every time we launch an 6:57:36 ad, we should expect on average on a 6:57:39 larger data set that we generate $3,000 6:57:41 in revenue. So then we can start to back 6:57:43 propagate this into our actual revenue 6:57:45 targets. Now more specifically, I would 6:57:47 actually delineate this down into new 6:57:49 customer revenue to be specific to 6:57:51 acquisition. So, if you have a 300K 6:57:55 monthly new customer revenue target, you 6:57:58 take 300K, you divide by your 3,000. 100 6:58:02 creatives per month are required. Now, 6:58:05 there's a few nice things about this. 6:58:07 Number one, you can flex the actual 6:58:09 volume of creative in conjunction with 6:58:11 spend, which is what you should be 6:58:12 doing. As you go through the year and 6:58:15 your forecast changes in terms of 6:58:17 expected new customer revenue, your 6:58:19 creative volume should flex accordingly. 6:58:22 So this should be exactly what your 6:58:23 revenue looks like, but also what your 6:58:25 creative volume looks like. Now, there 6:58:27 are two disadvantages to this simplified 6:58:29 model, which is that number one, it 6:58:30 doesn't take into consideration the 6:58:32 average time of an ad spending. And so 6:58:35 these ads might generate $3,000 in 6:58:37 revenue, but it might take 4 months. And 6:58:39 so that needs to be factored in because 6:58:40 obviously volume needs to be a lot 6:58:41 higher then. And it also doesn't factor 6:58:43 in the difference between a winner and a 6:58:45 loser. If you got 100 losing ads this 6:58:48 month, you will generate barely any 6:58:50 revenue. If you got a 100 winners, 6:58:52 you'll generate like multiple millions 6:58:54 in new customer revenue. And so, you 6:58:56 need to forecast losing ads versus 6:58:58 winning ads separately. And then you 6:59:00 need to add the time component. We've 6:59:02 built a model that does all of this. 6:59:03 Once again, if you're an ecom brand and 6:59:05 you want it, you can click the link in 6:59:07 the description, uh, fill out the form, 6:59:08 we'll send it out to you. Now, when 6:59:10 we're thinking through creative budget, 6:59:11 how much budget should go towards 6:59:13 production versus actual distribution of 6:59:15 the assets, if you're at 4 to 30k a 6:59:18 month in spend level, there is no reason 6:59:20 as to why your content shouldn't be 6:59:22 super founderled. And the actual budget 6:59:25 here on production is your time rather 6:59:27 than money because you really don't need 6:59:30 to go and pay for production on this 6:59:32 kind of spend level. This can all be 6:59:33 done literally through statics in most 6:59:35 businesses, but even just founderled 6:59:37 videos can get you to 30K a month in 6:59:39 spend easily. And honestly, I would 6:59:41 challenge that the founder of a business 6:59:42 at this size should be creating content 6:59:45 because content and creative is the 6:59:46 largest lever that exists within the 6:59:48 business. And so understanding what good 6:59:50 looks like, what good creative looks 6:59:51 like, getting an understanding for how 6:59:52 to piece it together, what the editing 6:59:54 process looks like, what the ideation 6:59:56 process looks like gives you the ability 6:59:58 to then KPI and train people moving 7:00:00 forward on one of the most critical 7:00:01 roles in the business. As you then move 7:00:03 into 30 to 100k per month in ad spend, 7:00:06 you want to be thinking of allocating 7:00:07 25% of media budget towards production. 7:00:10 So at 100K a month, that's 25K a month 7:00:12 into creative production. This is way 7:00:14 more than like 99% of brands are 7:00:17 actually spending. And that's also one 7:00:18 of the reasons why brands don't actually 7:00:20 scale and do that well. It's because 7:00:22 they significantly underweight the 7:00:23 allocation that needs to go into their 7:00:24 media budget. Now, this is an 7:00:26 uncomfortable range because the cost of 7:00:29 production is so high excessive to the 7:00:31 scale of the business right now that it 7:00:33 feels off. It feels like this is too 7:00:35 high. But it does democratize as you 7:00:37 achieve more scale. Once you crack 7:00:39 through 100K plus, this percentage 7:00:42 really should asmmotope down to around 7:00:44 about 10% of media budget. And so as a 7:00:47 function of revenue in the business, 7:00:48 let's say that you're at a 25% ME, 10% 7:00:51 of this would therefore be 2.5% of total 7:00:54 revenue in the business. So 2.5% of 7:00:56 total revenue should be going towards 7:00:58 media budget. This kind of spend 7:00:59 allocation allows diversity in the 7:01:01 creative that you're putting in. This is 7:01:03 enough budget to have UGC going into the 7:01:05 account, to have partnership ads going 7:01:07 into the account, to have a full-time 7:01:10 either international designer or someone 7:01:11 part-time uh in house on statics, and it 7:01:14 can likely support a lowcost uh agency 7:01:17 in some capacity to supplement with 7:01:19 another creative type in here as well. 7:01:20 Maybe it's VSSLs or something. I'll give 7:01:22 you a real example of a client that 7:01:25 unfortunately we had to part ways with 7:01:26 about uh 6 to 8 months ago because we 7:01:29 couldn't get this concept across to 7:01:31 them. And this is why I stress this so 7:01:33 much in a lot of content that we put out 7:01:34 now. And it's really important that 7:01:36 people understand this particularly 7:01:37 through the initial discussions when 7:01:38 they start talking with us because we 7:01:40 really don't want to onboard brands that 7:01:43 aren't across this concept. So this 7:01:45 brand was spending 350k 7:01:48 per month on Meta when they started 7:01:50 working with us. Their CAC was declining 7:01:52 pretty steadily over the course of the 7:01:54 last 2 years. Now, they only had about 7:01:58 15 ads live within the account on a 350K 7:02:01 a month spend, which is crazy. And so, 7:02:03 right out of the gates, we made a bunch 7:02:05 of media buying changes. The account was 7:02:07 super oversegmented. The previous media 7:02:09 buyer was doing so much over-the-top 7:02:11 stuff because he didn't have anything to 7:02:13 work with from an ad and creative 7:02:15 perspective. And so, when you don't have 7:02:17 anything that you can do on creative, 7:02:18 what do you do? Well, you go to the next 7:02:20 thing, and that's let's create 25 7:02:22 different campaigns. let's use big caps, 7:02:23 cost caps, let's try everything to try 7:02:25 to squeeze these 15 ads, which 7:02:27 ultimately isn't the lever that's going 7:02:29 to double this business. And so when we 7:02:30 did the audit, we identified this as the 7:02:32 bottleneck. And what we did immediately 7:02:34 is as we came in, we went and took all 7:02:36 old performers and started relaunching 7:02:38 and reworking them a little bit. Even 7:02:39 though at the time we weren't doing 7:02:41 creative production in house, we just 7:02:43 threw this on top cuz we knew that this 7:02:45 was the limiter that we had to get this 7:02:46 done. So we took the ad account up to 7:02:48 about 60 ads. Nowhere near where it 7:02:50 needs to be, but this is all we could 7:02:52 do. in terms of high quality assets that 7:02:54 have been turned off in the past. Now, 7:02:55 what the results ended up being for this 7:02:57 brand is we actually turned them around, 7:03:00 which was one of the most insane case 7:03:02 studies that we've done cuz we did it 7:03:03 with no new creative and we actually 7:03:05 took them from 350K a month in spend, we 7:03:08 increased their spend by about another 7:03:09 40K and we decreased their CAC by about 7:03:12 20 to 30%. And so, new customer revenue 7:03:14 expansion off the top of my head, and 7:03:16 there's a case study out there on this 7:03:17 somewhere, I think it was about like 7:03:19 28%. And so we were able to turn this 7:03:22 brand around substantially because of 7:03:23 this. Now the issue was that this 7:03:26 founder was very much so of the opinion 7:03:29 that this spend on meta is a revenue 7:03:31 driver. Any spend towards creative 7:03:33 production is a cost center. And so 7:03:36 because of that they refused to do any 7:03:38 more creative production. We introed to 7:03:40 multiple agencies. They went and chatted 7:03:42 to them. Uh they actually signed on and 7:03:44 then pulled out a of a lot of those 7:03:45 deals. And so we weren't able to get any 7:03:47 more ads, any more creative. And so 7:03:50 because of that, there's only so far 7:03:52 that you can push media buying over 7:03:53 time. And so performance started to come 7:03:55 down. And then as performance started to 7:03:57 trickle down due to these 60 ads 7:03:59 fatiguing, ads started to need to be 7:04:01 turned off. Spend started getting 7:04:02 distributed away from them. And the 7:04:04 account slowly fell back to what it was 7:04:05 previously. Now, the proposition that we 7:04:07 put forward to this brand well before 7:04:09 any of that happened was that if you 7:04:12 took just $10,000 of this budget, just 7:04:15 10K out of it, and you decreased from 7:04:18 Meta and you moved it over to creative 7:04:20 production, this could get us with an 7:04:21 agency that we were talking to about 40 7:04:24 assets. Not all unique, some were just 7:04:25 hook swap, some were statics, but it 7:04:27 would give us 40 assets. This would 7:04:28 allow us to almost double the amount of 7:04:30 active ads in the account, which would 7:04:32 likely unlock at least a 5% improvement 7:04:35 in efficiency. Now, a 5% improvement in 7:04:38 efficiency on 350K is the equivalent of 7:04:42 times 2.5 a 17.5K 7:04:45 saving. So, we could either drop 17.5K 7:04:48 out of this budget, which we already 7:04:50 did. we drop 10 so we've effectively 7:04:52 made 7 1/2 grand or we could keep the 7:04:54 budget where it is and this would 7:04:56 provide us with an incremental 3x was 7:04:58 their return at the time on this so we 7:05:00 would make an extra 40 to 50k a month 7:05:02 and so you would actually get revenue 7:05:04 expansion in this business by keeping 7:05:06 spend the same but redistributing it 7:05:08 over to production. It's a similar 7:05:09 concept to just changing between 7:05:11 channels but instead of it being a 7:05:12 channel it's creative with an output. 7:05:15 But unfortunately, this founder was very 7:05:16 set on the fact that they didn't need 7:05:18 new creative that we could just 7:05:19 repurpose these existing assets and that 7:05:21 would continue the business going. And 7:05:23 this all definitely could have been 7:05:24 fixed well ahead of time simply through 7:05:26 a slight allocation of this budget over 7:05:29 into production. So definitely avoid 7:05:32 being that brand that sees creative as a 7:05:34 cost center rather than a revenue 7:05:36 driver. Now, one last tactical piece of 7:05:38 advice here that applies right now as of 7:05:41 April 2026. This might change as the 7:05:44 algorithms change. This might not apply 7:05:46 into the future. But one of the best 7:05:49 ways to quickly test high volumes of 7:05:51 assets right now, we actually do this on 7:05:52 all of our own ads, is that you post 7:05:54 them as trial reels and you cut the CTA 7:05:56 off. So if there's a call to action at 7:05:58 the end, cut the call to action out, 7:05:59 throw it up as a trial reel on the 7:06:01 business page that's going to run 7:06:02 through or the personal page that's 7:06:03 going to run through. And then what you 7:06:04 will get back is two things. Number one, 7:06:06 you'll see how viral does the post 7:06:08 actually go? Does it get any views? Does 7:06:09 it not? So you're going to have 7:06:10 comparative data. But number two, you're 7:06:12 also going to get retention graphs, 7:06:13 which is going to show you how you've 7:06:14 retained viewers throughout the ad. Now, 7:06:18 the retention graph is going to show 7:06:19 you, was the hook bad, was the bridge 7:06:21 bad, was the body bad. And this lets you 7:06:23 iterate quickly on the next shoot 7:06:25 without needing to actually push the ad 7:06:27 straight into the ad manager. And you 7:06:28 can do this before it's perfected. So, 7:06:30 if there's some issues with the captions 7:06:32 or there's some stuff that isn't great 7:06:33 and it needs to go in for another round, 7:06:35 you can just take the V1, throw it as a 7:06:36 trial reel, cut the call to action off 7:06:38 in app and get feedback while that round 7:06:41 two version is actually getting worked 7:06:42 on. And then you could pass even more 7:06:44 feedback on during that round two within 7:06:47 24 hours before you've even got the 7:06:48 final asset. And so, we found this is a 7:06:50 really effective way to take bulk 7:06:52 amounts of creative and quickly test 7:06:54 particularly on hook variations as well. 7:06:55 So we can take like 20 hook variations, 7:06:57 throw it up as trial reels, what hooks 7:06:59 actually perform the best, and then only 7:07:00 put five ads into the account rather 7:07:02 than wasting 20 ads and spending money 7:07:03 on trying to figure it out. So now we 7:07:05 move into creative fatigue. So there's 7:07:07 two concepts to understand with creative 7:07:09 fatigue. Number one is the two reasons 7:07:11 why creatives fatigue. And the other 7:07:13 concept is to understand the maximum 7:07:15 spend threshold of an ad over time. So 7:07:18 when you think about launching an ad and 7:07:19 on the y- axis we'll put spend and on 7:07:22 the x- axis we have time. You launch the 7:07:25 ad, it does well. And so, one of two 7:07:27 things happen. Either you start manually 7:07:30 spending more on it, so you start 7:07:31 increasing budgets, or Meta will go and 7:07:33 automatically distribute spend to it. 7:07:35 And the spend goes up and up and up and 7:07:36 up. Eventually, there will be a limit in 7:07:39 how much daily spend that ad can hold. 7:07:42 You'll reach that limit. You'll probably 7:07:44 go past it. You'll correct. And then 7:07:46 you'll find the maximum daily spend 7:07:48 limit of the ad. Let's say it's at like 7:07:50 $500 a day or $1,000 a day. From there, 7:07:53 this ad will continue spending moving 7:07:55 forward in time. It will eventually 7:07:56 start fatiguing. It will start falling 7:07:58 off. You'll either start pulling budgets 7:08:00 or the CBO will pull budgets out of it 7:08:02 and then eventually you will just turn 7:08:03 it off entirely or Meta will stop 7:08:05 distributing spend. Now, this area under 7:08:08 the curve here is the total spend 7:08:10 capacity of the ad. So, you can go and 7:08:12 do derivative math and just get an 7:08:14 understanding of how much area there is 7:08:16 there and you will have total spend 7:08:20 capacity. Now, what's important to 7:08:21 understand is number one, every ad has a 7:08:23 total spend capacity. Every ad will 7:08:25 fatigue. Every ad will get turned off. 7:08:27 Number two is you can change what this 7:08:29 graph looks like over time. And so, if 7:08:31 you want, you can squeeze ads harder in 7:08:34 the short term, but they will fatigue 7:08:37 faster. And so, you could take this ad 7:08:39 all the way up to here in daily spend, 7:08:41 but it just means it's probably going to 7:08:42 fatigue and be turned off here. Or you 7:08:44 could go even more aggressive and you 7:08:45 could actually launch it at a high 7:08:46 budget, take it way up, but then it's 7:08:48 going to fall off. Or you could 7:08:49 obviously do the opposite and you could 7:08:51 under portfolio management theory, which 7:08:53 is what we did at the start, is that you 7:08:56 actually just want to introduce this ad 7:08:57 but keep it at $100 a day and you'd 7:08:59 rather have it live for 6 months than 7:09:02 die off within 6 weeks. Now, this is 7:09:05 ultimately more so going to be a product 7:09:07 of the inventory constraint and demand 7:09:09 within the business. So, if you have a 7:09:11 lot of inventory that you need to sell 7:09:12 through for the current profitability 7:09:14 position in that month, then obviously 7:09:16 you should go with option one here and 7:09:18 just squeeze the winning ad to sell 7:09:19 through as much product as possible, 7:09:21 even if it's at a degraded me. On the 7:09:23 other hand, if you don't have much 7:09:25 inventory, um, or if you're waiting for 7:09:27 more inventory to happen or if you're 7:09:28 growing too quickly that your cash 7:09:29 conversion cycle can't actually uh, keep 7:09:32 up with the growth rate, well, then why 7:09:34 would you do this? This would be a 7:09:35 ridiculous decision. you would instead 7:09:37 want to operate at higher efficiency and 7:09:39 let the ad breathe for 6 months. 7:09:41 However, if you're working with an 7:09:42 agency, they don't have any of that 7:09:45 wider context within the business unless 7:09:46 they're asking for it, unless you're 7:09:48 actively giving it to them. And so, they 7:09:50 will see an ad that's winning and they 7:09:52 will go, let's squeeze the hell out of 7:09:53 it because of two things. Number one, we 7:09:55 just started working with this brand and 7:09:57 this will give us a great 30 to 90day 7:09:58 case study. And number two is, well, 7:10:01 this is giving us the best rorowaz and 7:10:02 we report on rorowaz or efficiency and 7:10:04 profit and so let's do this. But the 7:10:06 issue is it completely screws up the 7:10:08 long-term jevity of the ad account and 7:10:09 the business because you've over 7:10:11 prioritized short-term gain for the 7:10:13 long-term stability in the ad account 7:10:15 and the inventory purchasing. So you're 7:10:17 just going to go out of stock here and 7:10:18 you're going to ruin a lot of what's 7:10:19 going on. Then the two drivers of 7:10:21 creative fatigue. Number one is the ad 7:10:24 itself just become stale. So this is a 7:10:26 function of it's just reached its 7:10:28 maximum spend capacity and it will 7:10:29 ultimately just achieve at the end of 7:10:31 its life cycle. Um, this could also be 7:10:34 just a function of the ad has the same 7:10:36 visuals, it has the same concepts, it's 7:10:38 targeting the same audience as the other 7:10:40 creatives in the account. And so you'll 7:10:42 get much more faster fatigue because 7:10:43 you're targeting the same audience with 7:10:45 the same value propositions. The second 7:10:46 reason ads fatigue is that the audience 7:10:48 is actually just too small. When we went 7:10:50 through the persona component of uh this 7:10:54 video, we talked about how specificity 7:10:56 in the persona is critical to being able 7:10:58 to script a better ad that resonates 7:11:00 with a very specific target demographic. 7:11:02 This is also going to generate way 7:11:04 better results because you're talking to 7:11:06 a few people rather than a lot. Making 7:11:08 very good top offunnel unaware ads is 7:11:10 incredibly difficult. The skill ceiling 7:11:12 is very high. Most people shouldn't try 7:11:14 to do it unless you're already a $50 7:11:16 million plus brand. Now with that, if 7:11:19 you speak to a very small persona, there 7:11:22 is obviously going to be a maximum daily 7:11:24 spend capacity that you can spend 7:11:26 talking to that persona. And so if we go 7:11:28 back to the example before of men trying 7:11:32 to run sub 4-hour marathons that have 7:11:34 gone to the gym, that are trying to cut 7:11:36 5 minutes off their time using creatine. 7:11:39 Very specific target demographic. Now, 7:11:40 there's still a lot of people there. A 7:11:41 lot of people are trying to run 7:11:42 marathons and trying to run sub4, but 7:11:44 you're not going to be able to spend, 7:11:45 particularly in the Australian market, 7:11:47 $10,000 a day on that creative. It's 7:11:49 just not going to happen. And so, you 7:11:51 need to find out what the spend capacity 7:11:53 is, keep it there, rotating creative so 7:11:56 that once these fatigue, we can continue 7:11:57 to hit that audience over and over again 7:11:59 because that is an audience that churns, 7:12:01 which is really important. So, people 7:12:02 will enter into that audience, then run 7:12:04 a marathon and enter out of it. So 7:12:06 there's constantly new people coming in 7:12:08 that we can continue to saturate, but we 7:12:11 don't want to oversaturate, over 7:12:12 prioritize, overspend, drive up 7:12:14 frequency, and then just ruin the 7:12:16 profitability of us targeting this 7:12:18 audience altogether. Now, an obvious key 7:12:20 learning here is that broader audiences 7:12:24 will scale much further. And so if you 7:12:28 have a top offunnel unaware ad that 7:12:29 speaks to a much larger TAM total 7:12:32 addressable market, you will obviously 7:12:33 be able to spend a lot more money on 7:12:34 that. In a B2B context, an example of 7:12:37 this is that if I have a persona call 7:12:39 out at the front of one of our ads that 7:12:42 says e-commerce CFOs in Australia owning 7:12:44 over $und00 million a year in your 7:12:47 business, that is such a tiny pool. 7:12:50 There's like how many people fit that 7:12:52 particular persona in Australia? It's 7:12:53 probably a thousand, maybe 2,000. And so 7:12:57 because of that, when I go and put that 7:12:58 ad in market, I can't spend a lot on 7:13:00 that ad. There would be no point in me 7:13:01 going and putting $10,000 a day behind 7:13:03 that creative because I would then just 7:13:05 be hitting the thousand people like 600 7:13:08 times a day. Makes no sense. But then if 7:13:10 I expand the persona call out to 7:13:12 e-commerce founders earning over 100k a 7:13:14 year, well then that's a much larger 7:13:17 target demographic. So the daily spend 7:13:19 capacity improves. Now does that mean I 7:13:21 should just do broad callouts? 7:13:23 Absolutely not. Because broad callouts 7:13:25 means you have to have better creative 7:13:26 that resonates with more people. And so 7:13:28 you're actually better off having the 7:13:30 more specific persona call outs that 7:13:31 force you to create better assets and 7:13:34 then move into broader audiences once 7:13:36 you hit a particular scale. In terms of 7:13:38 tactical things that you can do when 7:13:39 creatives fatigue, there are changes 7:13:41 that you can make. Number one is you can 7:13:43 make iterations. And so you can take an 7:13:45 asset that's already done well, which I 7:13:47 mentioned before, a hero ad that started 7:13:48 to fatigue, and you can just rotate new 7:13:50 hooks on, and you'll be able to get more 7:13:52 spend through that asset. Number two is 7:13:54 you can move the ads over into a cost 7:13:56 cap campaign and you can try to get more 7:13:58 spend through here. Now, this will also 7:14:00 work. It will fatigue the creative 7:14:01 faster and it's a short-term tactical 7:14:04 media buying strategy. It's not 7:14:06 something that's really going to 7:14:06 materially change the business. Number 7:14:08 three, you could go and duplicate 7:14:10 adsets. It's not something that I would 7:14:12 recommend. I'm not even going to go and 7:14:13 put it down here. I know it's something 7:14:15 that people still argue works. I think 7:14:17 it's too tacky as a medium buying 7:14:19 strategy. It used to work back in the 7:14:20 day and I could explain the concept as 7:14:22 to why it works and why it doesn't now. 7:14:23 But if it's something you want to do, 7:14:24 you can do it, but I wouldn't recommend 7:14:25 it. You can take the post ID of the 7:14:28 creative so it retains all of the 7:14:30 engagement and you can go and move that 7:14:32 into another campaign like a scaling 7:14:34 campaign to try to get more spend 7:14:36 through it. This sometimes works, but 7:14:37 once again, like this isn't going to 7:14:38 save the account. And then you could 7:14:40 rotate, which is kind of a variation of 7:14:42 number one. You could rotate formats. 7:14:44 And so if this is a video that's doing 7:14:46 really well, you could try to translate 7:14:48 it into an image. If it's a VSSL that's 7:14:51 doing really well, you could try to 7:14:53 translate it into UGC. If it's an image 7:14:55 that's doing really well, you could try 7:14:57 it translated into an unaware long form 7:15:00 copy static image, which is a completely 7:15:02 different strategy. So they're really 7:15:03 the four options. As a creative 7:15:05 fatigues, it is just worth noting that 7:15:07 once a creative fatigue fatigues, 7:15:08 there's not like a whole lot you can do 7:15:10 about it. you know, that's why you need 7:15:11 the portfolio management approach and 7:15:13 you need other ads that are doing well 7:15:14 so that you can fall back on them. There 7:15:16 are things that you can do though just 7:15:17 to squeeze a little bit more out of the 7:15:19 back end of a creative, which is what 7:15:21 these tactics are. Tying all of that 7:15:23 back into portfolio management on 7:15:25 accounts spending greater than 50k per 7:15:27 month really the job of the media buyer 7:15:30 is portfolio management across the ads 7:15:33 and diversifying risk here portfolio 7:15:36 management against the products and 7:15:38 assuring commercial objectives within 7:15:40 the business is aligned with what is 7:15:42 actually occurring within the ad account 7:15:44 and so the 8020 parto principles rule 7:15:47 applies into ads and so the question 7:15:49 becomes do we want this naturally 7:15:51 occurring within the account or do we 7:15:53 want to force back against it through 7:15:55 introducing some structure so that we're 7:15:57 not overleveraged on ads. Same thing 7:15:59 applies into products and then on 7:16:01 commercials this becomes are we making a 7:16:03 balance sheet or a P&L player. And what 7:16:05 I mean by that is that agencies will 7:16:08 always optimize to maximize profit. 7:16:10 Okay, the idea is how do we have the 7:16:11 highest efficiency? How do we have the 7:16:13 highest profit contribution? That is 7:16:15 what they are KPI on. But often in a lot 7:16:17 of businesses, if you do that, if you 7:16:19 overoptimize towards profit, you end up 7:16:21 end up underoptimizing towards the 7:16:24 balance sheet and you end up in a 7:16:25 position where you have a lot of unsold 7:16:27 inventory sitting on the balance sheet 7:16:28 that isn't selling through. And yes, you 7:16:30 are making profit and yes, you are 7:16:32 spitting out cash, but the cash gets 7:16:33 tied up in the balance sheet quite 7:16:35 quickly. And so you could have a P&L 7:16:37 that says $2 million in profit, but then 7:16:40 you have a balance sheet that has $3 7:16:42 million in unsold inventory, and this 7:16:44 business is actually negative $1 million 7:16:46 in cash because all of this cash that 7:16:48 they produced just moved into unsold 7:16:50 inventory. And this is fixable through 7:16:52 paid media through the account 7:16:53 structure. You just need to restructure 7:16:55 the account to sell through this stock. 7:16:57 It's not going to give you the best 7:16:58 return. Your efficiency isn't going to 7:17:00 be that good. You're going to have to 7:17:01 make creatives around this stock as 7:17:02 well, which is going to be a hard pill 7:17:04 to swallow, but it's going to allow you 7:17:05 to get out of this position and 7:17:07 rebalance the business. And this is 7:17:08 ultimately what the media buy needs to 7:17:10 be across. If the media buyer isn't 7:17:11 across inventory, that is not good. 7:17:14 Creative strategy is worthless without 7:17:16 the production. And so, moving into 7:17:18 creative production, teams, briefing, 7:17:20 and the process. I want to start with 7:17:22 the three content types. Number one is 7:17:24 winning content replication. So this is 7:17:27 taking your winners and replicating them 7:17:29 in some capacity. Okay? Whether that's 7:17:31 iterations on the hook, whether this is 7:17:33 translating them into different formats, 7:17:34 you're taking concepts, you're taking 7:17:36 ideas, you're taking creatives that have 7:17:38 already worked in the past and you're 7:17:39 replicating on them. This is by far the 7:17:42 highest impact content types that you 7:17:44 can do across paid as well as organic. 7:17:47 If you have organic content that's 7:17:48 performing well, what most people don't 7:17:50 do enough is just repost the same stuff, 7:17:53 redo the same stuff over and over and 7:17:54 over again because every single time you 7:17:56 do it, it performs just the same, but it 7:17:58 reaches a new audience. So, this is 7:17:59 where a lot of your effort should go. 7:18:01 Then you move into iterative content. 7:18:03 This is trying to fix underperformers 7:18:05 through changes in hooks, changes in 7:18:06 formats. You're working on tangential 7:18:09 creative ideas. So, if you have a 7:18:11 concept that's working, you're making a 7:18:12 slight change to a new persona, or 7:18:14 you're introducing a new offer and 7:18:16 you're trying to get it to work, this is 7:18:17 the second highest priority, prioritized 7:18:20 by leverage. And then down the bottom 7:18:22 here is net new concepts. This is 7:18:25 testing completely new formats, 7:18:26 completely new angles, completely new 7:18:29 concepts. You should be doing this and 7:18:31 you should be allocating time and energy 7:18:33 to it. How much you allocate is based on 7:18:35 the performance up here as well as your 7:18:37 risk profile. What I would not recommend 7:18:39 is some people do 0% here. This is a 7:18:41 terrible idea because all of this will 7:18:43 eventually fatigue and fall off and 7:18:45 you'll be in a very terrible position. 7:18:46 So you need to constantly be finding new 7:18:48 stuff to then flow up to the top of the 7:18:50 funnel here. But I would say that most 7:18:52 people actually overweight towards net 7:18:54 new concepts. I would say that I 7:18:56 actually do this myself in my own 7:18:58 organic content, which is that probably 7:19:00 80 to 90% of the content I put out is 7:19:04 net new. It's new concepts. It's new 7:19:06 ideas. It's new stuff I've never spoken 7:19:08 about before. And 10 to 20% is just the 7:19:10 exact same stuff over and over again. 7:19:11 And what ends up happening is all of the 7:19:13 winning content replication always does 7:19:15 unbelievably well, holds 80% of the 7:19:17 views. And this stuff doesn't really 7:19:18 work that well. It does allow for 7:19:20 continuous winners to be pushed up here, 7:19:23 but this is probably an overallocation. 7:19:25 And so my recommendation generally 7:19:27 speaking, and this is going to change 7:19:29 contextual to the business, is that you 7:19:31 want about 20 to 30% of effort down 7:19:34 here. another 20 to 30% of effort down 7:19:37 here and about 50 to 60% effort up here. 7:19:41 Now, when it comes to briefing in 7:19:43 creators, the key here is that the 7:19:45 quality of the output of the creator is 7:19:47 going to be a product of the quality of 7:19:49 the brief. Okay? So, you want to do two 7:19:51 things when you're sourcing creators. 7:19:53 Number one is you want to check what 7:19:55 they've made in the past. You want 7:19:56 examples, and if you're not doing this, 7:19:58 that's kind of crazy that this is kind 7:20:00 of self-evident, but you want examples 7:20:01 of what they've produced in the past. 7:20:03 But not only that, this is kind of where 7:20:05 the additional source is, is that you go 7:20:07 under that and you get the brief of the 7:20:10 example that they created because that's 7:20:12 going to give you a direct one for one 7:20:14 comparison of how does this creator 7:20:15 produce against a brief. Was the brief 7:20:18 super in-depth and therefore we need to 7:20:19 match that level? Did this creator make 7:20:21 something amazing with a terrible brief? 7:20:23 Okay, then we know that we can give them 7:20:24 relatively broad strokes and they're 7:20:26 going to pull it into something that's 7:20:27 going to be good. Now, generally 7:20:29 speaking, I want the highest quality, 7:20:31 most inep in-depth brief possible. If 7:20:33 they don't want to hold to the script, 7:20:34 that's fine. Shoot one that's held to 7:20:36 the script. Shoot one that feels more 7:20:38 natural. We can choose the best. We can 7:20:40 run them both. Um, but you really want 7:20:42 examples and the briefs associated with 7:20:44 those examples to be able to set 7:20:45 yourself up for success when you're 7:20:47 going and building out creative briefs. 7:20:49 When it comes to building out teams 7:20:51 around production, I've seen a lot of 7:20:52 different teams that all produce 7:20:54 enormous amounts of volume, but there's 7:20:56 a few different key hires that you 7:20:57 should think through when you're 7:20:58 thinking about maximizing volume within 7:21:00 a team. Uh, number one is a video editor 7:21:03 specifically for video ads. Now, if this 7:21:07 person can be AI native or someone that 7:21:09 uh has built a lot of AI creatives in 7:21:11 the past, this is even better because 7:21:12 this just unlocks the ability to make 7:21:14 way more assets and work with way more 7:21:17 footage that doesn't actually exist yet. 7:21:19 Another one is a designer for statics. 7:21:22 Okay, now both of these can be based 7:21:24 anywhere globally. Both of them can be 7:21:27 part-time contractors based on actual 7:21:29 outcome or they can be full-time within 7:21:31 the business. Then it has a third. You 7:21:33 want a creative strategist. Now, this is 7:21:36 a new role that didn't really exist two 7:21:38 to three years ago and therefore there 7:21:40 aren't actually many creative 7:21:41 strategists out there. So, what do you 7:21:43 actually look for when hiring for this 7:21:44 role? Well, you you look for number one, 7:21:48 copywriters, because creative strategy 7:21:50 is primarily copywriting. They are 7:21:53 writing the scripts. They are writing 7:21:54 the brief for the static image. They are 7:21:56 ideulating on new personas. This is 7:21:58 exactly what copywriters have been 7:21:59 doing, except copywriters are now 7:22:01 redundant because LLMs will just produce 7:22:04 most of it. And so these people are 7:22:05 coming up with the ideas and then 7:22:07 they're moving this into a claude skill 7:22:09 which needs to be built by you and brand 7:22:11 purposed to then output an actual script 7:22:14 that then goes into production. The next 7:22:16 option is you just poach a creative 7:22:18 strategist from somewhere else. There's 7:22:19 enough floating around now that you can 7:22:21 probably find one. Or the third is you 7:22:23 actually backfill this role from a 7:22:25 designer or a video editor. And so you 7:22:27 train a designer or a video editor up 7:22:30 into becoming a creative strategist. And 7:22:32 this actually allows for faster feedback 7:22:34 loops because they can ideulate on all 7:22:36 the ideas and then they can design and 7:22:38 then they can obviously brief in for any 7:22:40 video edits. These are really the two 7:22:42 pathways into creative strategist as it 7:22:43 currently stands. Now, in terms of a 7:22:45 side note here, what a lot of people 7:22:46 will do with their creative strategist, 7:22:48 which depending on the scale and volume 7:22:50 that you're at, depends on whether you 7:22:52 want to do this or not, is they will 7:22:53 actually pay out a percentage of ad 7:22:55 spend based on the creatives that they 7:22:58 obviously come up with and then brief 7:23:00 in. And so the creative strategist is 7:23:02 directly incentivized to maximize hit 7:23:04 rates because if their ads don't get 7:23:06 spent and don't perform, they don't get 7:23:07 paid. Typically, this sits at about 3% 7:23:10 of ad spend. So you'll pay out the 7:23:12 creative strategist on top of a base 7:23:13 salary with a percentage variable model 7:23:16 that is obviously directly proportional 7:23:18 to the results within the account. Um, 7:23:20 now I've seen this work really well on 7:23:22 mid-size brands. When you get really 7:23:23 big, this gets outrageous because these 7:23:25 creative strategists end up getting paid 7:23:26 like $80,000 a month. And so then the 7:23:28 model kind of breaks and you end up 7:23:30 moving away from it. A really important 7:23:32 part of this whole component of all 7:23:34 production is where does AI fit into the 7:23:37 mix. When you break down the creative 7:23:40 process end to end, you've got ideiation 7:23:42 as step number one. You then have 7:23:44 briefing. You then move into the actual 7:23:46 production and then you move into the 7:23:49 publishing into the account and 7:23:50 ultimately the analysis process that 7:23:53 feeds back into here on future ideation. 7:23:56 Now, a lot of people are going and 7:23:58 applying AI and they're building Claude 7:24:00 skills for this right here, the ideation 7:24:03 process. This right now, at least at the 7:24:05 time of recording, is not the place 7:24:07 where you want to automate because the 7:24:08 reality is is that all of the LLMs just 7:24:11 output the meme. And so no matter how 7:24:13 much context you provide, no matter how 7:24:15 much documentation you have, no matter 7:24:16 how much you build out your skills to go 7:24:18 and scrape Reddit and all of these 7:24:19 different forums to come up with unique 7:24:21 ideas, ultimately the ideation process 7:24:23 here ends up not being strong enough to 7:24:25 differentiate from competitors. And you 7:24:27 still need a creative strategist to 7:24:28 really do a lot of heavy lifting here to 7:24:30 validate the concepting. When you then 7:24:33 move into briefing and script writing, 7:24:35 this is when this can be heavily 7:24:36 leveraged with AI. That's when you want 7:24:38 to build a claude skill to be able to 7:24:41 walk you through like a 20 questionnaire 7:24:44 about who's your persona, who's the 7:24:45 concept, who's the offer, do you want 7:24:46 this to be long form, what stage of 7:24:48 awareness is it at? And so you're 7:24:50 answering all of these questions and 7:24:52 then ultimately at the end of it, it 7:24:53 pumps you out a script or a brief that's 7:24:56 going to be 99% of the way there and you 7:24:57 just have a 1% tweak. On production, 7:25:00 there's a lot of leverage right now in 7:25:01 AI as well. It's obviously going to 7:25:03 speed up workflows, but in VSSLs and 7:25:05 static image ads as well, you can pretty 7:25:07 much get away with AI for 90% of it. 7:25:10 Then in the analysis process, this is 7:25:12 also where AI becomes a little bit 7:25:14 dangerous because it won't overweight to 7:25:16 the hierarchy of metrics that you're 7:25:18 looking at and won't sit towards the 7:25:20 importance of the account structure and 7:25:21 aligning the context of the creative to 7:25:24 the commercial objectives of the 7:25:26 business. So there is leverage here. You 7:25:28 can use it in analysis. We use it in 7:25:30 analysis. uh but there was so much 7:25:32 context required that it still misses a 7:25:35 lot of the time and if you're using AI 7:25:37 analysis to make core decisions within 7:25:38 the business a lot of the time it 7:25:40 actually misleads you in the wrong 7:25:41 direction and so with AI application I 7:25:44 would try to skip the outsides and 7:25:46 really maximize throughput on the middle 7:25:47 want to make a quick note here on 7:25:49 content first brands which is brands 7:25:52 that are starting with organic content 7:25:54 first before they even touch paid I 7:25:56 think this is the approach for 26 and 27 7:26:00 because it's not due to the fact that 7:26:02 organic content outperforms paid, but 7:26:04 it's due to the fact that organic 7:26:06 content forces you to become good at 7:26:08 creative, which then compounds into paid 7:26:11 when you know how to make content that 7:26:12 people actually want to watch. And so my 7:26:14 recommendation to smaller founders that 7:26:15 are watching this video is that I would 7:26:17 be posting organically every single day. 7:26:20 I would be getting good at figuring out 7:26:21 what good content looks like. See what 7:26:23 actually gains views. Look at the 7:26:24 retention graphs. look at the share and 7:26:26 save rates because this is ultimately 7:26:28 going to improve your ability to make 7:26:30 ads that are actually at a good quality. 7:26:32 Another side note here is that when it 7:26:34 comes to creators, you want to make sure 7:26:36 that they're regional to where you're 7:26:38 actually targeting. So, as an example of 7:26:40 this, in Australia, you want Australian 7:26:43 accents. When you go to the US and you 7:26:44 try to translate an Australian accent 7:26:46 into the market, it just does not 7:26:48 perform as well. So, you need native US 7:26:49 accents. uh UK to AU actually translates 7:26:53 decently well, but still native accents 7:26:55 always outperform. So, you want to make 7:26:57 sure the creatives are native to the 7:26:58 actual region that you're targeting. It 7:27:00 can be the exact same script. It just 7:27:02 has to be a local creator. And then one 7:27:04 final recommendation when it comes to 7:27:06 production is on copywriting. 7:27:07 Copywriting is still one of the highest 7:27:10 leverage skills when it comes to 7:27:11 building out video asset scripts and 7:27:14 also building out static images, 7:27:17 particularly long form copy static 7:27:19 images. There are four books on 7:27:20 copywriting that I have read and I 7:27:23 recommend you read all of them if you're 7:27:24 wanting to become good at creative 7:27:26 strategy and copywriting. Number one is 7:27:27 Breakthrough Advertising by Eugene 7:27:29 Schwarz. Number two is Scientific 7:27:31 Advertising by Claude Hopkins. Number 7:27:33 three is Oglevie on Advertising by David 7:27:36 Oglevie. And then number four is 7:27:38 influence the psychology of persuasion. 7:27:40 If you read those four books, you will 7:27:42 become exponentially better at being 7:27:44 able to actually script out copy that 7:27:46 will convert better through ads. So if 7:27:48 we now go and put it all together, the 7:27:50 foundation is that you want to 7:27:50 understand how the ad platforms work. 7:27:52 Understand that the creative is the 7:27:54 targeting. Concept separation is what's 7:27:56 going to allow you to reach new 7:27:57 audiences. Then when it comes to 7:27:59 concepts, you want to understand how to 7:28:00 split up a concept. Start with personas, 7:28:02 move into angles, then move into the 7:28:04 offer. And as the final component, you 7:28:06 want to bring in the format. Hooks drive 7:28:08 80% of the view through within an ad. So 7:28:10 it's really important to spend time in 7:28:12 optimizing that portion of the ad and 7:28:14 usually hook testing across multiple 7:28:16 different videos. The three components 7:28:18 of a hook is the visual, the copy, and 7:28:20 the audio. You can change any three of 7:28:22 these variables and improve performance. 7:28:24 Then you move into the format. You want 7:28:26 to start with the concepting first, then 7:28:28 always move into formats. Statics are 7:28:30 great for quick testing on new concepts, 7:28:31 and then they can be validated and moved 7:28:33 into videos. Statics also are incredibly 7:28:36 underrated, and most people don't think 7:28:37 they perform well because it's a skill 7:28:38 issue. I have seen static image ads 7:28:40 spend $500,000 in lifetime spend. And 7:28:43 you want to always be careful on DPAs. 7:28:44 DPA are good, but they generally 7:28:46 overspend. and the rorowaz number over 7:28:47 inflates them. When it comes to testing, 7:28:49 you always want to be testing one 7:28:50 variable at a time. You want amount 7:28:52 spent to be the best proxy for 7:28:54 performance, then rorowaz and the 7:28:56 tertiary metrics. And the actual 7:28:58 structure of the ad account should be a 7:28:59 product of the complexity of the 7:29:01 business. When it comes to thinking 7:29:02 through volume verse quality, you want 7:29:04 as much volume as possible as long as it 7:29:06 meets the bar of quality that we would 7:29:08 classify it as a good ad. If it's a good 7:29:10 creative, no one will be able to guess 7:29:12 which one will win. So, put them all 7:29:13 into the account. When ads fatigue, it's 7:29:15 because you're either trying to push too 7:29:17 much spend through a niche audience and 7:29:19 it's driving up frequency, or the ad has 7:29:21 simply reached this lifetime spend, it's 7:29:24 falling off, and you need to diversify 7:29:26 by thinking through portfolio management 7:29:28 risk. When it comes to production, we 7:29:29 ran through all of the relevant team 7:29:31 structure that you should think through 7:29:33 as well as the fact that you want to be 7:29:34 putting majority of your production cost 7:29:36 towards iteration rather than net new 7:29:38 concept design because that's going to 7:29:40 drive most of the performance. And then 7:29:41 if we do one final breakdown of spend 7:29:44 per month and the recommendations on how 7:29:47 you should approach creative at 4 to 30k 7:29:49 a month, the founders creating the 7:29:51 content, it can be done on an iPhone. 7:29:52 You want it to come across natural. All 7:29:54 the leverage is going to be in how 7:29:55 strong the scripts are. You want to be 7:29:56 working on three core concepts. You 7:29:58 really don't need more than that at this 7:30:00 scale. You just want to do these well. 7:30:02 You want a high testing allocation. 7:30:04 Realistically, 60 to 80% of your 7:30:05 budget's going to go towards testing 7:30:07 because you just don't have enough spend 7:30:08 yet to even scale winners. And then you 7:30:11 can use trial reels as well as a hack 7:30:12 for validating ideas. Once you move into 7:30:15 30 to 100k per month, you want two to 7:30:17 three UGC creators. 25% of this spend 7:30:20 needs to be going towards production. 7:30:22 You're going to move into five to six 7:30:24 concepts. You want to start branching 7:30:26 out and d-risking. You're going to have 7:30:27 an and then likely a scaling 7:30:29 campaign. You're sitting at 30 to 100 7:30:31 new creatives per month here or you're 7:30:33 using our creative calculator and it'll 7:30:35 give you an exact amount. And then as a 7:30:37 little hack at the bottom, you want to 7:30:38 be introducing partnership ads. And then 7:30:40 if you're at over 100K per month in ad 7:30:42 spend, you really want a full creative 7:30:44 team. Someone who's doing scripting. You 7:30:46 want an editor. You want four to six 7:30:48 creators. And you want a designer. Now, 7:30:51 these could be part-time, but you want 7:30:53 these resources in place so that you 7:30:55 have the production machine in place to 7:30:57 churn out creative. That's really what 7:30:58 you want to be looking for at this ad 7:30:59 spend level is you need a system that 7:31:01 consistently produces creative on an 7:31:03 ongoing basis. If that isn't in place, 7:31:06 you're just constantly going to be 7:31:07 scrambling and creative will be the 7:31:08 bottleneck. You want at least 10% as a 7:31:11 minimum budget going towards creative at 7:31:12 this spend level. You want 8 to 10 7:31:14 active concepts within the account. You 7:31:16 want at least 100 new assets per month 7:31:18 based on the basic 1k per ad formula. 7:31:21 And you want a financial model in place 7:31:23 that's tracking your cost per asset, 7:31:25 ideally across the different sources. 7:31:27 So, how much is these creator pieces 7:31:29 costing you? And then what is the 7:31:31 average profit contribution per ad on 7:31:33 these creators? So we can start to see 7:31:35 if the cost per asset of these different 7:31:37 content types is actually worth it. And 7:31:39 you want to be tracking ROI at an 7:31:41 individual ad level based on ad spend 7:31:43 against the cost of production. For most 7:31:45 people right now, creative is the 7:31:47 constraint on meta. You don't know what 7:31:49 good looks like until you start doing 7:31:51 volume, building out concepts, 7:31:53 scripting, getting ads in the account, 7:31:54 and beginning that feedback loop. So, as 7:31:56 a wrap-up, if you're a performance 7:31:58 marketer and you have at least 2 years 7:31:59 experience, please reach out to us in 7:32:01 the description below. We're always 7:32:02 hiring. And if you're an e-commerce 7:32:03 brand or a retail business doing at 7:32:05 least $5 to $10 million a year at a 7:32:07 minimum, you can also click below or 7:32:09 reach out to us and we'll do a 7:32:10 completely free audit where everything 7:32:12 that you've learned in this video. We 7:32:13 will actually translate directly into 7:32:15 the ads that you're running directly 7:32:16 into your ad account. We'll pick it 7:32:17 apart so that you can know how to unlock 7:32:19 the current bottleneck that's stopping 7:32:21 you from growing. And also, please 7:32:22 subscribe. Most of the Google Ads 7:32:24 playbooks running inside agencies right 7:32:25 now are 2022 2023 strategies. They 7:32:28 haven't been updated for the massive 7:32:30 platform changes that have occurred over 7:32:32 the course of the last 2 to 3 years. 7:32:33 Almost every part of the platform has 7:32:36 shifted. Google tries to ship a new 7:32:38 product every 3 to 4 months and it 7:32:39 constantly changes the landscape of how 7:32:41 you should be approaching the platform. 7:32:43 In this video, we're going to be 7:32:44 breaking down how the platform works in 7:32:46 2026, what structure and segmentation 7:32:49 should look like in your account. We'll 7:32:51 go through PMAX. We'll go through where 7:32:52 standard shopping still has a place in 7:32:54 2026. We'll talk about smart bidding and 7:32:56 bidding strategies. We'll go into the 7:32:57 GMC feed and we'll rank everything that 7:32:59 matters in the order of importance. And 7:33:02 then finally, we'll give you a 90-day 7:33:04 playbook at the end, which is a 90-day 7:33:06 plan that you can roll out on your 7:33:08 Google account to be able to transition 7:33:10 it into a position that's fit for 2026. 7:33:13 The quick commercial frame of Google ads 7:33:15 that's really important to understand is 7:33:17 that almost everyone watching this video 7:33:19 is either overspending or underspending. 7:33:21 Now the overspend is very common and it 7:33:24 comes from the fact that Google sits 7:33:26 very bottom of funnel and all of the top 7:33:28 offunnel generation in the business is 7:33:30 actually coming from platforms like 7:33:31 Meta, Tik Tok, affiliates etc. And then 7:33:34 Google just captures all of the bottom 7:33:36 intent demand and then claims a really 7:33:38 high rorowass. And so as a function of 7:33:40 that, a lot of people just push up spend 7:33:42 on Google despite it not being that 7:33:44 incremental in the business and they end 7:33:46 up with overspend. Then there's the 7:33:48 other side of accounts where there are 7:33:50 people that understand this idea. In 7:33:52 fact, they're very indoctrinated into it 7:33:54 and they see Google ads just as a 7:33:55 bottomfunnel platform and that's all it 7:33:57 is. And so as a product of that, they 7:33:59 don't take Google spend over $10,000 a 7:34:02 month despite them spending a4 million a 7:34:04 month on Meta because the logic is well 7:34:06 yeah Google's bottom up funnel. I don't 7:34:08 want to scale it. Everyone overspends 7:34:09 here. Why would I spend more? But 7:34:12 they're in a category that is very 7:34:15 scalable on Google. There is tons of 7:34:17 cold traffic every single day searching 7:34:19 for relevant key terms that their 7:34:21 product could serve under. But because 7:34:23 they don't trust the platform, which 7:34:25 fair enough, they don't go and spend on 7:34:26 it and they don't go and maximize it. 7:34:28 So, it's really important to understand 7:34:29 that you're going to be in one of these 7:34:30 two camps. And by the end of this video, 7:34:32 you'll understand which camp you're in 7:34:34 and where you need to start moving spend 7:34:36 and structuring spend in the account. 7:34:38 There's really been five big shifts over 7:34:40 the course of the last few years in 7:34:42 Google that you need to be across. 7:34:43 Number one is that Pmax hit its ceiling. 7:34:46 So, when Pmax first launched, it was a 7:34:48 pretty promising campaign. In fact, 7:34:50 Google was likely giving discounts at 7:34:52 auction so that people would preference 7:34:54 into the new product. And so when you 7:34:55 used Pmax, you got ridiculous results. 7:34:57 We saw some crazy results on accounts 3 7:34:59 4 years ago. Then everyone started to 7:35:01 weight all their spend into Pmax and the 7:35:03 arbitrage opportunity there kind of went 7:35:05 away. You also have no clear visibility 7:35:07 into where spend is going. And then what 7:35:09 you end up finding is that as we scaled 7:35:12 performance max campaigns on a lot of 7:35:13 large accounts to 100 200 $300,000 a 7:35:16 month in ad spend is that we didn't 7:35:17 start to see the returns show up on 7:35:20 backend revenue. And this is where we 7:35:21 started to dive into the true 7:35:22 incrementality of the platform to 7:35:24 understand, okay, even though Pmax says 7:35:27 that we're still at a five rorowaz and 7:35:29 we've doubled budget, why hasn't backend 7:35:31 revenue moved? And that became a real 7:35:33 core testing framework for us to start 7:35:36 thinking through redistributing budgets 7:35:38 out of PMAX into standard shopping. 7:35:41 Standard shopping pretty much was phased 7:35:43 out of a lot of accounts uh when PMAX 7:35:46 started to first arise, but now standard 7:35:48 shopping is back and it's because once 7:35:52 you scale Pmax far enough, you just see 7:35:54 diminishing returns and you can't scale 7:35:55 it any further. And so if you want to 7:35:57 continue to grow cold new customer 7:35:59 acquisition on the platform, it will 7:36:01 require standard shopping being layered 7:36:03 into the structure. Number three is 7:36:05 match type definitions have changed. 7:36:08 Exact match in search campaigns isn't 7:36:10 exact match anymore. It now matches to 7:36:12 the same meaning. And so there's 7:36:14 actually flexibility. Phrase match got a 7:36:16 little bit broader as well. And then 7:36:17 broad match is just as broad as it's 7:36:19 always been. And so this affects a lot 7:36:21 of the old strategies that people used 7:36:23 to do. For example, if you've been 7:36:24 running Google ads for a while or you 7:36:25 might still have this in your account, 7:36:27 SCAGs was a really common strategy back 7:36:29 in the day where you would have a single 7:36:31 keyword ad group for every keyword and 7:36:33 you would monitor at an ad group level. 7:36:34 I wouldn't recommend running this 7:36:36 anymore. In fact, this stopped working 7:36:37 in like 2020. Now, you might have an 7:36:39 account where it works and therefore 7:36:41 you've just continued to run it, but 7:36:42 there's a better way to do things. And 7:36:43 so, we'll dive into that later on when 7:36:45 we talk to search campaigns. Number four 7:36:47 is that Google has introduced AIMAX. 7:36:50 They've also introduced demand genen. 7:36:52 And then number five is DSAs are being 7:36:54 deprecated. So, on account structure, 7:36:57 first rule of account structure, 7:36:58 consolidation will always beat 7:37:00 segmentation. And this is a product of 7:37:02 how the account structure works in 7:37:04 segmenting and siloing data. So let's 7:37:07 say you have three campaigns. This is 7:37:09 three shopping campaigns. And let's say 7:37:10 it's all the same products as well. 7:37:12 You're just testing different bidding 7:37:13 strategies or you have some kind of 7:37:14 strategy going on here. What happens is 7:37:16 that this top campaign maybe has 50 7:37:18 conversions every 30 days. This next one 7:37:21 has 20 and then this next one has 10. 7:37:24 Now the issue here is that all of these 7:37:27 campaigns learn in isolation. Now there 7:37:29 is a little bit of platform level data 7:37:32 sharing but just ignore it for now to 7:37:33 simplify the model down and just imagine 7:37:35 they are completely siloed out. So these 7:37:38 50 conversions are teaching this 7:37:39 campaign who to optimize for but they're 7:37:41 not getting shared with this campaign 7:37:43 and it's not getting shared with this 7:37:44 campaign. So they're isolated out. What 7:37:48 that then means inherently is that this 7:37:50 campaign down the bottom here with only 7:37:52 10 conversions isn't going to perform 7:37:54 very well because it's modeling off 10 7:37:57 conversions. It's using the 7:37:58 psychographic data points of just 10 7:38:00 users to decide how to adjust bids in 7:38:03 real time on future search terms, which 7:38:05 is crazy. And so because of this, this 7:38:07 campaign inherently won't perform that 7:38:09 well. And you'll often end up finding 7:38:11 this when you see a cam account 7:38:13 structure that looks something like this 7:38:14 is that as you go down in the conversion 7:38:17 volume, you also go down in return on 7:38:19 ads. And that is usually causally 7:38:21 related to the volume going through the 7:38:23 campaign. And so how we could 7:38:25 immediately get better performance in an 7:38:27 account like this is we simply delete 7:38:30 campaign B. We delete campaign C and we 7:38:33 move all the products and the budget up 7:38:36 to here. And now in the top campaign we 7:38:39 have 90 conversions per month rolling 7:38:42 through. Because this is such a larger 7:38:44 data set, it will be able to model to 7:38:47 users better, create more accurate bids 7:38:50 at the auction, and therefore produce a 7:38:52 higher return on ad spend than if we 7:38:54 were to segment out. And so the question 7:38:56 becomes, why do we even segment out? Why 7:38:58 don't we just always run like one PMAX 7:39:01 campaign and that's it? Because that's 7:39:03 how we'll see the best ROI. And the 7:39:05 reason being is that unfortunately, most 7:39:07 businesses aren't that simple. And the 7:39:08 complexity of the business needs to 7:39:09 match the complexity of the ad account. 7:39:11 And so if you have as an example here 7:39:14 three different product categories, one 7:39:16 of them is grade C inventory that we 7:39:18 either have to burn discount by 80% or 7:39:21 give away. Well, do we want that getting 7:39:23 grouped into this single campaign and 7:39:25 getting no ad spend? No. We probably 7:39:26 want a little bit of spend behind those 7:39:28 products so that we can turn them over 7:39:30 so we don't have to put them on heavy 7:39:32 discount. Therefore, we launch a grade C 7:39:35 inventory campaign here that doesn't do 7:39:38 very well. It doesn't hold many 7:39:40 conversions, but we don't care cuz at 7:39:41 least it's just moving over a little bit 7:39:43 of stock. We then also might have new 7:39:46 arrivals that are constantly coming in. 7:39:48 Now, we want that surfaced right at the 7:39:50 top of Google Shopping because we want 7:39:52 to not only turn over our new arrivals 7:39:54 quickly, but we want the seasonality of 7:39:57 our ranging to be relevant to today. We 7:39:59 don't want super old products that are 7:40:00 no longer relevant to the season to be 7:40:03 popping up on Google Shopping. So, as a 7:40:05 product of that, we have a new arrivals 7:40:07 campaign. And then lastly, we have the 7:40:09 core campaign at the top here that maybe 7:40:12 we started with when we were a smaller 7:40:14 business. And then as we've scaled and 7:40:15 grown, this has just grown in ad spend. 7:40:17 And this is why the account is 7:40:19 structured in this way despite it not 7:40:22 meeting the rule of consolidation over 7:40:23 segmentation. And so the real key here 7:40:26 is that you always want to be as 7:40:28 consolidated as possible whilst 7:40:32 introducing commercial realities into 7:40:35 the attic. This is fundamentally why an 7:40:38 agency needs to intricately understand 7:40:41 your business down to the grade C 7:40:43 inventory level or else they will just 7:40:45 segment with no reason behind it and 7:40:48 just decrease performance for the sake 7:40:50 of adding complexity into the ad account 7:40:52 or they will do the opposite and they 7:40:54 will consolidate up under this rule. 7:40:57 However, you would actually be able to 7:40:59 better align the ad account with the 7:41:01 commercial objectives of the business if 7:41:03 you did have a little bit of 7:41:04 segmentation. So that is how you need to 7:41:05 be thinking through this problem within 7:41:07 the account. A good way to think through 7:41:09 this problem is that the goal is always 7:41:12 one campaign that's never realistic but 7:41:14 it's the default and then every 7:41:16 additional campaign that you layer into 7:41:18 the account needs heavy justification. 7:41:20 Now the justifications that count is 7:41:24 number one brand versus non-brand. So 7:41:26 you don't want your one Pmax campaign to 7:41:30 have brand searching. So, if you want 7:41:31 placements on brands, but then you also 7:41:33 want cold, this needs to be two separate 7:41:35 campaigns. And I would always recommend 7:41:37 this. Number two is geography. So, if 7:41:40 you're selling in multiple different 7:41:41 countries, that should always get 7:41:43 segmented out at a campaign level. 7:41:45 Number three is different margins or 7:41:48 efficiency goals. A 70% 7:41:51 collection in terms of gross margin 7:41:54 cannot be bundled with another 7:41:56 collection that has 35% gross margin. 7:41:58 Because if these share the same bidding 7:42:00 strategy, let's say you're just running 7:42:01 a 300% target rorowaz. Well, this is 7:42:05 like barely profitable at that rorowaz. 7:42:08 This is incredibly profitable at that 7:42:10 rorowaz. Now, because Google, at least 7:42:12 with most people's setups, doesn't have 7:42:15 visibility into the margin profile. It 7:42:17 will just optimize towards revenue. And 7:42:20 if this is generating more revenue, this 7:42:22 is where all your spend goes despite it 7:42:24 not even being profitable. And so you 7:42:25 need a degree of segmentation in the 7:42:27 account based on the margin profile. And 7:42:29 then number four is product categories. 7:42:32 And the reason being is that you can 7:42:33 have very very different products and 7:42:35 they don't belong together and they 7:42:36 don't belong learning together as well 7:42:38 in the account. As a super extreme 7:42:40 example, let's say that you sell office 7:42:42 chairs and then you also sell plants. 7:42:45 Now yes, you can buy plants for an 7:42:47 office. They have some kind of 7:42:48 similarity and that's probably why 7:42:50 you're selling both products. But the 7:42:51 reality is plants are going to learn 7:42:54 towards a very different audience than 7:42:56 your office chairs are. And so as a 7:42:57 product of that you do want the data 7:42:59 segmented so they can learn in isolation 7:43:01 to their own unique persona. Now if we 7:43:03 quickly talk about single keyword ad 7:43:05 groups you do not want to be putting one 7:43:07 single keyword into an ad group anymore. 7:43:10 It does not play to the way that the 7:43:12 accounts want to be structured by 7:43:13 Google. Now, Google and Facebook say a 7:43:15 lot of stuff because they want you to 7:43:16 spend a lot of money and their advice 7:43:19 isn't necessarily what you should be 7:43:20 doing 80% of the time, but there is 7:43:23 method in 20% of the stuff that they 7:43:25 say. If they say consolidating keywords 7:43:28 down is beneficial and then you test it 7:43:30 and it works well or we can be pretty 7:43:32 confident that consolidating keywords 7:43:34 down is beneficial. And that is what we 7:43:36 see constantly across all the accounts 7:43:38 that we work on. We have still tried 7:43:40 SKAGs in fact because we onboard clients 7:43:42 that have this running in the account 7:43:44 and then we transition over to single 7:43:46 topic ad groups instead and we always 7:43:49 see better performance. So let me give 7:43:50 you an example and I won't give an 7:43:52 e-commerce example. I'll give a 7:43:54 service-based example as search 7:43:56 campaigns are way more uh relevant to 7:43:59 service-based businesses. Quick side 7:44:00 note here. If you're an e-commerce 7:44:02 brand, we only work with e-commerce 7:44:03 brands. All the content is about 7:44:04 e-commerce brands. video is really 7:44:06 orientated towards e-com and retail 7:44:08 businesses. In e-com and retail, you 7:44:10 really want 80 to 90% of your spend to 7:44:12 be in shopping, not in search. And the 7:44:14 reason for that is that you always end 7:44:15 up with higher quality traffic and 7:44:17 higher intent audiences on the website. 7:44:19 It is due to the pre-click information. 7:44:22 When you look at a search ad, what do 7:44:24 you get? You get a headline, you get a 7:44:26 description that no one reads, and then 7:44:28 you usually get a couple extensions that 7:44:29 can help. So, if I go and search for buy 7:44:32 red dress, size medium for wedding, I 7:44:36 get a search ad and it says red dress 7:44:38 for wedding size medium. Now, it matches 7:44:41 my intent. It tells me what it's going 7:44:42 to be, but I have no idea what the price 7:44:44 is. I have no idea what it looks like. I 7:44:46 have no idea what the other options are. 7:44:47 I'm very limited. And so, I'm trusting 7:44:49 that when I click on that search ad and 7:44:51 charge the advertiser $1 to $2, that 7:44:53 whatever's going to be on the other side 7:44:54 is what I actually desire. And in 7:44:56 fashion particularly, it is very visual 7:44:58 based. And so that is generally not a 7:44:59 good idea. It's not a good idea to just 7:45:01 run ads and say, "Hey, we sell dresses." 7:45:03 Because you'll end up with a lot of 7:45:04 people coming in the door and a lot of 7:45:05 people go, "But this isn't the style I 7:45:06 like." The great thing about shopping 7:45:08 ads is you show the user an image of 7:45:11 what the product looks like, the price, 7:45:13 the title, potentially any promos or 7:45:15 reviews that are currently active, the 7:45:17 brand name, and then any other 7:45:19 extensions that drop down. On top of all 7:45:21 of that, five of your competitors pop up 7:45:23 right next to you. So the user does a 7:45:25 comparison across six listings and 7:45:26 chooses the one that's most relevant to 7:45:28 them. So they're already pre-bought into 7:45:31 your listing being the best. Now on 7:45:32 search, there's almost no information. 7:45:34 So you get worse conversion rates, which 7:45:36 is why shopping will always outperform 7:45:38 search in pretty much any e-commerce 7:45:40 business up to the point until you 7:45:42 completely saturate the shopping network 7:45:44 and then you go into search for 7:45:46 additional volume. Back to single topic 7:45:48 ad groups. What this looks like is we 7:45:49 have three ad groups. We have general 7:45:51 plumbing, emergency plumbing, and then 7:45:53 affordable plumbing. Where this 7:45:55 previously on a single keyword ad group 7:45:57 set up, you would just have phrase 7:45:59 matcher, exact match for these key 7:46:00 terms, and that's it. This whole ad 7:46:02 group is just bidding on the key term 7:46:04 general plumbing. Instead, this is a 7:46:06 topic. And so under this sits 10 to 20 7:46:11 themed keywords. So all keywords that 7:46:13 are relevant and themed to this topic. 7:46:15 And then the same thing for emergency. 7:46:17 And then the same thing for affordable 7:46:18 plumbing. And then the ads here are 7:46:20 tuned to whatever that theme is. So 7:46:22 they're contextually relevant to all the 7:46:24 keywords that we're bidding on. I'll 7:46:25 tell you why hyper segmentation within 7:46:28 Google doesn't work anymore. And why 7:46:29 scaggs has turned into stag, which is I 7:46:32 used to run scags back in 2019. This was 7:46:34 actually the first Google campaign I 7:46:37 ever launched on the platform was 7:46:38 running a SCAG campaign. So I'm very 7:46:40 familiar with it. The reason why this 7:46:41 used to work and for new performance 7:46:43 marketers or new people running brands, 7:46:45 you won't even remember this, but ads in 7:46:47 Google didn't used to be responsive ads 7:46:50 where you put in 15 headlines and four 7:46:52 descriptions and tons of images and then 7:46:53 it will just dynamically adjust what 7:46:55 headlines serve to the user. Instead, 7:46:57 back in the day, you would provide just 7:47:00 three headlines, just one or maybe two 7:47:02 descriptions, and that was it. And there 7:47:04 was no ranking or sorting them around. 7:47:06 You would choose which one's the first 7:47:08 headline, which is the second, which is 7:47:09 the third and then that is the ad that 7:47:11 served. So you would choose exactly what 7:47:12 the ad looked like. You would choose the 7:47:14 exact keyword it would place on and that 7:47:16 would be that makes total sense, right? 7:47:18 We want our ads to be contextually 7:47:19 relevant to the exact keyword that we're 7:47:21 placing on. And that's why SCAGs 7:47:22 existed. That's why it did well. But now 7:47:25 that's not how it works. Now you create 7:47:27 a responsive ad. And a responsive ad has 7:47:30 15 headlines. It has four descriptions. 7:47:33 It has a bunch of other asset variations 7:47:35 that you put in there. And because of 7:47:36 that, there is 2,700 7:47:40 possible combinations of this one ad 7:47:43 that can serve to a user. And so Google 7:47:45 is going and split testing all of these 7:47:47 headlines, all of these descriptions, 7:47:48 all of these assets against each other 7:47:50 in different combinations to try to 7:47:51 figure out which of these 2,700 7:47:54 combinations is going to perform the 7:47:55 best. Now, to properly test this, you 7:47:57 realistically need to serve this ad 7:47:59 against about a quarter million 7:48:01 impression. Now, are you going to get a 7:48:03 quarter million impressions on this ad 7:48:05 if you're just bidding on one keyword 7:48:07 with exact match in a SCAG setup? 7:48:09 Absolutely not. And so, the only way to 7:48:11 get out of learning and to get the 7:48:13 campaign understanding what combination 7:48:15 of an ad should be served to a 7:48:17 particular audience is through higher 7:48:19 volume getting put through at an ad 7:48:21 group and an ad level. The added benefit 7:48:23 too is that because this is dynamic, it 7:48:26 can just dynamically adjust based on the 7:48:28 different keywords. So, we don't need to 7:48:30 set up one ad for this keyword, one ad 7:48:32 for this keyword. We just set up a large 7:48:34 ad and then Google dynamically figures 7:48:36 out what headlines and descriptions work 7:48:38 based on which keyword it is within the 7:48:40 themed group. A quick note on account 7:48:42 hygiene, you always want a proper naming 7:48:44 convention set up just so someone can 7:48:46 easily jump in the account, have 7:48:47 context, understand what's going on. The 7:48:49 naming convention that we like to use is 7:48:51 obviously the agency tag at the start so 7:48:53 we understand what campaigns we launched 7:48:55 and what anyone else might have 7:48:56 launched. the region, campaign type, is 7:48:58 it a PMAX, is it a shopping, is it a 7:49:00 search? The audience, is this a cold 7:49:02 campaign, is this a brand campaign, is 7:49:04 this a theta campaign? And then the goal 7:49:07 at the end, which normally we'll 7:49:08 actually just drop this off cuz the goal 7:49:10 in 99% of the case is just sales. It's 7:49:12 conversions. But in some cases where 7:49:14 we're doing some different campaign 7:49:15 testing, uh, this might change out and 7:49:17 therefore we'll specify. I now just want 7:49:19 to rapidfire you a bunch of mistakes 7:49:21 that I see in Google accounts all the 7:49:22 time. Before we dive into the large 7:49:24 section of the video where we start 7:49:25 breaking down account structure, 7:49:27 performance max campaign, shopping, etc. 7:49:29 So, number one is G4 being used as the 7:49:31 primary conversion. You shouldn't use G4 7:49:34 events in the account because there's 7:49:36 about a 10 to 15% sometimes even higher 7:49:38 attribution gap. So, it misses about 10 7:49:40 to 15% of the conversions that would get 7:49:42 tracked if you just had direct snippet 7:49:44 code installed on the website. Number 7:49:45 two is just running one asset group on a 7:49:47 performance max campaign. We'll go into 7:49:49 why that's a mistake later on, but 7:49:51 always make sure that you have a degree 7:49:52 of segmentation within the PMAX under 7:49:54 the asset group level. Next one is 7:49:56 running broad match with the maximize 7:49:59 clicks bidding strategy. Reason being is 7:50:01 that broad match requires a smart 7:50:03 bidding strategy to actually work well. 7:50:05 So when you're using a non-smart bidding 7:50:06 strategy on broad match doesn't work. 7:50:08 Number four is display being enabled in 7:50:11 search campaigns. this is a big rookie 7:50:14 error and you're just wasting budget on 7:50:15 the display network which is arguably 7:50:17 one of the worst advertising networks 7:50:19 that exist like you don't want spend 7:50:21 going there particularly on cold. The 7:50:23 next one is no negative keywords. This 7:50:25 applies to no branded negative keywords 7:50:27 on cold campaigns but also no negative 7:50:30 keywords in general on your search 7:50:32 campaigns as well. The next one is the 7:50:34 auto apply recommendations 7:50:36 are left on on the account. So, Google 7:50:39 is just randomly going and applying 7:50:41 stuff into the account that almost 7:50:42 always is not in your best interest. So, 7:50:44 they're the main mistakes to look out 7:50:45 for in your account. You're probably 7:50:47 making one of these mistakes right now. 7:50:48 All right, so let's talk about 7:50:50 performance max. We're going to cover 7:50:51 the mechanics, the traps that you'll 7:50:53 fall into, rules for segmenting asset 7:50:56 groups, brand exclusion lists, search 7:50:58 themes, audience signals, and the 7:51:00 conditions in which we have performance 7:51:02 max in a campaign versus when we don't. 7:51:03 It's worth noting that Pmax is not the 7:51:05 enemy. If you go back really far on the 7:51:07 BlueSense YouTube channel, you'll find a 7:51:09 video that I put out called something 7:51:10 like Pmax campaigns are terrible. And 7:51:12 that was a bit of a biased view that I 7:51:13 had at the time, but it was true and it 7:51:14 was contextual to the moment in which 7:51:15 that video was made, which is that we 7:51:17 were testing Pmax a lot on accounts. And 7:51:20 what we were finding is that they 7:51:21 weren't as incremental as shopping 7:51:23 campaigns when you read the rorowaz 7:51:25 number. And so if you went into an 7:51:28 account and you saw Pmax was at a 5x and 7:51:31 let's say you saw that shopping sitting 7:51:34 next to it was also at a 5x. The test 7:51:37 that we were running at the time was 7:51:39 what happens if we go and scale this 7:51:40 Pmax campaign by 2x over the course of a 7:51:43 30-day period. So we ramp it up. And 7:51:45 then on another account, what happens if 7:51:47 we ramp up the standard shopping? And 7:51:49 what we saw every single time we ran the 7:51:51 test was that the incremental revenue 7:51:53 return on the back end of the website 7:51:55 was way better on the standard shopping 7:51:57 ramp. And so we at the time put 7:51:58 ourselves into this perspective and view 7:52:00 that Pmax is just a bad campaign type. 7:52:03 It just doesn't scale anywhere near as 7:52:04 well on cold audiences despite what the 7:52:07 inplatform numbers say. Because the 7:52:09 caveat of this whole thing is that when 7:52:10 you go and scale the PMAX campaign, the 7:52:12 row still says five. You go and scale 7:52:14 the standard shopping, it drops. It goes 7:52:16 down to like a four and a three. But the 7:52:18 actual back-end revenue is better in 7:52:20 this circumstance. And the reason why 7:52:22 that occurs is that Pmax because it 7:52:24 retargets a lot, it ends up 7:52:26 overcrediting just for more purchases as 7:52:28 you put more and more spend through it 7:52:29 rather than truly driving new customer 7:52:31 acquisition in the business. Whereas the 7:52:32 standard shopping campaign drives very 7:52:34 top ofunnel traffic that then might not 7:52:36 get attributed to standard shopping. It 7:52:38 might go and get attributed into PMAX 7:52:40 because it goes and retargets, follows 7:52:41 up and gets the final click before the 7:52:43 conversion. And so this is really the 7:52:45 dichotomy of what Pmax was like within 7:52:47 the first 1 to two years of launch. But 7:52:49 I will say that Pmax isn't the enemy. 7:52:51 PMAX with the default settings that it 7:52:54 has when you launch it for sure is. But 7:52:56 if you set PMAX up with the correct 7:52:58 asset group structure, you have brand 7:53:00 exclusions in place, you understand that 7:53:02 audience signals are used for signals, 7:53:04 not for targeting, and you use the feed 7:53:06 only strategy where it's appropriate in 7:53:08 the business, then it's a perfectly good 7:53:09 tool to have within the structure. 7:53:11 However, caveats are Pmax is a great 7:53:14 tool for smaller accounts. If you're 7:53:16 spending like 15 to 20k a month or let's 7:53:19 say even less than this, Pmax is good. 7:53:21 It'll get the account moving. It'll get 7:53:23 it scaling to this level. But quite 7:53:25 honestly, Pmax does hit a ceiling. You 7:53:27 scale and scale and scale it and then 7:53:29 eventually you can't push it past a 7:53:30 certain point. At least you can't push 7:53:32 it past a point where incremental 7:53:33 returns on the back end of the website 7:53:35 actually follow what the platform's 7:53:36 telling you. And the real big issue here 7:53:37 is that once PMAX does plateau, you go 7:53:40 and scale and scale it up and then you 7:53:42 find this point in which you can't scale 7:53:43 it any further. The natural response is 7:53:45 either well let's just raise budgets 7:53:47 even further and see what happens and 7:53:49 just let it run for a longer period or 7:53:51 let's raise the target return on ad 7:53:53 spend on the campaign to try to get it 7:53:55 to operate at a higher efficiency here 7:53:57 so then we can afford to put more budget 7:53:59 in. But the issue is both of those moves 7:54:01 just pushes more spend into warmth and 7:54:04 repeat traffic and the PMX rows might go 7:54:06 up but it just becomes a glorified 7:54:08 retargeting campaign. I would start to 7:54:10 reframe the question on Pmax away from 7:54:12 should I run performance max campaigns 7:54:14 to instead what percentage allocation 7:54:17 should I have to Pax versus standard 7:54:18 shopping because the answer for most mid 7:54:20 to upper market ecom and retail 7:54:22 businesses is not a lot to Pmax 20 30 7:54:25 40% maybe but a lot of spend should be 7:54:27 driven directly through standard 7:54:28 shopping and if you're a lead genen 7:54:29 business watching this every time I say 7:54:31 shopping just imagine I'm talking about 7:54:32 search so the single biggest mistake 7:54:34 that's made in Pmax is just running one 7:54:37 asset group when you go and set up the 7:54:38 campaign It just by default creates one 7:54:40 asset group. And so if you're lazy, you 7:54:42 just click through and you don't 7:54:43 actually set up multiple. An even bigger 7:54:45 mistake is that people will set up 7:54:47 multiple asset groups, but they'll 7:54:49 forget to customize the listing group. 7:54:52 And so you just have all the products 7:54:53 bundled into every asset group anyway, 7:54:54 and it doesn't even matter. So what do 7:54:56 these actually mean? If you've never 7:54:57 even opened up Google Ads before, or 7:54:58 you're a founder trying to understand 7:55:00 this better, so you can have more 7:55:01 in-depth conversations with your agency. 7:55:03 The asset group is effectively the same 7:55:06 thing as different ads in Meta or 7:55:09 different ads within a search or 7:55:13 shopping campaign. When we talk about 7:55:14 shopping campaigns though, we 7:55:15 technically don't have different ads. We 7:55:18 have different products that come 7:55:20 through from the feed. And so the 7:55:22 product feed within Pmax is called 7:55:25 listing groups. And so this is 7:55:27 effectively how we are bundling all of 7:55:30 the products that are flowing through 7:55:32 from the GMC feed. With each asset 7:55:34 group, we have a connected listing 7:55:36 group. So the asset group has all of the 7:55:39 assets, so the images, the headlines, 7:55:41 the descriptions, etc. All of this 7:55:43 information provided in the asset group 7:55:46 allows the campaign to go and take the 7:55:47 headlines, titles, descriptions and 7:55:49 serve onto search, serve onto display, 7:55:52 serve onto YouTube as the little side 7:55:54 banners, serve on to discover, serve 7:55:57 onto Gmail, and any other random 7:55:59 placements that it wants to go and put 7:56:00 you on, it will use these assets to do 7:56:02 so. So, this constitutes a large portion 7:56:05 of what your ads actually look like. But 7:56:06 if you're an e-commerce brand and you 7:56:08 were listening earlier, majority of your 7:56:10 performance should be coming from 7:56:12 shopping. And so to get placements onto 7:56:14 shopping, it uses the listing groups and 7:56:17 just places these over onto shopping. 7:56:19 Now, it's really important that when you 7:56:20 build out multiple asset groups, you 7:56:23 segment the listing group accordingly 7:56:25 based on how it's structured. Now, the 7:56:27 question you should be asking is, wait, 7:56:29 why are we segmenting at an asset group 7:56:32 level when the number one rule of 7:56:33 account structure on every single 7:56:35 platform is consolidation beats 7:56:37 segmentation? Why are we segmenting? And 7:56:39 there's four core reasons. Number one is 7:56:42 we want creative relevance. So if I go 7:56:46 to the website and I look at jeans, when 7:56:50 you start retargeting me across display, 7:56:52 across all these different channels that 7:56:53 Pmax can place on, you want to retarget 7:56:56 me with photos and headlines of jeans. 7:56:58 You don't want to retarget me with 7:57:00 random other stuff that's not related to 7:57:02 what I was actually looking at. Now, if 7:57:03 you just group everything together into 7:57:05 one asset group, I will just get served 7:57:08 random stuff. But if you segmented out 7:57:10 and you had a jeans asset group and then 7:57:12 you just had the jeans products from the 7:57:14 listing group in there, then when I go 7:57:16 and look at jeans, it will dynamically 7:57:17 retarget me with relevant assets. This 7:57:19 is also the case on cold targeting. So 7:57:22 even though I don't love it, I don't 7:57:23 love this going out and just targeting 7:57:25 cold on display and YouTube, etc. But 7:57:28 when it does inherently go and do that, 7:57:29 it will go out and if it's all grouped 7:57:31 together, it will just go out and serve 7:57:33 all of your fashion products to anyone 7:57:35 that it thinks is interested in the 7:57:36 brand. Instead, if you have 7:57:37 segmentation, it will go out and take 7:57:39 your jeans products and serve it to 7:57:41 people that it believes are interested 7:57:43 in market right now for buying jeans. 7:57:45 Much more contextually relevant, much 7:57:47 more relevant on a creative angle and 7:57:50 therefore you see better performance. 7:57:51 Number two, which I kind of hinted 7:57:53 towards a little bit there, is audience 7:57:55 signals. So on each of these asset 7:57:57 groups when you set them up, you have to 7:58:00 also select an audience signal. Now, the 7:58:02 big mistake here is people think 7:58:03 audience signals is like targeting in 7:58:06 meta ads back a couple years ago where 7:58:07 you choose your interests and then it 7:58:08 goes out and targets those interests. 7:58:10 That's not how it works at all. Audience 7:58:11 signals are just directional signal that 7:58:13 the campaign uses at the start before it 7:58:15 has a lot of conversion data. So, when 7:58:17 you first launch the campaign, you just 7:58:19 got to give it a little bit of direction 7:58:20 rather than it going off and just 7:58:22 randomly testing. Google tries to not 7:58:24 waste your budget in doing so. And so, 7:58:25 it says, "What kind of people should we 7:58:27 be targeting?" And so you go well for 7:58:28 this asset group you should be targeting 7:58:29 people that are interested in jeans in 7:58:30 market for denim uh that are this age 7:58:33 that look like this that are interested 7:58:34 in this stuff. And so then out of the 7:58:36 gates rather than wasting budget on 7:58:37 testing random people it goes straight 7:58:39 to directionally the people that are 7:58:41 interested. Now obviously audience 7:58:42 signals work better when you're more 7:58:44 granular and you have that segmentation 7:58:45 at an asset group level because if it's 7:58:47 all consolidated you would just go 7:58:49 people in fashion but if it's segmented 7:58:51 you can go no no no no people that are 7:58:52 in market for jeans right now and are 7:58:54 making searches for gene related 7:58:56 products. Number three is search themes. 7:58:59 This is effectively the same thing as 7:59:00 audience signals, but search themes is 7:59:02 saying, "Hey, if they've searched for 7:59:04 keywords like this, then target them." 7:59:06 And then number four is visibility into 7:59:09 performance breakdowns. So when you 7:59:10 introduce segmentation at the asset 7:59:12 group level, you get the ability to 7:59:14 actually see performance on each asset. 7:59:16 So rather than it all being bundled 7:59:17 together, we can go, "Oh, wait a second. 7:59:20 Tops are currently outperforming jeans. 7:59:22 That makes sense because of the 7:59:23 seasonality that we've checked against 7:59:25 the keyword planner and the demand 7:59:26 forecasting of the business. All right, 7:59:28 let's think about how we can 7:59:30 redistribute budgets accordingly or make 7:59:31 changes based on this data insight. If 7:59:33 you don't have asset group segmentation, 7:59:35 you don't get that insight into the data 7:59:38 siloed in that way. So what are some 7:59:39 general rule of thumbs for segmentation 7:59:41 here? Number one is you generally want 7:59:43 one asset group per product type. So in 7:59:46 furniture, if you have chairs, tables, 7:59:48 lighting, decor, you want that split out 7:59:50 at an asset group level. In for example 7:59:52 hardware you would want hand tools, 7:59:54 power tools, fasteners, hardware 7:59:56 accessories split all out into different 7:59:58 asset groups. Inside each asset group, 8:00:01 really important that you customize the 8:00:04 listing group accordingly so that if you 8:00:06 are just putting hardware accessories in 8:00:08 here, the actual products in there are 8:00:11 just hardware accessories. So you want 8:00:12 to make sure that that segmentation is 8:00:13 set up. Number three, you really want 8:00:15 and this is going to depend on actual 8:00:17 skew count in the business. So take this 8:00:19 with some flexibility, but you want at 8:00:20 least 15 products in an asset group or 8:00:24 else you are really just oversegmenting. 8:00:26 What you definitely don't want like a 8:00:28 real red flag here is you don't want one 8:00:29 product in an asset group. And the 8:00:32 reason being is it just won't get any 8:00:33 spend. And so if you have a bunch of 8:00:35 asset groups and then you're throwing in 8:00:36 a couple asset groups, like let's say 8:00:38 you have a top performing hero product 8:00:39 and you're like, I want this in its own 8:00:41 asset group. If there's only one 8:00:42 product, it might actually get not as 8:00:44 much spend as it should. And you 8:00:46 obviously want to take into 8:00:47 consideration the exact same thing that 8:00:49 you take into consideration in account 8:00:50 structure, which is that consolidation 8:00:52 will always beat segmentation. When PMAX 8:00:55 first rolled out, we did tons of testing 8:00:57 on asset groups. We had accounts where 8:00:58 we were launching like 200 asset groups 8:01:00 with this hyperderee of segmentation. 8:01:03 Different asset groups for different 8:01:04 listing groups, different search themes, 8:01:06 different product types, like the list 8:01:07 goes on. What we ended up finding was 8:01:09 that the simpler structures always 8:01:10 performed better. And so you want 8:01:12 segmentation. You want it out by 8:01:14 different product types, etc. But don't 8:01:16 go and throw 400 asset groups in the 8:01:18 account. I'd even say like as a general 8:01:19 rule, have minimum three. Have maximum 8:01:22 20. When you start to go over 20, I 8:01:24 don't see many businesses that have 8:01:25 enough complexity to be able to warrant 8:01:27 that degree of segmentation. So we then 8:01:28 move into search themes. How do you 8:01:31 select your search themes? How important 8:01:32 is this? How does it work? There's three 8:01:34 things to know. Number one is that 8:01:36 search themes function as inputs to 8:01:38 Pmax, but they are signals and they're 8:01:40 not targeting. So when you go and put in 8:01:42 a bunch of keywords or a bunch of search 8:01:44 themes that you think are relevant, it's 8:01:46 not guaranteeing that the PMAX goes in 8:01:47 places and bids on those key terms. Once 8:01:49 again, it's just directional signal. 8:01:51 Number two is that when you set these 8:01:52 up, you want to prioritize the key terms 8:01:55 that are actually converting in the 8:01:57 account. So you want to go have a look 8:01:58 at your search campaigns that you've run 8:02:00 historically, what key terms converted 8:02:01 well, and use them. Don't make stuff up 8:02:03 here. If you can use data from 8:02:05 historical performance in the account, 8:02:06 that's always going to be better. And 8:02:07 then, as I said before, and this might 8:02:10 be a little bit controversial, but 8:02:11 honestly, this is what the data shows 8:02:13 us. After running PMAX on over 250 8:02:16 accounts, consulting on another 400 ecom 8:02:18 accounts from 7 8 9 10 figures is that 8:02:21 search themes and audience signals don't 8:02:23 really matter if you're a mature 8:02:25 account. Kind of the reality. If you're 8:02:27 already spending 30, $40, $50,000, 8:02:29 $100,000 a month on payax, your audience 8:02:32 signals, your search themes, they're not 8:02:33 really going to move the needle. It's 8:02:34 like it's a 1enter at best. And so, if 8:02:36 you get an audit from an agency that's 8:02:38 going in and going, "Oh, your audience 8:02:40 signals can be tweaked a little bit to 8:02:41 look like this, or your search themes 8:02:42 could look like this." It's going to 8:02:44 make literally no impact to the 8:02:45 business. Where search themes actually 8:02:47 matter is on a new launch. And so this 8:02:50 is either literally a new launch of a 8:02:52 new business and this is the first 8:02:53 campaign that you're launching or new 8:02:55 launch of a new product or a new asset 8:02:57 group or a new category. Okay, if 8:02:59 there's no historical data, that's where 8:03:01 the PMAX then goes and looks at the 8:03:03 search themes and the audience signals 8:03:05 and goes, okay, what is the user telling 8:03:07 us? What direction should we go in? And 8:03:09 then it starts there. And then the same 8:03:10 thing applies for audience signals as 8:03:13 well. So, Google will just bypass the 8:03:15 audience signals once it has real 8:03:16 conversion data to work off. Where 8:03:18 they're actually helpful is when you're 8:03:20 warming up a new campaign or a new asset 8:03:22 group, when you have a new account and 8:03:23 the algorithm has no signal. But after 8:03:25 30 days of conversion data pulling up 8:03:28 within the campaign, the audience 8:03:29 signals done its job and you could 8:03:31 remove it from the campaign and it won't 8:03:32 impact performance at all. Now, when it 8:03:34 comes to brand exclusions on Pmax, this 8:03:36 is relatively straightforward. Simply 8:03:38 exclude your brand name from your PMAX 8:03:40 campaigns. There's actually two ways to 8:03:42 do this right now. Now, if you go into 8:03:43 campaign settings and then you scroll 8:03:45 down and then you click brand exclusions 8:03:47 in there, you can then click add new 8:03:50 brand. You need to put your brand in. 8:03:51 Google will go and index it, find all 8:03:53 relevant branded key terms and then you 8:03:54 can select it as a list. That's called 8:03:56 applying a brand list exclusion to PMAX. 8:03:59 This was the only way in which you could 8:04:00 actually exclude your brand for the last 8:04:02 like year. Then recently, Google went 8:04:05 and rolled out negative keywords on 8:04:07 Pmax, which is amazing. So now you can 8:04:10 go in and you could just negative 8:04:11 keyword your brand directly in here. Now 8:04:13 what I like to do is both. Let's put 8:04:15 fail safes in place so that if we miss a 8:04:17 negative keyword variation, it doesn't 8:04:19 go and place on it because the brand 8:04:20 list scoops it up. And then vice versa, 8:04:22 if the negative keywords stop working 8:04:24 for whatever reason on the campaign, at 8:04:25 least we got the brand list there to 8:04:27 actually protect us. If I was to go 8:04:28 through and order the importance of all 8:04:31 of the variables that matter in Pmax, 8:04:34 number one is the actual GMC feed. It is 8:04:37 the titles. It is the descriptions. It 8:04:40 is the imagery within the feed. Assuming 8:04:42 that this is e-commerce and the Pmax is 8:04:45 putting majority of spend into shopping, 8:04:47 right? So, I'm going to put an asterisk 8:04:48 here because this is contingent on the 8:04:49 PMAX actually prioritizing shopping and 8:04:51 this obviously being an e-commerce 8:04:53 brand. Number two is brand exclusions. 8:04:56 If you don't have brand exclusions on 8:04:57 your PMAS campaigns, they are just 8:04:59 retargeting campaigns. Number three is 8:05:02 asset and listing group structure as 8:05:05 well as obviously the assets in the 8:05:07 asset groups. If you have terrible 8:05:09 headlines, terrible images, terrible 8:05:10 descriptions, the ads are going to look 8:05:12 terrible and therefore they're not going 8:05:13 to work. At the end of the day, the 8:05:14 creative of how the ad shows up is 8:05:16 vitally important. And so the assets and 8:05:18 the structure and the copyrightiting etc 8:05:20 in there is going to be third. Number 8:05:21 four is the bidding strategy, which 8:05:25 we'll talk about bidding strategies in 8:05:26 more detail later on. And then number 8:05:28 five is the audience signals and the 8:05:32 searchs. So before we move on from PAX, 8:05:34 there is one variation of Pmax that 8:05:36 exists and you've probably heard it 8:05:37 before if you're deep in the Google ad 8:05:39 space. If you're not, if you've never 8:05:41 even opened a Google ad account before, 8:05:42 it's going to be new to you. This is a 8:05:44 feed only PMAX. Now, how this works is 8:05:46 those asset groups that we were talking 8:05:48 about before, you delete them. And when 8:05:50 you delete the asset groups, all that's 8:05:51 left is the listing group. And the 8:05:54 listing group is the products flowing 8:05:56 through from the Google Merchant Center 8:05:57 feed. And so what happens is PMAX can't 8:06:00 place anywhere except on shopping and 8:06:02 except on display and it will only do 8:06:04 display retargeting because it can't 8:06:06 pull through headlines or descriptions 8:06:08 or anything. All it can show on display 8:06:10 is the GMC feed. So the product tiles 8:06:12 popping up. This is effectively what 8:06:14 smart shopping campaigns were back in 8:06:15 the day. And for those that haven't been 8:06:17 around long enough on Google ads, smart 8:06:18 shopping is what Pmax used to be where 8:06:20 it would only place on shopping and do 8:06:23 display remarketing. Now, a little tip 8:06:25 here for feed only campaigns is there's 8:06:26 two things that end up destroying them, 8:06:28 which is that number one, URL expansion 8:06:30 being on within settings. You need to 8:06:32 make sure URL expansion is off or it 8:06:34 just goes and starts adding URLs, 8:06:36 scraping information, building asset 8:06:37 groups automatically without you doing 8:06:39 it. And then number two is having 8:06:41 automatically created assets on within 8:06:44 settings as well. So you need to make 8:06:45 sure those two settings are off if 8:06:46 you're going to run feed only. Now why 8:06:48 would you run feed only? It's so that 8:06:50 you can force spend into the shopping 8:06:51 placement. If you don't want pmax 8:06:53 placing on search, YouTube, anything 8:06:54 else, you're like, we just want Pmax 8:06:56 placing on shopping. That is it, then 8:06:58 this is the effective hack around being 8:07:00 able to do so. There's a lot of reasons 8:07:02 as to why you might want to pivot into 8:07:04 feed only. There's a lot of reasons as 8:07:06 to why you might want to run feed only 8:07:08 on an account. I'm not going to go into 8:07:09 all the reasons here as this is a little 8:07:10 bit more of a complex nuance strategy 8:07:11 that's going to be context dependent on 8:07:13 the account. But let me give you one 8:07:14 example, which is that we had a client 8:07:16 uh spending a couple hundred,000 on PMAX 8:07:18 per month and then the issue was is that 8:07:20 clicks went through the roof out of 8:07:22 nowhere. Return on ad spend dipped by 8:07:24 about 20%. And we what's going on? Is 8:07:26 this bot traffic? What's actually 8:07:28 occurring here? And what was happening 8:07:29 is the PMAX campaign went rogue and 8:07:32 started distributing about 10% of spend 8:07:35 into YouTube where it was previously 8:07:36 only distributing 2%. and it went and 8:07:38 started distributing like 10% of spend 8:07:40 into Gmail. Gmail cold, which is crazy. 8:07:44 And so, as a product of that, click 8:07:45 volume looked really high and good. The 8:07:48 clicks actually weren't translating to 8:07:50 the website. Here's a little side note 8:07:52 is that Google PMAX campaigns will place 8:07:55 on Gmail, but when they place on Gmail, 8:07:58 and you'll see it is just like a fake 8:07:59 email at the top and you click on it and 8:08:01 it just shows you an email that it puts 8:08:03 together using the assets and the GMC 8:08:05 feed is that the click onto the email, 8:08:07 me just opening the email up, that's 8:08:09 what the platform counts as a click, not 8:08:11 an outbound click to the website. And so 8:08:14 when you place on Gmail, you get a ton 8:08:17 of clicks that seem to be like 8:08:19 10-centent clicks, but they're not 8:08:20 actually clicks. These people aren't 8:08:22 actually going to your website. They're 8:08:23 just opening the email that was in their 8:08:25 inbox. And then what you also end up 8:08:26 with is a lot of customers getting 8:08:27 really angry saying, "We have 8:08:29 unsubscribed. Why are these emails 8:08:31 popping up in my inbox?" And then you 8:08:32 have to say, "Well, they're not actually 8:08:33 an email. They're an ad." And then the 8:08:34 customers don't believe you. So it's a 8:08:36 bit of a nightmare all around. So in 8:08:39 that circumstance, what did we do? We 8:08:41 pivoted the PMAX out into a feed only so 8:08:44 it couldn't physically place on Gmail. 8:08:47 So it couldn't physically go and 8:08:48 overspend on the YouTube channel. And 8:08:51 because of that, we fixed the issue and 8:08:52 resolved it and that account continued 8:08:54 to scale. And so this is really an 8:08:56 ability to crutch and solve problems if 8:08:59 problems arise with the primary PMAX 8:09:02 campaign. You might also decide to have 8:09:04 multiple PMAX campaigns for different 8:09:06 categories and then some categories are 8:09:08 feed only, some categories aren't. The 8:09:10 last comment I'll make on Pmax is T 8:09:13 roaz. So most people's understanding of 8:09:16 bidding strategies is fairly 8:09:17 rudimentary. And so you have target 8:09:20 return on ad spend. The other option 8:09:22 here is that you have max conversion 8:09:24 value. Now what people will do is they 8:09:25 will launch on max conversion value to 8:09:28 accumulate data and then once data has 8:09:30 accumulated they will roll into a target 8:09:33 row strategy. And this is how the 8:09:35 platform works and this is generally 8:09:36 what you should do. Reason being is that 8:09:39 when you're using max conversion value, 8:09:41 this is what performance looks like over 8:09:43 time. When you switch to target rorowaz, 8:09:46 this is what performance looks like over 8:09:48 time. It becomes a lot more stable. So 8:09:50 the campaign rather than making big bets 8:09:52 that are going to pay off and you're 8:09:53 going to see better efficiency and then 8:09:54 it makes big bets and you're going to 8:09:55 lose and you have this instability week 8:09:57 on week, target rorowaz will flatten it 8:09:59 out. Now notice the actual performance 8:10:02 over this whole time period is the same. 8:10:05 You don't necessarily get better 8:10:07 performance with target rorowaz against 8:10:09 maximized conversion value. Performance 8:10:11 ends up averaging to somewhat similar. 8:10:13 The only core difference is the 8:10:15 reliability of the campaign on a 8:10:17 day-to-day basis. Now, most people want 8:10:20 better reliability. They want consistent 8:10:21 revenue coming from the platform every 8:10:23 single day. And therefore, target 8:10:24 rorowaz is a good feature to roll into. 8:10:26 The trap people fall into is that they 8:10:28 roll into target rorowaz. They maybe set 8:10:31 it 400% because their trailing return on 8:10:34 ad spend is maybe 420 430. So they set 8:10:36 it a little bit under and that's fine. 8:10:37 That's all well and good. And the 8:10:38 campaign starts to stabilize and it hits 8:10:40 that rorowaz. It's a lot more stable on 8:10:42 a day-to-day basis and we go fantastic. 8:10:44 We don't want to scale. We don't have 8:10:45 the budget to maybe we're agency side 8:10:47 and the client is strict on financial 8:10:49 year budgets that the board is given. 8:10:51 And so as a product of that, what can we 8:10:53 do? Well, we can increase the target 8:10:55 rorowaz to try to squeeze more 8:10:57 efficiency and more revenue out of the 8:10:59 existing campaign. So you'll come in and 8:11:00 you'll go, "Okay, let's bump this to 8:11:02 425." And then the rorowz goes up a 8:11:03 little bit. And then you go, "Okay, 8:11:04 cool. Let's bump this again to 450." And 8:11:07 then the rorowes goes up a little bit. 8:11:08 And then you keep doing this and maybe 8:11:10 you end up at about 500%. At this point, 8:11:12 spend starts to pull back. It can't 8:11:14 achieve that rorowaz until you find 8:11:16 equilibrium and you find, okay, this is 8:11:18 the max we can go to. Now we're at 500%. 8:11:21 Zoom out, look back, and you go, this 8:11:22 was incredible media buying. Well done 8:11:24 to the agency. Well done to whoever did 8:11:26 this because at the start of the quarter 8:11:29 we were getting a four rorowaz on this 8:11:30 campaign spending $1,000 a day. Now 8:11:33 we're getting a five rorowaz and sorry 8:11:35 this is cuz we're looking in 8:11:37 percentages. This is a five rorowaz. 8:11:38 This is a six rorowaz. Um now we're at a 8:11:41 six rorowaz at $1,000 a day. Great. 8:11:45 Reality is 8:11:47 this down here was probably a better 8:11:49 position for the campaign. And the 8:11:51 reason for that is because this was 8:11:54 likely going after more cold traffic 8:11:56 because the target return on ad spend 8:11:58 trap is that as you increase target 8:12:00 rorowaz for the campaign to achieve this 8:12:04 return what it does is it restricts who 8:12:06 it bids on and what it restricts is cold 8:12:08 audiences. If you think about the entire 8:12:12 population in here and then you think 8:12:14 this is the people that I want to target 8:12:17 that I can achieve a particular 8:12:18 efficiency on. Now this target 8:12:21 demographic is probably going to have a 8:12:22 lot of warm people in it is the reality. 8:12:24 There's probably going to be people that 8:12:25 have already visited the website before 8:12:26 probably some existing customers 8:12:28 probably some people that have seen your 8:12:30 ads on Facebook. And if you just go and 8:12:32 target this narrow amount of people you 8:12:34 will be able to achieve a 7x return. But 8:12:36 then if we want to scale spend, what 8:12:39 happens is that this opens up and we 8:12:41 target more cold users. And as we go out 8:12:44 and out and out to colder and colder 8:12:46 users, we get worse efficiency. The 8:12:48 opposite then occurs when we increase 8:12:50 target row. So if we go and increase it, 8:12:53 let's say this circle is a seven and we 8:12:55 want to go and take it to an eight, what 8:12:57 happens is the circle shrinks and we 8:13:00 actually go in and just start targeting 8:13:02 this area of people. And so even though 8:13:04 efficiency might look better, even 8:13:06 though we're happy, we're actually not 8:13:08 genuinely acquiring new customers and 8:13:09 we're just tightening the pool of users 8:13:11 that we're targeting. This is where you 8:13:12 start to get into feeder strategies. The 8:13:15 core premise of a feeder strategy is 8:13:18 that you have a campaign with a really 8:13:20 low target rorowaz. It could be as low 8:13:23 as, let's say, 50%. So that's a 0.5 8:13:26 rorowaz. And then the idea is that when 8:13:28 we go back to this pool, this campaign 8:13:30 is going to target everyone in the 8:13:32 entire pool. It's going to reach 8:13:33 everyone that's searching for these key 8:13:35 terms. And then you have a second 8:13:37 campaign which has a high target rorowaz 8:13:40 which might be like let's say 600%. And 8:13:43 this campaign will only target warm 8:13:45 audiences. It will only target people 8:13:46 that are in market and highly likely to 8:13:48 buy and have probably visited the 8:13:49 website before. Idea is that this 8:13:51 campaign drives all the traffic captures 8:13:53 as much as possible and then this 8:13:55 campaign converts [clears throat] them. 8:13:57 So you're effectively feeding traffic 8:13:59 from the top of funnelunnel acquisition 8:14:00 campaign into the bottom of funnel 8:14:02 campaign. The added advantage here is 8:14:04 that because most people don't operate a 8:14:05 campaign on such a low target rorowaz is 8:14:07 that you're going to reach people that 8:14:09 no one else is even entering auctions 8:14:10 on. And so where this might be the total 8:14:13 pool of users that are searching for 8:14:15 relevant key terms to your business. 8:14:17 Let's say it's fashion and let's say 8:14:18 it's dresses. So this is everyone 8:14:20 searching for dresses on a day-to-day 8:14:21 basis. Most people are only bidding on 8:14:25 this half of users. And the reason being 8:14:27 is that these half of users have really 8:14:29 high intent, have searched for a lot of 8:14:32 purchase relevant key terms recently, 8:14:34 have visited a lot of websites. And so 8:14:36 Google knows these people are very 8:14:37 likely to buy. And so anyone that has a 8:14:40 high target rorowaz strategy is going to 8:14:42 go and bid here. And everyone's going to 8:14:44 be bidding against each other. But all 8:14:46 of these users from Google's perspective 8:14:48 look super low intent. They haven't 8:14:49 visited any websites before. or this is 8:14:51 their first time searching for this kind 8:14:53 of key term. And so because of that, 8:14:54 almost no one is out here bidding on all 8:14:56 of these auctions for all of these 8:14:58 users. And so the idea, the arbitrage 8:15:00 opportunity is that you can just go and 8:15:02 bid on all these users. It's super cheap 8:15:04 to do so. You can drive tons of traffic 8:15:06 over. Now, yes, a lot of it might not be 8:15:07 that good, but a lot of it might be 8:15:10 good, and we can therefore funnel it in 8:15:12 and convert it on second or third click 8:15:14 using the bottom of funnel PMAX 8:15:16 campaign. And this actually ends up 8:15:18 being a really effective strategy that 8:15:20 we see work on a lot of accounts. The 8:15:22 reason why a lot of people adopt this 8:15:23 strategy is because PMAX was sold to the 8:15:26 community as a replacement for smart 8:15:28 shopping back in 20122. But 3 years 8:15:31 later, we're sitting here and we're 8:15:33 measuring all the results. We're looking 8:15:35 at the findings. I'm talking to other 8:15:37 agency owners who own some of the 8:15:38 biggest Google Ads agencies in the 8:15:40 world. And everyone pretty much says the 8:15:43 same thing behind closed doors, which is 8:15:44 that Pmax substantially overattributes 8:15:47 in the platform. 50% or more of Pmax 8:15:49 conversions are repeat or warm customers 8:15:51 that Pmax didn't actually genuinely 8:15:53 acquire. Standard shopping now 8:15:55 substantially underattributes cuz Pmac 8:15:58 takes Pmax takes all the credit for it 8:15:59 when they sit next to each other. And so 8:16:01 the combined effect is that Pmax looks 8:16:03 like it's scaling well in an account, 8:16:06 but it's actually being held up by 8:16:08 standard shopping doing all of the cold 8:16:10 acquisition work. Now, generally when 8:16:11 you're running a feeder strategy, the 8:16:13 actual approach here is that you run 8:16:15 standard shopping as the cold campaign 8:16:18 because it's just better at actually 8:16:20 placing and reaching cold audiences and 8:16:22 then you run PMAX as effectively the 8:16:24 high Tores retargeting campaign. All 8:16:26 right, so here's three different 8:16:28 strategies that are in market that a lot 8:16:29 of people use that I like that you can 8:16:31 do with standard shopping. Now, the same 8:16:33 rules as Pmax and everything else that 8:16:35 we've talked about still applies here. 8:16:36 Okay, you want consolidation over 8:16:37 segmentation. If you are going to 8:16:38 segment, it should be based on those 8:16:40 four variables that I mentioned earlier. 8:16:41 Um, but here are some strategies in 8:16:43 market that seem to work decently well. 8:16:45 Number one's pretty straightforward, 8:16:47 which is you just have all your top 8:16:48 sellers in one standard shopping 8:16:49 campaign and then you have everything 8:16:51 else in another campaign. The idea here 8:16:53 is that you want to force into Parto's 8:16:55 principle and have 80% of your spend in 8:16:57 the top 20% of products. Um, but it also 8:17:00 allows you the flexibility to continue 8:17:01 to ensure testing budget down here so 8:17:04 that you can roll new winners up. This 8:17:05 is particularly just relevant if you 8:17:07 have a really large skew count and 8:17:09 you're not fashion. If you're in 8:17:10 fashion, I probably wouldn't do this. 8:17:12 And there's a lot of reasons as to why. 8:17:14 And if you're 5 10 product business, you 8:17:17 also shouldn't do this. You don't have 8:17:18 enough products to really warrant this 8:17:20 degree of segmentation. So, this is a 8:17:22 very specific case for a business that 8:17:24 has a lot of products that isn't in 8:17:26 fashion. This is a structure that works. 8:17:28 We then have the feeder strategy, which 8:17:29 I talked about before, which is that you 8:17:31 have standard shopping then feeding into 8:17:33 a high trow pass campaign. And then you 8:17:35 also have the flow boost labelizer. It 8:17:37 couldn't be a Google Ads video if I 8:17:38 didn't mention this. Uh because this is 8:17:40 very common on any people that are deep 8:17:42 into Google Ads. What this really is is 8:17:44 just splitting up your ad groups or your 8:17:47 campaign into different performance 8:17:50 levels of products. So you have 8:17:52 overindex index, near index, under 8:17:54 index, no index, which this means that 8:17:56 it's above expected performance, it's 8:17:58 hitting target, it's just below target, 8:18:00 it's below target, or there's zero 8:18:01 performance at all. Now, if you don't 8:18:03 have a lot of products, you could just 8:18:04 do this manually, and you can just 8:18:05 segment the products out and give them 8:18:06 custom labels. But what you can do 8:18:07 instead is you can set up an automated 8:18:09 uh labelizer script, which will just 8:18:11 autotier them on 30-day rolling 8:18:14 performance in real time. So, 8:18:15 effectively, your products are going to 8:18:16 get shuffled around at an ad group or a 8:18:19 campaign level based on their real-time 8:18:21 performance, and then they will have a 8:18:23 bidding strategy and a budget that is 8:18:25 relevant to how hard you want to push 8:18:26 the product in that given state. Once 8:18:28 again, I don't recommend this to 8:18:29 majority of businesses because it 8:18:31 introduces a lot of segmentation for not 8:18:34 a huge amount of upside unless you have 8:18:36 like thousands of SKs. When you have 8:18:38 thousands of SKs, this becomes a really 8:18:40 effective way to manage it and be able 8:18:42 to stay on top of it. But for most 8:18:44 people, this is just over complicating 8:18:45 Google Ads and you don't need to go this 8:18:47 far. Then we've got search campaigns. 8:18:48 Transparently, search campaigns is not 8:18:51 something that we focus on that much. 8:18:53 It's because of what I said before, 8:18:54 which is that for 90% of e-commerce 8:18:56 brands, 90% of your focus should be 8:18:58 going into shopping because that's where 8:18:59 the greatest leverage is going to be. 8:19:01 Now, that breaks in one particular 8:19:03 circumstance in ecom, which is in ecom 8:19:05 businesses that have a B2B component. 8:19:08 So, if you sell to businesses in some 8:19:11 capacity, but it is through an ecom 8:19:13 store, you do actually see arguably 8:19:15 better performance on search campaigns 8:19:17 over shopping. I'll give you an example 8:19:19 of this which is that if you're selling 8:19:21 bulk eyelashes to like technicians and 8:19:24 small businesses in that case you end up 8:19:26 with better performance on search and 8:19:28 the reason for this is because when you 8:19:30 are making a business search you often 8:19:32 don't click on a shopping listing 8:19:34 because it doesn't match the price 8:19:35 discrepancy that you have in your head. 8:19:37 So, if I'm going to go and make a 8:19:39 business purchase, typically I want 8:19:41 something in bulk, right? If I'm going 8:19:43 to buy like tea for my coffee shop or my 8:19:46 restaurant, I either have a known 8:19:48 supplier or I'm going to go to Google 8:19:49 and try to find a supplier. Now, when 8:19:51 you're going and finding a supplier and 8:19:52 you see a bunch of shopping listings 8:19:54 that say tea is $5, $10. I'm not going 8:19:56 to click on them because I know that do 8:19:58 these people actually do bulk rates, do 8:20:00 they do business deals, you just know 8:20:02 that B2B business does not take place on 8:20:04 the shopping network. it's intuitive to 8:20:06 the user and so because of that you go 8:20:08 to search listings and you start 8:20:09 clicking on search ads and so that is 8:20:11 the case where B2B actually makes search 8:20:13 campaigns more important. So there's a 8:20:15 few different components of search 8:20:16 campaigns. Number one what we talked 8:20:17 about earlier which is you want to be 8:20:19 using topic ad groups not keyword ad 8:20:21 groups on negative keywords. The first 8:20:23 30 days of a search campaign is the 8:20:25 highest leverage point for negative 8:20:27 keywords because the campaign's going to 8:20:29 go out very broad and target a bunch of 8:20:31 keywords and there's going to be a lot 8:20:32 of stuff in there that is actually 8:20:33 irrelevant that you don't want to place 8:20:35 on. And so as a product of that in the 8:20:36 first 30 days I would be reviewing 8:20:38 negative keywords weekly dependent on 8:20:40 budgets. If budgets are really high this 8:20:42 can literally go down to a daily level 8:20:44 so that you're not allowing any kind of 8:20:46 keyword bleed and waste money. From 8:20:48 month one onwards you can drop this to 8:20:50 weekly-nightly. 8:20:52 You actually don't want to tweak with 8:20:54 negative keywords too much as every time 8:20:56 you do it, it does reset learnings and 8:20:58 impact performance a little bit. On top 8:20:59 of this, you also want a universal 8:21:01 negative keyword list that's just 8:21:03 applied at the account level. So, every 8:21:05 single new campaign inherits it. We have 8:21:06 one of these internally. You can 8:21:07 literally just go and search up for one. 8:21:09 I think they'll be all over the internet 8:21:10 or you could ask an LLM to try to build 8:21:12 one for you. It just has stuff like job 8:21:14 related queries. So if a key term has 8:21:16 jobs or careers or interns or hiring or 8:21:18 salary stuff like that, you just don't 8:21:20 want to place on that key term because 8:21:22 it's not relevant. And just a quick note 8:21:23 on responsive search ads, which is the 8:21:26 only thing that you can run these days, 8:21:27 is reiterating what I said before. There 8:21:29 is 2,700 possible combinations of a 8:21:33 search ad when you set it up. If we 8:21:35 assume it takes 100 impressions to be 8:21:37 able to determine whether a specific 8:21:39 combination works, which is honestly 8:21:40 really low, like I wouldn't want 100 8:21:42 impressions of data, but if you assume 8:21:44 that, then it takes 270,000 8:21:46 impressions to be able to actually exit 8:21:49 learning and have the campaign 8:21:50 understand which combo works. And so the 8:21:53 implication of that is that you want to 8:21:55 minimize segmentation because the more 8:21:58 segmentation you have, the longer these 8:22:00 RSAs will be in learning. And in most 8:22:01 accounts, they're just in learning 8:22:03 indefinitely. One final note here that 8:22:05 isn't spoken enough and I actually think 8:22:06 it's one of the largest levers that 8:22:08 exist within search campaigns is nothing 8:22:10 to do with Google ads. It's not even in 8:22:12 the account. It is the landing page. And 8:22:14 the reason being is that if you think 8:22:16 through search campaigns and Google ads 8:22:18 more as a whole, if we just step back 8:22:20 out of the platform and we get out of 8:22:22 the weeds and all the tactics and what 8:22:23 we're actually doing here and we just 8:22:24 think through spending to acquire 8:22:27 customers, what we are looking at is 8:22:30 that we are charged a cost per click on 8:22:32 Google and then we are driving this to 8:22:35 the website and then we are getting two 8:22:37 things out the back. Ultimately, we're 8:22:39 getting revenue, but revenue is a 8:22:40 function of the conversion rate on the 8:22:44 website and the average order value of 8:22:47 those conversions that are occurring, 8:22:49 that ultimately will then equal rev. 8:22:52 Now, when it comes to CPCs, there is a 8:22:54 little bit more nuance within this, 8:22:56 which is that if we just arbitrarily 8:22:58 optimize for the lowest CPC, it doesn't 8:23:00 necessarily improve this equation. In 8:23:03 fact, CPCs are actually a vanity metric, 8:23:06 which seems counterintuitive because you 8:23:08 look at this and you go, "Well, all that 8:23:09 actually matters is we need the highest 8:23:10 average order value, the highest 8:23:11 conversion rate, and the lowest CPCs." 8:23:13 Not true. And that's where people get 8:23:14 into trouble. CPCs, and you see this on 8:23:17 any large data set and analysis, CPCs 8:23:20 and ROI, and you can measure this on 8:23:22 acquisition, me on just rorowaz in the 8:23:24 platform, doesn't matter what it is, 8:23:25 there is very little relationship 8:23:27 whatsoever. You look at accounts and 8:23:29 some accounts have very high CPCs with a 8:23:30 very high rorowaz and vice versa. You 8:23:32 see this on a product level, you see 8:23:33 this on a campaign level. On the 8:23:35 peripherals, it's true. If CPCs are like 8:23:37 $100, yeah, you'll never be profitable. 8:23:39 And then vice versa, if CPCs are like 10 8:23:41 cents, it's probably junk traffic and 8:23:43 you'll never be profitable. But once we 8:23:45 tighten into the middle, CPCs don't 8:23:48 really give us any good indication of 8:23:50 performance. And that is because the 8:23:51 quality of this traffic 8:23:54 changes. You can have high quality 8:23:56 clicks, you can have low quality clicks. 8:23:57 And the quality of the clicks is 8:23:59 ultimately based on the automated 8:24:01 bidding that is occurring in the 8:24:03 platform. And you are trusting that the 8:24:04 campaign is going to bid correctly on 8:24:07 users that are likely to convert from 8:24:08 you. Which means that all of Google 8:24:11 ultimately just comes down to the 8:24:14 auction that you have against 8:24:16 competitors. You and all of your 8:24:18 competitors are bidding on a specific 8:24:20 key term to rank. And at the end of the 8:24:22 day, whoever can pay the most amount of 8:24:25 money to win the auction wins the 8:24:26 listing, gets the click, gets the 8:24:29 conversion. Now, if Google is just all 8:24:31 about who can spend the most money at 8:24:33 auction to get the placement and get the 8:24:35 conversion, then how do we make sure 8:24:37 that we can pay the most amount of 8:24:39 money? Well, it is by having the highest 8:24:41 conversion rate possible on the website. 8:24:43 If we are converting 2x higher than all 8:24:45 our competitors, well, we can spend 8:24:46 double the amount on a click and we'll 8:24:48 make the same amount of money as them. 8:24:50 And that's how we obviously flood Google 8:24:52 and beat them. Same thing for average 8:24:53 order value. If our average order value, 8:24:55 then we squeeze more revenue out of the 8:24:57 same amount of clicks and therefore we 8:24:58 get an advantage and we can spend more. 8:25:00 And so ultimately all of the stuff that 8:25:02 we're doing in the platform, the 8:25:03 segmentation, the asset groups, the 8:25:06 audience signals, the optimization of 8:25:08 the GMC feed, which we're about to go 8:25:09 into, all of these things are trying to 8:25:12 improve the quality, which is improving 8:25:15 the auctions that we enter. So we enter 8:25:17 better auctions. we don't waste on 8:25:19 auctions that don't matter on lowquality 8:25:20 users and it's ultimately improving our 8:25:23 quality score on the platform so that 8:25:25 our bid that goes into the auction is 8:25:28 lower artificially than competitors 8:25:30 because Google's giving us an advantage 8:25:32 and so that's ultimately everything 8:25:33 we're trying to do in the platform is 8:25:35 here but at the end of the day with all 8:25:37 of that in place we're still going to be 8:25:38 restricted based on the fact that if a 8:25:40 competitor has higher conversion rates 8:25:42 and higher average order value they can 8:25:43 just still out bid us with worse ads 8:25:45 with a worse campaign structure with 8:25:47 worse everything and and they will still 8:25:48 beat us because they can afford to. And 8:25:50 so when it comes to search campaigns, 8:25:53 what very little people are doing is 8:25:56 just split testing landing pages. It's 8:25:58 just that simple is that when you have 8:26:00 an ad group or a campaign set up, rather 8:26:03 than just having one ad, have two, 8:26:05 duplicate it, have two ads, but ad one 8:26:07 goes to one landing page, landing page 8:26:09 one, and this goes to landing page two. 8:26:10 And then we monitor conversion rates 8:26:12 over a 30 to 60 day period. And then the 8:26:13 landing page with a better conversion 8:26:15 rate, that gets selected. And then what 8:26:17 do we do? We rotate in another test and 8:26:19 we have add three. This gets turned off 8:26:21 and we roll in. And let's say that this 8:26:23 landing page was even better. Amazing. 8:26:25 And then slowly we're increasing the 8:26:27 average conversion rate of the campaigns 8:26:29 which is allowing us to bid more. 8:26:31 Ideally, if we can do some average order 8:26:32 value optimization too, even better. And 8:26:34 so our revenue per visitor from the 8:26:36 campaign improves, which gives us the 8:26:38 ability to unlock scam. I strongly 8:26:39 believe that landing page testing on 8:26:41 search campaigns is actually one of the 8:26:42 biggest levers that you have. Before we 8:26:44 move into bidding, two quick side notes. 8:26:46 Make sure you're excluding brand on 8:26:47 search campaigns. And then when it comes 8:26:49 to bidding strategies, there's so many 8:26:51 different tactics here, but generally 8:26:52 speaking, I just go with a smart bidding 8:26:53 strategy like target rorowaz or max 8:26:55 conversions or something similar. Let's 8:26:57 talk about bidding strategies at an 8:26:59 entire account level. So, what are your 8:27:01 different options? Which one should you 8:27:02 go through? How do you understand the 8:27:04 differences? Let's start with the ones 8:27:05 that I really don't like because we only 8:27:07 work with established businesses. But if 8:27:08 there's people watching this that aren't 8:27:10 established, I have to mention it, which 8:27:11 is that if you have a new account with 8:27:13 no conversion data, you actually won't 8:27:15 even be able to use smart bidding 8:27:16 strategies, at least the last time I 8:27:17 checked. And so, you'll be forced into 8:27:19 using a manual bidding strategy. You 8:27:21 should use manual CPC or maximize click. 8:27:23 If you are an established business, you 8:27:26 should not be using either of these 8:27:27 strategies unless the agency is very, 8:27:29 very competent and has some kind of 8:27:31 complex intricate strategy as to why 8:27:33 they are applying this. We will 8:27:35 sometimes do some clever tricks and 8:27:37 strategies in an account to try to 8:27:38 squeeze out more performance. And it 8:27:40 might involve using something like 8:27:41 maximize clicks. It's rare. You won't 8:27:43 see it in 90% of accounts that we 8:27:45 manage, but it's a test that will 8:27:46 sometimes run to see if we can get a 8:27:48 squeeze or take advantage of an 8:27:51 arbitrage opportunity. This really 8:27:53 should just be beginners and ideally you 8:27:54 get off these bidding strategies as 8:27:56 quickly as humanly possible. Then you 8:27:58 roll into maximize conversions or 8:27:59 maximize conversion value. These are 8:28:01 what's called expansive bidding 8:28:03 strategies. And the reason why it's 8:28:04 called expansive and target rorowaz and 8:28:07 target CPA are restrictive is because 8:28:10 target CPA and target rorowaz restrict 8:28:14 who you actually enter the auction on 8:28:17 based on an efficiency target. You're 8:28:19 putting a limit on it and saying don't 8:28:20 enter auctions unless you can guarantee 8:28:22 this particular return. But on expansive 8:28:24 bidding strategies, they can go and 8:28:26 enter whatever auction they want. Now 8:28:27 obviously it's going to do it within uh 8:28:29 the constraints of actually getting you 8:28:30 results but it will be more aggressive 8:28:33 and it will go into auctions that 8:28:34 otherwise you wouldn't have gone into if 8:28:36 there was a restriction on the campaign. 8:28:38 Now there are benefits and disadvantages 8:28:39 of that. The benefits are you will win 8:28:41 auctions that you otherwise wouldn't 8:28:42 have won. The disadvantages is you will 8:28:44 lose auctions that you otherwise 8:28:45 wouldn't have lost. And so this is 8:28:47 effectively a higher risk model to run 8:28:49 on. Now a core thing to understand here, 8:28:51 easy way to conceptualize the difference 8:28:52 between expansive and restrictive 8:28:54 bidding strategies is percentage 8:28:56 allocation of budget towards testing. So 8:28:58 if we were to just theorize this for a 8:29:00 second so you can understand the 8:29:02 concept, let's say that maximize 8:29:03 conversion value campaigns, they put 8:29:05 about 30% of budget towards testing. And 8:29:08 what I mean by testing is that it's 8:29:10 testing new keywords, new audiences, new 8:29:13 if it's performance max, new placement 8:29:15 types, new regions where people haven't 8:29:17 bought from before, more new 8:29:19 psychographic data points. It is testing 8:29:21 stuff that based on historical 8:29:23 conversion data doesn't seem like the 8:29:25 best thing to test. It isn't it hasn't 8:29:27 converted before, but it's adjacent. 8:29:29 It's tangential. It seems like ah it's a 8:29:31 worthy test. Let's put it in and see 8:29:33 what happens. And so because of that, 8:29:34 this campaign continues to test new 8:29:36 audiences and find new audiences that 8:29:38 work, which unlocks scale. And so it 8:29:41 will go and test on all of these 8:29:42 different people and it go actually 8:29:43 people in this region in this target 8:29:45 demographic are actually learn working 8:29:47 quite well. Let's double down, spend 8:29:48 some more money there. And that's what 8:29:50 allows you to unlock scale and continue 8:29:51 to increase budgets. Whereas with a 8:29:53 restrictive bidding strategy, there's 8:29:55 about a 0% testing budget. Now, this 8:29:58 percentage allocation really is 8:29:59 contextual to how aggressive the target 8:30:01 rorowaz is against the actual rorowaz in 8:30:03 the campaign. If you're achieving a four 8:30:05 rorowaz and your target rorowaz is a 8:30:07 4.5, you'll definitely have 0% testing. 8:30:09 It's going to put all the budget into 8:30:10 ensuring that it can actually drive the 8:30:12 efficiency you want. If instead there's 8:30:13 a big discrepancy, so maybe you're 8:30:15 hitting a five rorowaz, but your target 8:30:16 row is on the campaigns at two, well 8:30:18 then you actually might have some 8:30:19 testing budget in there and it's going 8:30:21 to operate more like max conversion b. 8:30:23 Now, the advantage here is that you 8:30:24 don't have any testing budget, so you're 8:30:25 going to be more efficient. Uh the 8:30:26 disadvantage is that you're going to be 8:30:28 siloed into a particular audience and a 8:30:30 particular target demographic and it's 8:30:31 not going to evolve over time as the 8:30:33 business evolves and as the platform 8:30:34 evolves. And so just being on target row 8:30:37 permanently for an extended period of 8:30:38 time can be a dangerous position to be 8:30:40 in. Ideally, we like to rotate between 8:30:42 the two over extended time periods so we 8:30:44 don't pull the account into a position 8:30:46 where it's too siloed in in terms of 8:30:48 data and it starts to negative feedback 8:30:50 loop. But we also don't want to just be 8:30:51 on maximize conversion value all day and 8:30:53 not be as efficient as we could be. Now 8:30:55 the question also becomes now that you 8:30:57 understand the difference between 8:30:58 expansive and restrictive well which of 8:31:00 these should we use? Should we use 8:31:02 maximize conversions or should we use 8:31:04 maximize conversion value? And then that 8:31:06 will obviously also change which version 8:31:08 of these bidding strategies you roll 8:31:09 into because TCPA is a version or a 8:31:12 subcomponent of max conversions. Tores 8:31:15 is a sub component of max conversion 8:31:17 value. Now in e-commerce I would 8:31:19 generally 90% of the time recommend that 8:31:21 you just go for value. And the reason 8:31:22 being is that when you go for 8:31:24 conversions is ultimately optimizing for 8:31:25 the lowest CPA possible. And the issue 8:31:28 with optimizing for the lowest CPA or 8:31:30 the lowest CAC cost to acquire a 8:31:32 customer, it will optimize towards 8:31:34 pushing your cheapest product because if 8:31:36 you have a, let's say, a $40 t-shirt, 8:31:39 it's much easier to sell a $40 t-shirt 8:31:41 than it is to sell a $200 jumper. And so 8:31:44 as a product of that, all the spend just 8:31:47 ends up going here because you could 8:31:48 probably achieve like $20 CPAs on this 8:31:51 t-shirt. But on this jumper, it might 8:31:53 cost you $60 to be able to actually 8:31:55 acquire an order. Now, the rorowaz here 8:31:57 might be very similar. It might be the 8:31:58 same, but because this has a lower CPA 8:32:01 against the spend. And in fact, in this 8:32:02 instance, you can see the rorowz here is 8:32:03 actually better. We would prefer budget 8:32:06 goes into the jumper. We're getting like 8:32:07 a 3.5x ROI here. 3.5x. Over here, we're 8:32:11 going to get 2x. But campaign doesn't 8:32:13 care. It's not optimizing for rows. It's 8:32:15 not optimizing for value. It's 8:32:16 optimizing for the lowest CPA and the 8:32:19 most amount of conversions, most amount 8:32:20 of orders. And so, as a product of that, 8:32:22 all the budget goes here, which is not 8:32:24 what we want. Hence, max conversions can 8:32:26 actually, and it typically does end up 8:32:29 putting a lot of your spend into 8:32:30 products that you don't want to put 8:32:32 money into. So, a few more quick notes 8:32:34 on bidding is that you will see a target 8:32:36 rorowaz spiral. Um this is super common 8:32:40 in Google ad accounts which is that when 8:32:42 you look at the account over time spend 8:32:44 will be like here and then it will just 8:32:46 slowly trail off and campaigns will just 8:32:49 die. And the reason for this is that if 8:32:53 we overlay rorowaz rorowaz might have 8:32:56 dipped and then because the target 8:32:58 rorowaz is sitting somewhere like here 8:33:01 the account is no longer hitting the 8:33:03 rorowaz goal and so spend pulls back. 8:33:06 And what this really is is the bullseye 8:33:10 analogy that I kind of gave before, 8:33:12 which is that you are targeting these 8:33:14 people and then as you stop achieving 8:33:16 the target rorowaz, you need to shrink 8:33:19 the circle even further to make sure you 8:33:21 still achieve it. And so the circle gets 8:33:23 smaller and smaller and therefore spend 8:33:24 starts to fall off. So once you get into 8:33:26 this circumstance, if you leave target 8:33:28 rorowaz where it is, entire campaign 8:33:30 declines and just dies. If you start 8:33:32 pulling target rorowaz back, you need to 8:33:34 pull it back aggressively or else it 8:33:36 continues to die. What people do is 8:33:37 they're not aggressive enough in pulling 8:33:38 the troz back. So the entire account 8:33:41 just continues to decline. Now the 8:33:42 reason why this spiral continues is 8:33:44 because as you are not achieving the 8:33:47 target rorowaz, conversion data falls 8:33:49 off and target rorowaz uses primarily 8:33:52 the last 30 to 90 days of conversion 8:33:55 data to be able to model who it should 8:33:57 target. Now, as your conversion volume 8:33:59 starts to decline, the modeling accuracy 8:34:02 starts to decline and this is where you 8:34:05 end up in these negative feedback loops 8:34:07 because let's say in the last 30 days 8:34:08 you were getting 100 conversions and it 8:34:10 was using those 100 conversions to 8:34:12 optimize and figure out who to target 8:34:13 and then now this suddenly drops for 8:34:15 whatever reason. Maybe it's seasonality, 8:34:16 maybe a product went out of stock that 8:34:18 was good, something happened, your 8:34:19 conversions drop to 80. Well, now you 8:34:21 have less conversion data to be able to 8:34:23 model off and because of that you have 8:34:25 small sample size bias. Therefore, the 8:34:28 targeting gets worse. Therefore, the 8:34:29 conversion volume falls further. And as 8:34:31 the conversion volume falls further, you 8:34:33 stop hitting the target rorowaz. And 8:34:34 therefore, the spend starts to pull 8:34:36 back. And as the spend pulls back, you 8:34:37 get even lower volume. And then even 8:34:39 lower volume. And as the volume 8:34:40 declines, the accuracy of the bidding 8:34:42 declines and the bidding model starts 8:34:44 performing worse. And you just negative 8:34:46 death spiral. This is what you need to 8:34:47 be really cognizant on and careful 8:34:49 because if you don't know mechanically 8:34:51 why this happens, how it exists. You see 8:34:53 campaigns all the time that just do 8:34:55 this. We on board campaigns that are 8:34:56 like here a lot of the time and I say in 8:34:59 the audit I'm like this is exactly 8:35:00 what's happening right now. You need to 8:35:02 fix it soon or you're going to end up 8:35:03 here pretty soon. And these Google 8:35:05 campaigns aren't going to be driving any 8:35:06 volume for you anymore. Also, a specific 8:35:08 hack worth knowing is that there's stuff 8:35:11 called portfolio bidding strategies. And 8:35:13 this is where you set the bidding 8:35:14 strategy up at the account level. Now, 8:35:17 the reason why you would do this is it 8:35:19 gives you access to an extra feature 8:35:20 which is kind of cool. You also get 8:35:22 access to this feature in SA3, but 8:35:24 that's only for enterprise businesses, 8:35:26 so it doesn't apply to most people. Um, 8:35:27 and portfolio bidding strategies 8:35:29 actually allow you to merge learnings, 8:35:32 which is really cool and really critical 8:35:35 as an implementation if you have too 8:35:37 much segmentation in the account. What 8:35:38 am I talking about? Right at the start 8:35:40 of the video, if you remember, I drew 8:35:42 out campaign ABC and I spoke about how 8:35:45 these campaigns don't share learnings. 8:35:48 They're isolated. Hence, segmentation is 8:35:50 not good. You can actually fix this. You 8:35:53 can take down these silos and allow the 8:35:56 campaigns to share learnings with each 8:35:59 other. The way that you do this is you 8:36:01 set up a portfolio bidding strategy that 8:36:04 then gets applied to all of the 8:36:05 campaigns and the learnings get housed. 8:36:08 They all learn together. This is a 8:36:10 really cool additional strategy that you 8:36:12 can layer in if you want the 8:36:13 segmentation but you want the learnings 8:36:15 being pulled together. Now the other 8:36:17 benefit is that you can use a for 8:36:19 example target rorowaz strategy or a 8:36:21 maximize value strategy plus implement a 8:36:24 max and a min CPC. So if you don't want 8:36:28 the campaign for example spending over 8:36:30 $10 on a click and let's say that you 8:36:32 work in high-end furniture and so there 8:36:35 are sometimes very expensive bids in 8:36:37 that niche where you might go out and 8:36:38 just bid 10 20 $30 on a user. You're 8:36:40 like we don't want them at all. We never 8:36:42 want a $30 click. It's not going to be 8:36:44 profitable for us. I don't care how high 8:36:45 intent they are. You can go up and apply 8:36:47 a portfolio bid strategy where you can 8:36:49 specify the target rorowaz and then put 8:36:52 a max CPC on the campaign. On search 8:36:55 campaigns, there's also a setting called 8:36:57 smart bidding exploration. I actually 8:37:00 really like this as a concept. 8:37:01 Unfortunately, it's not available on 8:37:02 Pmax or on shopping. But the idea here 8:37:05 is that even with a target rorowaz, it 8:37:07 can go outside of the target rorowaz in 8:37:09 10, 20, 30% bands and do testing, which 8:37:12 I think is really cool because it's 8:37:13 effectively a middle ground between 8:37:15 using a restrictive bidding strategy and 8:37:17 an expansive bidding strategy where 8:37:18 you're kind of getting the best of both 8:37:20 worlds. You're restrictive, you're going 8:37:21 to hit target rorowaz, but you're like, 8:37:22 "Hey, go out sometimes and make some 8:37:25 auction plays that otherwise you 8:37:27 normally wouldn't of to be able to learn 8:37:29 some more things and to be able to 8:37:30 ultimately unlock scale." So, if you're 8:37:32 using search, I'd recommend looking into 8:37:33 that setting. Limited by budget. You 8:37:35 will see this all over your Google ad 8:37:37 account on campaigns saying that you 8:37:39 could spend more spend more on these 8:37:41 campaigns. Now, this is a product 8:37:42 supposedly of the campaign having the 8:37:45 ability to enter more auctions where it 8:37:46 didn't because of a restriction in 8:37:48 budget. So, this is an indicator that 8:37:51 you may be able to spend more on the 8:37:53 campaigns. This is not a rule that you 8:37:56 can spend more on the campaigns. I have 8:37:58 almost always seen on every account, no 8:38:00 matter how high we get the spend, 8:38:01 everything's always limited by budget. 8:38:03 We have campaigns spending5 $10,000 a 8:38:06 day on PMAX in a very small total 8:38:08 addressable market in Australia, and it 8:38:10 says limited by budget. So, you need to 8:38:12 be thoughtful about this. Use it as an 8:38:15 indicator, but don't use this as the 8:38:17 effective bible for being able to 8:38:19 determine whether there's incremental 8:38:21 spend available within a campaign. Just 8:38:23 a quick note on budget changes, uh, 8:38:25 which is that the rule that you've 8:38:26 always heard, increase budgets in 20% 8:38:28 increments. Uh, it's real. It applies on 8:38:31 Google. On Meta, Tik Tok, you have a bit 8:38:32 more flexibility. You can increase 8:38:34 budgets way faster than that. On Google, 8:38:36 I've generally seen it always be a 8:38:37 pretty bad idea. When you mess with 8:38:38 budgets too heavily, it's way more of a 8:38:40 stable platform. It drives conversions 8:38:42 way more consistently. And as a product 8:38:44 of that, it wants slower, more 8:38:46 methodical changes, and you're not just 8:38:48 throwing budgets around. Doubling 8:38:49 budgets overnight, not a good idea. 8:38:51 Having a target CPA overnight, not a 8:38:54 good idea. Now, bringing budgets down 8:38:56 quickly, we've generally seen is fine. 8:38:58 If you're really high budgets and you 8:38:59 just have them, it's generally okay. You 8:39:01 don't have much instability. You don't 8:39:03 have much of a performance drop off. But 8:39:04 going upwards in budgets, you want to 8:39:06 move slowly and you want to move slowly 8:39:08 in terms of bidding strategy changes as 8:39:09 well. We lastly have GMC. Now, I would 8:39:12 argue that Google Merchant Center or 8:39:14 feed optimization is actually one of the 8:39:16 most important levers when it comes to 8:39:18 shopping. And the reason being is that 8:39:20 how does shopping know what key terms to 8:39:22 place on? It's using the feed. How does 8:39:24 Google determine your quality score for 8:39:26 where your bids will sit? It depends on 8:39:28 your feed. So the most important thing 8:39:30 when it comes to GMC optimization is the 8:39:33 product title and the product title 8:39:35 structure. Most GMC titles are not 8:39:36 optimized. This is really 8:39:38 underleveraged. How do you actually want 8:39:40 to structure the title? You want to 8:39:41 structure it something like this. And 8:39:42 there isn't a golden rule, but this at 8:39:44 least gives an indication as to a title 8:39:46 buildout. Let's say that by default you 8:39:48 have a dress and normally dresses in 8:39:51 fashion everything has a name. I'm just 8:39:52 going to call this the Nathan dress just 8:39:54 for the sake of this example. What will 8:39:56 by default get pulled into GMC and seren 8:39:58 shopping is something like Nathan dress 8:40:01 size 12 and then maybe the color will 8:40:04 get appended as well. Sometimes you 8:40:06 might also have a brand name getting 8:40:08 automatically appended to the front or 8:40:10 the end of the key term too. So maybe at 8:40:12 the start you have the brand here as 8:40:14 well. And so that's the default title 8:40:16 structure. Now, the issue with this 8:40:18 title structure is that the brand means 8:40:20 nothing if we're not ranking on branded 8:40:22 keywords because we have brand excluded. 8:40:24 So this doesn't mean anything. Also, the 8:40:26 brand is at the bottom of the shopping 8:40:27 listing anyway. So this like really is 8:40:29 redundant and not needed. Nathan means 8:40:31 nothing to anyone. Okay, unless people 8:40:33 are product aware and know the exact 8:40:35 product that you're selling, this means 8:40:37 nothing to anyone. We then have dress, 8:40:39 which is the first point in which this 8:40:40 actually becomes relevant, but it 8:40:41 doesn't tell us anything about what type 8:40:43 of dress this is. And then we have the 8:40:45 size, which is also kind of irrelevant, 8:40:47 but yeah, it's a nice to have to make 8:40:49 sure that people know that it actually 8:40:50 has the size that they're in. And so 8:40:51 this ultimately just won't perform very 8:40:53 well. So instead, what we want to do is 8:40:55 take the brand name, push it to the 8:40:57 back. If no one even knows who you are, 8:40:59 don't have it at all. But if you're a 8:41:00 big retail business, still probably have 8:41:02 it. Push it to the back. And then at the 8:41:04 front we want to start having relevant 8:41:06 keywords and context for not only the 8:41:09 user but mainly for the algorithm. So 8:41:11 how I would start to reshape this is 8:41:14 this can stay if this is the unique 8:41:15 name. Okay, we can keep this but let's 8:41:17 start to pull some keywords in. And 8:41:19 there's really two options here. We can 8:41:20 have the keywords right at the start or 8:41:22 we can still start with the the product 8:41:24 name and type and then go into the 8:41:26 keywords. And so in this case we're 8:41:27 going and we're adding formal dress to 8:41:29 the start which is going to add way more 8:41:30 context and allow us to actually place 8:41:32 very well on this key term. Now, why did 8:41:34 we put this key term? It's based on 8:41:35 actual keyword research of what's 8:41:37 performed historically well on the 8:41:38 account. So, we're not just making stuff 8:41:39 up and putting into the title. The 8:41:42 titles are always built based on actual 8:41:44 conversion historical data out of search 8:41:46 term reports. So, you look at the search 8:41:47 term reports, you what is done well. 8:41:49 Okay, formal dress performs unbelievably 8:41:51 well. This is a formal dress. So, let's 8:41:53 append it to the start of this title. 8:41:55 Having as much attributes as possible as 8:41:57 well as long as it fits within the title 8:41:58 length is also good because this will be 8:42:01 used to in placements of key terms. So 8:42:03 if someone for example searches for a 8:42:04 blue dress, this will then get 8:42:06 prioritized in the rankings where 8:42:08 previously it wouldn't have because it 8:42:09 wasn't actually listed within their 8:42:11 title or any of the attributes of the 8:42:13 product. And so you want to be thinking 8:42:15 through still have the product title or 8:42:16 whatever the product's called. Sometimes 8:42:18 in other industries, if we're talking 8:42:19 like CPG, you can get way more flexible 8:42:22 with this. So let's say that instead 8:42:23 this is just creatine powder and the 8:42:25 website it's called creatine powder. 8:42:27 Well, this is where we'd want to start 8:42:28 going and doing keyword research and 8:42:30 finding out what key term variations of 8:42:32 creatine powder is working. And we might 8:42:34 find that dissolvable 8:42:36 creatine powder is a very big search key 8:42:39 term which performs well. And so we go 8:42:41 dissolvable creatine powder. Then we 8:42:43 look for another key term that's doing 8:42:45 well and we might find that it's clear 8:42:47 and then flavored and then portable. And 8:42:50 then at the very end we can put our 8:42:51 brand name. And so what we are doing is 8:42:53 chunking in as many relevant key terms 8:42:55 as possible into the title so that we're 8:42:57 maximizing the surface area in which we 8:42:59 replace on. A little additional hack 8:43:01 here is that if this creatine powder has 8:43:03 multiple variants, if you have let's say 8:43:06 a 50 g version, then you have a 100 g 8:43:08 version and then you have a 500 g 8:43:10 version. Each of these depending on the 8:43:12 how the feed is set up, each of these 8:43:14 will have its own separate shopping 8:43:16 listing. As a product of that, you can 8:43:18 have different titles, different photos, 8:43:21 different descriptions for all three of 8:43:23 these variants despite it ultimately 8:43:24 going to the same landing page with the 8:43:26 same product. And so you can have one of 8:43:28 these focused on a pool of keywords. 8:43:32 Then you can have the other one focused 8:43:33 on a different pool of keywords and then 8:43:35 this one focused on a different pool of 8:43:36 keywords. So you're effectively finding 8:43:37 different audiences and maximizing your 8:43:40 surface area across Google by leveraging 8:43:42 all the different variants and spinning 8:43:44 off the titles accordingly. On images, 8:43:46 there's two things you want to fix. You 8:43:48 want to make sure that the aspect ratio 8:43:49 actually matches the platform. There's a 8:43:51 really common issue in fashion, which is 8:43:53 that the images by default will be a 9 8:43:56 by6. And so when they pull through into 8:43:58 the feed, they'll get autocropped. And 8:44:00 often the autocropping doesn't actually 8:44:01 look very good. It isn't correct. And so 8:44:03 instead, you want to make sure that 8:44:04 you're overriding the feed with a 8:44:06 supplementary feed that has images in a 8:44:08 4x5, which is now the default image 8:44:12 ratio that Google wants to receive. The 8:44:14 other important consideration of images 8:44:16 is that the idea is that the image takes 8:44:19 up about 70% of the actual real estate 8:44:21 in a shopping listing. So, it is 8:44:22 arguably one of the most important 8:44:23 variables for being able to generate the 8:44:25 click. Other than obvious price 8:44:27 sensitivity on the price down the bottom 8:44:28 and then the title, which might sway 8:44:30 people, the image is really ultimately 8:44:32 what's going to generate whether someone 8:44:33 clicks or not. Now, do we necessarily 8:44:36 want to just maximize CTR on the 8:44:38 listing? Not necessarily. If we're 8:44:40 maximizing CTR on the wrong people, you 8:44:42 can think of the image the same as a 8:44:43 hook on Facebook. The idea of a hook 8:44:45 isn't to hook everyone. It's to hook the 8:44:47 right person. It's the same thing as 8:44:49 what Eugene Schwarz says with headlines. 8:44:52 The idea of a headline or the start of 8:44:53 an ad is not to get everyone to read, is 8:44:56 to get the right person to read. And the 8:44:57 same thing with the image. So, we want 8:44:59 to make the image as clickbaity as 8:45:01 possible, but to the right audience. 8:45:02 Now, within the restrictions of brand in 8:45:04 most businesses, what that actually ends 8:45:06 up looking like is just making the image 8:45:08 look different from the rest of the 8:45:10 listings that are appearing next to it. 8:45:12 So, how do we build contrast in the 8:45:14 imagery that we're flowing through into 8:45:16 the feed versus what the competitors are 8:45:17 doing? So, if the competitors, for 8:45:19 example, just have white backgrounds and 8:45:21 products getting placed on a photo 8:45:22 shoot, can we have a photo shoot? Sure, 8:45:25 but have the background be a different 8:45:26 color so that there's contrast built 8:45:28 within how we actually place across 8:45:30 Google Shopping. Besides the image and 8:45:32 the title, there's the description and 8:45:33 all the other attributes that need to 8:45:34 get cleaned up within the feed. That 8:45:36 should honestly already be in place. If 8:45:37 it's not, do it. It's a onetime setup 8:45:39 and then you're good. Another thing on 8:45:40 the feed is you do just want to make 8:45:41 sure you have multiple feeds for each 8:45:43 different currency. A big issue that I 8:45:44 see in a lot of audits is that all of 8:45:47 the different regions will just be 8:45:48 running through one feed with one 8:45:50 currency. Now, the issue there is that 8:45:51 if you're placing in other currencies, 8:45:53 it won't place in the localized 8:45:54 currency. it would do a live conversion 8:45:57 and show you the live conversion rate on 8:45:59 the shopping listing and say this has 8:46:00 been converted from AUD or this is being 8:46:02 converted from USD plus tax. That 8:46:06 immediately kills click-through rates 8:46:07 and it kills conversion rates because 8:46:08 people know that they're buying from an 8:46:10 international store when they might not 8:46:12 actually be. You just haven't set your 8:46:13 feeds up correctly. So, you finished the 8:46:14 video, you want to go away, you want to 8:46:16 start applying the stuff that you've 8:46:17 learned. What should the 90-day roll out 8:46:19 of these changes actually look like? 8:46:20 Number one, going back to the start of 8:46:22 the video, you want to make sure your 8:46:23 tracking is set up correctly. Make sure 8:46:25 you don't have J4 events. Make sure 8:46:27 you're triggering back events with 8:46:28 enhanced conversions on auto taggings 8:46:30 on. Everything's up to scratch. Number 8:46:32 two is you want to start building out a 8:46:35 structure map. So how many campaigns do 8:46:37 we think we need? We always the goal is 8:46:40 one and then what is the commercial 8:46:42 reason for each additional campaign? Are 8:46:43 we splitting brand versus nonbranded and 8:46:45 do we have naming consistency across 8:46:46 this structure that we're going to 8:46:47 implement? Number three is you should do 8:46:49 a Pmax specific check. You should check 8:46:51 that your asset groups are split by 8:46:53 product type and that the listing groups 8:46:54 are split out. You want to make sure 8:46:55 your brand exclusions on. You want to 8:46:57 make sure that audience signals and 8:46:58 stuff are present, but it's not 8:46:59 absolutely urgent. And you want to make 8:47:00 sure that your target rorowaz isn't 8:47:02 absurdly high, which is just causing it 8:47:04 to be a retargeting campaign. You want 8:47:05 to make sure that your bidding strategy 8:47:07 is the correct one based on everything 8:47:09 that I ran through in bidding 8:47:11 strategies. So, we ran through whether 8:47:12 you should be expansive or restrictive 8:47:14 and what the difference is between 8:47:15 maxing for conversions and maxing for 8:47:17 value. You want to go and audit and make 8:47:19 sure there's no wasted spend across the 8:47:22 account. Is there a campaign in there 8:47:23 that's just spending like $400 a month 8:47:25 and it doesn't even need to be there and 8:47:26 you could just roll the spend up? We'll 8:47:28 kill it. Is there a bunch of keywords 8:47:29 that are spending on search terms that 8:47:31 actually aren't profitable? Well, then 8:47:32 negative keyword them. Do you have 8:47:34 display enabled on the search campaigns? 8:47:36 Well, then turn it off. Then number six, 8:47:38 you want to do a health check on all the 8:47:39 ad assets that have been set up in the 8:47:41 account. Do we have site links? Do we 8:47:42 have call ads? Do we have promos that 8:47:44 aren't stale or outdated? Are we 8:47:46 maximizing assets within every single 8:47:48 campaign? Because this is ultimately 8:47:49 going to increase the real estate. So, 8:47:51 how much space your ads take up on the 8:47:53 landing page, which is always going to 8:47:55 be an easy quick win to improve 8:47:56 performance. And then lastly, we need to 8:47:58 do a feed check. So, you want to check, 8:48:01 are your titles optimized? Are your 8:48:03 images optimized? Can we do anything 8:48:04 more here? And if we can, do this in a 8:48:07 slow roll out. Don't suddenly roll all 8:48:09 the titles over tomorrow. Uh, if you do 8:48:11 that, it resets learning phases on every 8:48:13 campaign. You'll be in a bad position. 8:48:14 So, you want to slowly roll titles over 8:48:17 over time. See how the campaign's 8:48:18 reacting. See also in real time if those 8:48:21 listings are improving. So, look at 8:48:22 clickthrough rates, look at CPCs, look 8:48:23 at returns, see if the changes are 8:48:25 actually making a material impact. 8:48:27 Because if they're not, then we might 8:48:28 want to test some different titles with 8:48:29 some different keywords to make sure 8:48:31 that we can really maximize the account. 8:48:32 So, that's everything you need to know 8:48:34 on Google Ads for 2026. If you're a 8:48:36 performance marketer, email us at 8:48:37 hiringbluensedigital.com.au 8:48:39 to apply. [snorts] If you're a brand 8:48:41 doing over $5 million a month, click the 8:48:42 link in the description to book a free 8:48:44 audit. The single most common mistake on 8:48:46 an underperforming Google ad account is 8:48:48 that you either number one make a change 8:48:50 that you shouldn't have and things 8:48:52 actually get worse, number two you don't 8:48:54 make a change when you actually should 8:48:55 have or number three you make a bunch of 8:48:57 changes but the dip actually had nothing 8:48:59 to do with the platform in the first 8:49:01 place. The job of this 1hour training is 8:49:04 to teach you the order in which you 8:49:06 investigate what change you should 8:49:08 actually make. We'll start at the very 8:49:09 top of the business and then we'll drill 8:49:11 our way down each layer at a time. Ask 8:49:13 why at every layer and keep asking why 8:49:16 until we reach the root cause. Then you 8:49:18 can be confident in the changes that 8:49:20 you're actually making on Google that 8:49:21 they're making an impact and that you 8:49:23 needed to make them in the first place. 8:49:25 What this root cause analysis will 8:49:26 actually look like throughout the span 8:49:28 of this video is it will start at the 8:49:29 business level. We'll look at business 8:49:31 level KPIs, then move down to channel 8:49:33 checks, then go to campaign types, then 8:49:35 go into the specific campaign that we 8:49:37 think is causing the issue. Drill into a 8:49:39 metric, drill into a submetric, find the 8:49:41 cause of that submetric changing, and 8:49:44 then ultimately identify the root cause 8:49:45 so that we can make the appropriate 8:49:47 change to fix the business level KPI. 8:49:49 Now, there's three rules before you go 8:49:51 and touch anything. Number one is check 8:49:53 change history. Don't ever go and do an 8:49:55 audit or start making changes after 8:49:57 you've already just made a bunch of 8:49:59 changes. Also, if you're managing the 8:50:01 account with multiple people or even if 8:50:03 the client might have gone in and 8:50:04 tweaked something, you need to be across 8:50:05 it. So, make sure to always check change 8:50:07 history before you even start this 8:50:09 process in the first place. Number two 8:50:10 is you want to extend time horizons. And 8:50:13 so, what a lot of people will do is 8:50:14 they'll look at two short time horizons 8:50:16 to be able to see any kind of trend. So, 8:50:18 they end up working off just small 8:50:20 sample size bias and they make decisions 8:50:21 that they shouldn't. So don't look at a 8:50:23 1-day period of performance and then go 8:50:25 in and go through this whole diagnostic 8:50:27 process. Make sure that you're zooming 8:50:28 out as much as possible contextual to 8:50:30 the conversion volume of the account. 8:50:32 And then number three is check 8:50:34 conversion latency. So this is an issue 8:50:37 that just exists in Google, which is 8:50:39 that when a conversion occurs, the 8:50:41 conversion by default doesn't actually 8:50:43 get attributed to the day in which they 8:50:45 purchase, but instead it gets attributed 8:50:47 to the day in which they click. And so 8:50:48 if I just give you a quick timeline in 8:50:50 case you're unfamiliar with this, let's 8:50:51 say over here on the first of the month 8:50:53 I click on an ad. Then on the 3rd of the 8:50:55 month I click on another ad on Google 8:50:57 and then by the 7th I actually end up 8:50:59 buying. This is the moment of purchase. 8:51:01 Well, what happens in most platforms is 8:51:03 the purchase will then get attributed 8:51:05 into the platform on the 7th. And when 8:51:07 you open up and you look at the seventh, 8:51:09 you'll see oh purchase happened here. On 8:51:10 Google instead, the purchase actually 8:51:12 gets attributed back here. And so what 8:51:14 then occurs is that when you're looking 8:51:16 at the last 7 days or the last 3 to 4 8:51:18 days of data is rorowaz always looks 8:51:20 terrible. Conversion volume always looks 8:51:22 terrible because we haven't given all of 8:51:24 these clicks enough time to purchase in 8:51:26 the future and then get attributed 8:51:28 backwards. And so if you're beginner 8:51:30 level at Google or you're just opening 8:51:31 up Google for the first time, you'll 8:51:32 pretty much always see the last 30 days. 8:51:34 If you look at rorowaz, if you look at 8:51:35 conversion value, it'll be relatively 8:51:37 stable and then the last few days it 8:51:38 just drops off. And if you don't know 8:51:40 that this is what's occurring, you'll go 8:51:41 into the platform and go, "The last 8:51:43 three days are a disaster. The last 8:51:44 three days are always a disaster in 8:51:46 every single account that occurs." Now, 8:51:47 there is a way to get around this. And 8:51:49 what you do is you pull out a custom 8:51:51 column called conversions by conversion 8:51:55 time. So, the core keyword here to 8:51:57 always be looking for is conversion 8:51:59 time. Whenever you're looking at 8:52:00 conversion time, it is the day in which 8:52:02 they actually purchase, not attributing 8:52:04 backwards to the click. So when you use 8:52:06 this, you will get a better read on the 8:52:08 last 3 to four days of real-time 8:52:10 performance and therefore be able to 8:52:11 make better decisions and not get looped 8:52:13 into a diagnostic process that actually 8:52:15 is to do with conversion latency rather 8:52:17 than anything within the business 8:52:18 actually getting worse. So starting off 8:52:20 at level one on business signal, the 8:52:22 first thing we need to understand is 8:52:24 Google ads actually broken and the first 8:52:26 way that we check this is we do a cross 8:52:28 channel check. Now, a really extreme 8:52:30 example of this would be, let's say, 90% 8:52:33 of your ad spend is on Meta and only 10% 8:52:36 is on Google. Well, when revenue drops, 8:52:38 when there's a dip in efficiency, when 8:52:40 any top level business KPI decreases, 8:52:43 it's probably to do with Meta and not to 8:52:45 do with Google. Just considering that 8:52:47 Google is only driving 10% of ad spend, 8:52:49 probably half of this is just branded. 8:52:51 And so, maybe there's 5% of new customer 8:52:53 acquisition coming through Google as a 8:52:54 platform. If revenue dips, probably not 8:52:56 to do with Google. So don't go and 8:52:57 troubleshoot the platform because of it. 8:52:59 So we need to have more of a omni 8:53:01 channel understanding of the media mix 8:53:03 to be able to go, okay, what's happening 8:53:04 on the other platforms? Could something 8:53:06 on the other platforms be causing the 8:53:08 dip? Now let's go into a more realistic 8:53:10 scenario where maybe you're 65 Meta and 8:53:13 then 35 Google. Well, now if there's a 8:53:15 revenue dip, it's probably to do with 8:53:17 Meta, but we can't guarantee it. Could 8:53:18 also be to do with Google. And so we 8:53:19 need to do our due diligence. Yes, we're 8:53:21 going to go and troubleshoot Meta and 8:53:22 that's a whole another process in 8:53:23 itself. But we also need to do some 8:53:25 troubleshooting on Google and be like, 8:53:26 okay, does a revenue dip have anything 8:53:28 to do with this portion of the media 8:53:30 span. But before we even do that, we 8:53:32 need to start asking ourselves questions 8:53:34 on this side first, which is did 8:53:36 anything materially change on meta, 8:53:39 within the business, on Pinterest, on 8:53:41 Tik Tok. Did anything happen outside of 8:53:43 Google that could have caused this? 8:53:45 That's obvious. Did we come off the back 8:53:47 of a sale? Did we do a complete creative 8:53:50 refresh on Meta and then right after 8:53:52 that performance fell off? Did we change 8:53:53 the offer on Meta? Did we do something 8:53:56 within the business that would have 8:53:58 primarily driven this decrease in 8:54:00 performance and broken something? Once 8:54:01 we have answered that question, and this 8:54:03 shouldn't be a two-c should be an 8:54:05 investigation in itself. If your agency 8:54:08 side asks the client, did anything 8:54:09 happen within the business that we're 8:54:11 not across? We then also need to do our 8:54:12 own due diligence. Go into Meta, go into 8:54:14 Tik Tok, look at change history, see 8:54:16 what's happened. Check that the website 8:54:17 isn't broken. Ultimately understand of 8:54:20 the revenue equation which metric 8:54:22 decreased. So revenue equals conversion 8:54:24 rate time average order value time 8:54:26 sessions. So of these three metrics, 8:54:28 which one changed? Did average order 8:54:30 value decrease? Did conversion rate 8:54:31 decrease? Or did sessions decrease? Now 8:54:33 if conversion rate decreased, we want to 8:54:35 have a look at the website. If sessions 8:54:36 decreased, it's probably to do with 8:54:37 where the traffic is coming from and one 8:54:39 of the platforms has decreased 8:54:40 materially in click volume. And so we 8:54:42 need to figure out, okay, where's that 8:54:43 click volume falling off from? If it's 8:54:45 an average order value change, this 8:54:46 might be a change in the product 8:54:47 portfolio prioritization, which might be 8:54:49 directly in Google Ads. Okay, some 8:54:51 products might be getting pushed more 8:54:52 than others all of a sudden. Or this 8:54:53 might be to do with a website change in 8:54:55 terms of a structured upsell or 8:54:56 cross-ell that's changed materially. So 8:54:58 we want to do all this investigation on 8:55:00 these three numbers, figure out which 8:55:02 one changed, what could have changed it, 8:55:03 does it have anything to do with the 8:55:05 other platforms before we even get to 8:55:07 Google. Now the prioritization of 8:55:08 platform checking is just based on 8:55:10 spend. So let's say that in this example 8:55:13 actually 65% of spend was on Google and 8:55:16 35 was on meta. Well then we would 8:55:17 obviously do a business level diagnostic 8:55:20 which of the three metrics in the 8:55:21 revenue equation have gone down and then 8:55:23 we would check Google first. Okay Google 8:55:24 would be the first one that we do and 8:55:25 then we go through this whole process. 8:55:27 But for most people at least 80 to 90% 8:55:30 of the businesses that we work with and 8:55:31 that we order and that we interact with 8:55:33 on a day-to-day basis majority of their 8:55:34 spend skews to meta. Therefore you 8:55:36 should do the meta analysis first and 8:55:38 then you should go to Google. Now after 8:55:39 you do the cross channel check and you 8:55:41 look at the revenue equation and you 8:55:42 understand which of these metrics 8:55:43 decreased the next step at this level is 8:55:46 to understand is this an attribution 8:55:48 metric that is decreased or is it a true 8:55:49 business KPI. So is it revenue? Is it 8:55:52 profit contribution? Is it me or a any 8:55:55 efficiency level number that's indexed 8:55:56 on the actual revenue of the business? 8:55:58 Or are we talking about some kind of 8:56:00 attributed number that could have some 8:56:02 conversion time lag or where there could 8:56:04 just be an attribution error in the 8:56:06 setup for whatever reason tracking is 8:56:08 broken. That's why it looks bad. Has 8:56:09 nothing to do with the actual business. 8:56:10 So we always want to reconcile these two 8:56:12 numbers too in real time which is if it 8:56:15 is a true business KPI that's decreased 8:56:16 which is how we operate. We would only 8:56:18 look at this. We would rarely look at 8:56:20 this as a means to go through a 8:56:21 diagnostic process. And let's say 8:56:23 revenue has decreased. Well, then the 8:56:25 quick check to do is on all of our 8:56:27 attribution figures, does anything 8:56:28 correlate? What's gone down? Is meta 8:56:31 suddenly attributing way lower and we've 8:56:32 lost one in terms of return on ad spend. 8:56:34 So, it's dipped from four to three. Has 8:56:35 Google suddenly dipped off? Like what 8:56:37 has occurred? And make sure that we're 8:56:38 looking at conversion time, not default 8:56:41 conversions within Google, which will 8:56:42 just always show that the last 3 to 4 8:56:44 days have been bad. And then the final 8:56:46 level of this analysis is seasonal 8:56:49 patterns. So understanding is the dip to 8:56:52 do with seasonality. And this is 8:56:53 ultimately why accurate forecasting is 8:56:56 so critical and why it's a component of 8:56:58 what we do at BlueSense for clients 8:57:00 because we need to understand if revenue 8:57:02 decreases is this in line with just the 8:57:04 forecast and expectation of seasonality 8:57:06 within the business. And if so, okay, 8:57:07 cool. That is what we expected to 8:57:09 happen. And so there's no need to go 8:57:10 through a 5-hour diagnostic process to 8:57:12 try to figure out what's going on. If 8:57:14 you don't have forecasting in place, 8:57:15 then seasonality can just get you and 8:57:17 you end up wasting so much time trying 8:57:18 to diagnose what's actually going on 8:57:20 when the reality is this is just a 8:57:21 seasonal fluctuation that always will 8:57:23 occur in the business. So this is level 8:57:25 one. Ultimately, depending on how good 8:57:27 you are, this should take you about 5 8:57:29 minutes. You should be able to do the 8:57:31 cross channel check. You should be able 8:57:32 to prioritize other platforms first. You 8:57:34 can look at the revenue equation and 8:57:35 understand which lever is causing the 8:57:37 impact. Look to correlate attributed 8:57:39 numbers with true business KPIs. Has it 8:57:42 just been blatantly obvious that yeah, 8:57:44 attribution in meta has fallen off a 8:57:46 cliff and revenues fallen. Okay, it's 8:57:48 probably meta and then understanding of 8:57:50 seasonality and forecasting. So this is 8:57:52 the first step and often you won't get 8:57:54 past this step because on most 8:57:55 businesses where Google is 20 to 25 to 8:57:58 30% of spend, Google usually isn't the 8:58:00 reason why revenue is decreasing. In 8:58:02 fact, Google is one of the most 8:58:03 consistent platforms out of all of the 8:58:05 advertising platforms. Meta's all over 8:58:06 the place. Okay, you do a creative 8:58:08 refresh. You might get a winner. 8:58:09 Suddenly, you can triple ad spend 8:58:11 overnight. Your winner suddenly 8:58:12 fatigues. Ad spend has to pull back. 8:58:14 Efficiency falls off a cliff. Okay, meta 8:58:16 is all over the place. Particularly if 8:58:17 you don't have a consistent process in 8:58:19 place to be able to introduce creatives 8:58:20 within a testing structure, scale 8:58:22 conservatively and use portfolio 8:58:24 management when it comes to creatives, 8:58:26 which by the way, we have a whole video 8:58:27 on. It's 2 and 1/2 hours. It's called 8:58:29 creative strategy in 2026. I recommend 8:58:32 you watch it if you want a better 8:58:33 understanding of the metaite on Google. 8:58:35 Let's say you make a pass this step. 8:58:36 nothing has happened anywhere else. It 8:58:38 seems to be definitely a Google issue. 8:58:40 Well, then that takes us into level two 8:58:42 and three. You drill into Google Ads the 8:58:44 same way every time, which is that you 8:58:46 start at the account level first. You 8:58:48 sort by campaigns by cost descending. 8:58:51 So, you want the highest spending 8:58:52 campaigns at the top. The campaign 8:58:53 you're looking for is somewhere in the 8:58:55 top five by spend. Below the top five 8:58:58 spending campaigns, even if you doubled 8:59:00 performance on them or performance fell 8:59:02 off a cliff, it wouldn't actually 8:59:03 materially move the account. I'll give 8:59:05 you an example. If you have a campaign 8:59:06 that's holding, let's say, 10% of 8:59:08 budget, even if you got a 100% increase 8:59:12 in performance on this campaign, it's 8:59:14 not going to significantly impact the 8:59:16 account really at all. It's going to 8:59:18 have a 5% impact. And then if Google is 8:59:20 only 40% of your media mix, we're 8:59:22 talking about singledigit percentages. 8:59:24 And so, if there's a large material 8:59:25 change at a business level, it is not to 8:59:27 do with a campaign that's holding 10% of 8:59:30 your Google spend, which is why you want 8:59:31 to go straight to the top and start at 8:59:33 the highest spending campaigns. Now, is 8:59:35 there a reality where this could be the 8:59:36 reason why the business dropped? For 8:59:38 sure, but it's just not likely. And so, 8:59:39 we want to start at the most likely 8:59:41 reasons as to why performance has dipped 8:59:43 and then move our way through into 8:59:45 checking all the small things that 8:59:47 realistically is probably just a waste 8:59:48 of time. Hence why we want to dep 8:59:50 prioritize them. Here's the 5minute 8:59:52 check process. Number one, take your 8:59:54 date range and look at the last 90 to 8:59:56 180 days and switch to a weekly view. 8:59:58 Then you can just stay in the overview 9:00:00 tab of Google, so top left, and you can 9:00:02 just look at all the graphs. And what 9:00:04 you want to do is rotate through each 9:00:06 different metric here and understand how 9:00:08 it's moving. You want to look at cost. 9:00:10 Has cost materially changed over the 9:00:12 course of the last 90 days. Conversions, 9:00:14 has conversions changed, conversion 9:00:16 value, cost per conversion, return on ad 9:00:18 spend, which is conversion value divided 9:00:20 by cost, and then click-through rate. 9:00:22 Have any of these metrics materially 9:00:24 changed when you're looking at them in a 9:00:26 graph view over the course of the last 9:00:28 90 to 180 days? What has gone up? What 9:00:30 has gone down? What has gone sideways? 9:00:32 From there, you want to go away from the 9:00:34 overview tab and you want to go into the 9:00:36 campaign tab on the left. And now we 9:00:39 want to drill down at an individual 9:00:40 campaign level. And so we're looking at 9:00:42 each individual campaign. Once again, 9:00:44 the top five campaigns, the top five 9:00:45 biggest spenders, start at the top, work 9:00:46 your way down. Which campaign has 9:00:48 contributed the most to these metrics 9:00:50 changing. So let's say conversions has 9:00:52 slowly pulled off and it happened on a 9:00:54 specific date. So you want to identify 9:00:55 when is the inflection point? Is it the 9:00:57 3rd of March? And then from the 3rd of 9:00:59 March onwards, did it start trailing 9:01:01 off? And what you'll find if this is a 9:01:03 Google Ads issue is that maybe there is 9:01:05 one or two campaigns where conversions 9:01:07 or rorowaz or click-through rate started 9:01:09 declining after a particular date or 9:01:11 there was a material impact. From there 9:01:13 we drill one step deeper which is we 9:01:15 then go down to the ad group level. If 9:01:17 there are multiple ad groups or in Pmax 9:01:20 asset groups we do the exact same 9:01:22 exercise. We look at all of these 9:01:23 metrics. We look over the time period 9:01:25 and we go which asset group or ad group 9:01:28 has contributed to this decline in the 9:01:30 metric. Is it all of them? Is it just a 9:01:32 specific one? Cool. Now we know exactly 9:01:34 what in the account has caused a 9:01:37 material change in a topline KPI where 9:01:39 we can now start to move through and 9:01:41 identify what metric has caused it, what 9:01:43 submetric and then what is the root 9:01:45 cause. Now the real two keys that we 9:01:47 want to know off the back of this is 9:01:49 number one, what campaign type has 9:01:52 fallen off? Is this Pmax campaigns? Is 9:01:54 this shopping campaigns? Is this search 9:01:56 campaigns? Is this display? Is this 9:01:57 YouTube? etc. and then off the back of 9:01:59 that which specific campaigns and which 9:02:02 specific ad groups. Now the reason why 9:02:04 the campaign type matters is because the 9:02:07 failure modes of each campaign type is 9:02:10 different. So a Pax campaign will 9:02:12 decline for different reasons to a 9:02:14 shopping campaign. A shopping campaign 9:02:15 will decline for different reasons than 9:02:17 a search campaign. And so it's important 9:02:18 for us to understand the campaign type 9:02:20 and then we can go into troubleshooting 9:02:22 it. So what are the specific failure 9:02:24 modes of a PMAX campaign? Number one is 9:02:27 warm and cold drift. So because 9:02:29 performance mass campaigns can retarget 9:02:31 people and it can place across all the 9:02:33 different channels. What can happen is 9:02:35 Pass campaigns can start retargeting 9:02:37 people more or start going into cold 9:02:39 targeting more dynamically at its own 9:02:41 will. As an example, the campaign might 9:02:43 be labeled a cold campaign, but it now 9:02:45 starts serving mostly to existing 9:02:46 customers. And so you want to go and 9:02:48 check the search term report. Now brand 9:02:50 should be excluded anyway, but it's good 9:02:52 to just double check. And then number 9:02:53 two is you want to go and check the 9:02:55 audience report. If brand terms have 9:02:57 started to get introduced into the 9:02:58 campaign or if the audience report shows 9:03:00 a skew towards existing customers then 9:03:02 the brand exclusion is broken or the 9:03:05 target rowaz has been pushed too high on 9:03:08 the account on this campaign sorry which 9:03:10 is causing it to rep prioritize warm 9:03:12 audiences. Number two is feed 9:03:13 disapprovals. In e-commerce specifically 9:03:15 80 to 90% of the performance in Pmax is 9:03:17 going to come from shopping and shopping 9:03:19 performance is going to come from feed 9:03:20 quality. And so if something has gone 9:03:22 down in the feed, if a product has gone 9:03:24 down, if there's some kind of 9:03:25 disapproval that's occurred, that's 9:03:27 obviously going to be a main contributor 9:03:28 to why the campaign's performance 9:03:30 started to decrease. So as a product of 9:03:32 that, you want to go and look at the 9:03:34 feed. So you want to check products just 9:03:36 within the campaign, see if any top 9:03:38 spending products have gone out of stock 9:03:40 or have been turned off or being 9:03:42 disapproved. And then number two is go 9:03:43 into the GMC as well and just double 9:03:45 check everything there. Number three is 9:03:47 high rorowaz narrowing the targeting. 9:03:49 This is something that we've gone 9:03:50 through in the other Google video that 9:03:51 we put out. But as you increase target 9:03:54 rorowaz on a campaign, it doesn't 9:03:56 magically mean that you just suddenly 9:03:57 get better efficiency on a campaign. 9:03:59 That's not really how it works. What's 9:04:00 actually happening is that Google is 9:04:02 narrowing the targeting to a smaller 9:04:05 subset of buyers that are more likely to 9:04:06 convert at a higher efficiency. So, it's 9:04:08 only going to enter auctions where it 9:04:10 knows it's going to win and that person 9:04:11 is a high likelihood to purchase. When 9:04:13 it does that, what typically happens is 9:04:16 that you're just narrowing in on a 9:04:18 warmer audience. And so if you start 9:04:19 increasing and increase and increasing 9:04:21 target rorowaz, it's just going to 9:04:22 narrow the pool of audience that you're 9:04:23 targeting. Spend will likely pull back 9:04:25 and you'll probably get worse new 9:04:27 customer acquisition in the campaign. 9:04:28 You can actually double check this with 9:04:29 third party attribution tools these 9:04:31 days. You can also just see it natively 9:04:32 within Google ads too. But if you look 9:04:34 at any NC rorowaz numbers on like a 9:04:36 triple whale or any tool that you use, 9:04:38 uh NC rorowaz numbers on a pmax campaign 9:04:41 with a high target rorowaz will 9:04:43 typically be really bad. You could have 9:04:44 the exact same campaign and have a low 9:04:46 target rorowaz and NC rorowaz will be 9:04:48 better. And so the actual target rorowaz 9:04:50 is not indicative of the performance of 9:04:52 the campaign on cold audiences. It is 9:04:53 actually indicative of how hard the 9:04:55 campaign will go on just retargeting 9:04:57 warm people. And then number four here 9:04:59 is product bloat. What this means is 9:05:01 that and this isn't going to be 9:05:02 applicable to most people but if you're 9:05:03 rapidly increasing the skew count on the 9:05:05 website and all of these SKs and new 9:05:07 products are flooding into this 9:05:09 campaign, you can just end up with so 9:05:10 many products in the campaign that it 9:05:12 impacts learning. A subset point of this 9:05:15 is that if you have products going in 9:05:17 and out of stock all of the time, that 9:05:19 also resets learnings of the campaign, 9:05:21 particularly if it's a high spending 9:05:22 product. Let's say a product is holding 9:05:24 15% of total spend in the PMAX campaign 9:05:26 and this product went out of stock for 4 9:05:28 days and then come came back in stock. 9:05:30 This is not a good position to be in 9:05:32 because the learning phase of this 9:05:34 individual product resets, but it also 9:05:36 impacts the overall campaign as well. 9:05:38 And so we actually have in one of our 9:05:39 onboarding videos for clients a 9:05:41 disclaimer around this exact point which 9:05:43 is that if you have products that go in 9:05:44 and out of stock all the time, please 9:05:46 let us know because it might materially 9:05:48 change the way that we decide to 9:05:49 structure the account because we don't 9:05:50 want one product going in and out of 9:05:52 stock impacting the performance of all 9:05:53 the other products that sit in the same 9:05:55 campaign. If it is a search campaign 9:05:57 that has failed, number one, you want to 9:05:58 look at search terms. So what search 9:06:00 terms are we spending on and has this 9:06:02 materially changed over the course of 9:06:04 time from the inflection point? So there 9:06:06 was a point in time in which the 9:06:07 campaign stopped performing. What 9:06:08 happened before and after in the search 9:06:11 term portfolio of key terms that are 9:06:13 getting most spend and has anything 9:06:14 materially changed? So you go into the 9:06:16 search term report of the search 9:06:17 campaign. You look at the time period 9:06:19 beforehand. You sort by spend and you go 9:06:21 okay what are our top spending search 9:06:22 terms here and what was efficiency and 9:06:24 now we look at after that time period 9:06:26 and we go what does it look like now? 9:06:28 Has it materially changed? If it hasn't 9:06:29 materially changed if all the search 9:06:30 terms look similar performance looks 9:06:32 similar across them then you can move 9:06:33 on. But often what can be the case 9:06:35 particularly if you're using broad match 9:06:37 and smart bidding strategies is it might 9:06:39 be a wild change in search term uh spend 9:06:42 allocation which has actually caused the 9:06:43 impact in performance. Number two you 9:06:45 want to look at auction insights. It 9:06:46 might just be the case that some 9:06:47 competitors came in and launched search 9:06:49 campaigns and have started to drown you 9:06:51 out of the auction which has increased 9:06:52 CPCs or has decreased your ability to 9:06:56 spend because you're no longer entering 9:06:57 auctions correctly. Um you're sorry 9:06:59 you're no longer entering as many 9:07:01 auctions. Number three is target 9:07:03 rorowaz. Exact same thing as with pmass 9:07:05 campaigns. If you just started to 9:07:06 squeeze this up, it would have narrowed 9:07:08 the audience targeting. Number four, you 9:07:09 want to make sure display is turned off. 9:07:12 Or else you could have just had spend 9:07:14 getting allocated into display which was 9:07:16 causing a performance drop off. In 9:07:18 regards to shopping, it's all the same 9:07:19 stuff as Pmax except for the warm versus 9:07:22 cold. So you've got Troz, you've got 9:07:24 GMC. This is likely the biggest and most 9:07:26 important thing to check, right? Has the 9:07:28 products gone out of stock? Did we 9:07:29 change titles or descriptions recently? 9:07:32 Is there any kind of errors or warnings 9:07:34 within the GMC feed? Do we have now too 9:07:36 many products flooding into the 9:07:38 campaign? One that's probably not 9:07:40 causing anything, but it's worth 9:07:41 checking is is there a lot of spend 9:07:43 getting distributed into search partners 9:07:45 and is this performing poorly? That then 9:07:47 takes us into level four and level five. 9:07:51 The important concept to understand at 9:07:53 this stage when we drill into a specific 9:07:54 campaign and we start looking at one 9:07:56 metric is that every campaign level KPI 9:07:58 that you care about like revenue like 9:08:00 rorowaz like even conversion rate is the 9:08:03 product of two or three underlying 9:08:05 metrics. And this is why understanding 9:08:07 the formulas that constitute every 9:08:09 metric in e-commerce becomes really 9:08:11 helpful for being able to troubleshoot 9:08:13 and do bottleneck analysis. If you 9:08:14 understand for example that average 9:08:16 order value isn't just average order 9:08:19 value but it is a blend of new customer 9:08:21 average order value and returning 9:08:22 customer average order value and then 9:08:24 you understand that each of these sits 9:08:25 on distribution curves and then you 9:08:27 understand that the actual mechanics 9:08:29 that impact each individual order here 9:08:30 is units per transaction and average 9:08:32 unit retail then all of a sudden you can 9:08:34 go through a troubleshooting process on 9:08:36 average order value that is much more 9:08:38 in-depth comprehensive and aligned to 9:08:40 the actual root cause than anyone else 9:08:42 because some people will try to 9:08:43 troublesoot average order value by just 9:08:45 looking at oh what offers change but if 9:08:47 you understand that no actually we need 9:08:48 to drill into NC because that's the one 9:08:50 that dipped then we need to look at the 9:08:51 distribution curve and how that changed 9:08:52 and then this part of the distribution 9:08:54 curve changed what actually caused the 9:08:55 change was it units per transaction or 9:08:57 average unit retail was up what 9:08:59 mechanically caused the dip in and then 9:09:01 we can actually troubleshoot this which 9:09:03 is the root cause this isn't the root 9:09:05 cause and so the deeper you can go in 9:09:07 your metric understanding ultimately the 9:09:09 better you will be at being able to do 9:09:10 root cause analysis and actually 9:09:12 troubleshoot the business and So in 9:09:14 Google Ads, if revenue is down, 9:09:15 attributed revenue, it's because of cost 9:09:17 per click conversion rate and average 9:09:18 order value because that is ultimately 9:09:20 the revenue equation, but we're taking 9:09:22 clicks and we're going into CPC in the 9:09:24 platform. If conversions are down, it's 9:09:26 because either clicks drops, conversion 9:09:28 rate dropped, or tracking broke in some 9:09:30 way, and that's why the attributed 9:09:32 conversions aren't there anymore. The 9:09:33 job at this layer of troubleshooting is 9:09:36 to figure out which of the underlying 9:09:38 metrics moved because each one points to 9:09:40 a different cause. There's really two 9:09:42 formulas that you want to keep top of 9:09:44 mind. Now, I could write infinite here 9:09:45 and every metric derives into 9:09:47 submetrics, but these are the two that 9:09:49 you want to keep top of mind because 9:09:51 this will get you 90% of the way most of 9:09:54 the time, which is CPA equals CPC 9:09:57 divided by conversion rate. So, these 9:09:59 are the two submetrics of cost per 9:10:01 acquisition. And then clicks equals 9:10:03 impressions times by click-through rate. 9:10:05 And so if click volume goes down, it's a 9:10:07 product of either impressions 9:10:08 compressing, which means we're entering 9:10:10 less auctions and showing uh less of a 9:10:12 degree, or click-through rate is down, 9:10:14 which means that other people's listings 9:10:16 on shopping on search, whatever, are 9:10:17 suddenly more convincing than ours, and 9:10:19 so we're losing out on the actual clicks 9:10:20 in the auctions that we're entering. So 9:10:22 these are the first two that I would 9:10:24 actually start with in terms of 9:10:25 troubleshooting, which is, is our 9:10:27 rorowaz and efficiency down? Is that the 9:10:29 issue? Are we seeing a drop off in 9:10:31 performance, but volume is relatively 9:10:33 the same in terms of click volume? Okay, 9:10:35 cool. Then we need to drill into CPCs 9:10:37 and conversion rates. Now, what are the 9:10:39 common causes of CPCs being down? Number 9:10:42 one is that your quality score could 9:10:44 have materially changed. The way that 9:10:46 you check this is you open up keywords, 9:10:48 you add quality score as a column, you 9:10:50 open up keywords, and you add the 9:10:52 following columns. You add expected CTR, 9:10:57 you add ad relevance, and you add 9:10:59 landing page experience. If any of these 9:11:03 three have materially changed post 9:11:06 intervention, so post the point in which 9:11:08 performance dropped off, then that is 9:11:10 the thing that you need to troubleshoot. 9:11:11 Ad relevance is often a really easy 9:11:13 problem to solve. You just make need to 9:11:15 make the ad more relevant to whatever 9:11:16 search terms are actually being placed 9:11:17 on. Expected CTR is a product of actual 9:11:21 current CTR and then it's just 9:11:22 extrapolating it into the future. And so 9:11:24 this just means that your ads need to be 9:11:26 better and you need to improve them. uh 9:11:27 landing page experience is actually the 9:11:29 hardest one to materially change because 9:11:30 it actually requires dev work and the 9:11:33 reync cycle on this. So how often Google 9:11:36 will scrape the website and then 9:11:37 actually change this score is kind of 9:11:40 pretty unknown. Uh I've been in 9:11:41 positions myself where landing page 9:11:43 experience has been the bottleneck and 9:11:44 what actually caused a dip in 9:11:45 performance and it's taken weeks if not 9:11:48 months to actually be able to turn it 9:11:49 around. We fixed everything on the 9:11:51 website but Google just wasn't 9:11:53 rescanning uh the website and rescoring 9:11:56 us. Now number two is a new competitor 9:11:58 could have entered the auction and this 9:12:00 is driving up CPCs. And so the way to 9:12:02 check this is really easy. You just go 9:12:04 into auction insights and you can look 9:12:06 at auction insights prior to uh the 9:12:09 point in which performance dipped and 9:12:10 then auction insights after the point in 9:12:12 which performance dipped and then you 9:12:13 can see has someone new gone and entered 9:12:15 the auction. And then number three is 9:12:17 you might actually just be placing on 9:12:19 premium auctions. What this means is 9:12:21 that you might have made a change to 9:12:23 target rorowaz or any of your bidding 9:12:25 strategies and as a product of that you 9:12:27 are now entering into more expensive 9:12:29 auctions. So it's becoming more 9:12:30 expensive to get clicks. Not necessarily 9:12:33 a bad thing because they might be higher 9:12:36 quality clicks. And so this is actually 9:12:38 an important thing to understand about 9:12:40 CPCs in Google ads is that it's 9:12:42 relatively a vanity metric. In fact, 9:12:44 it's kind of a vanity metric as well in 9:12:46 meta and Tik Tok. And the reason being 9:12:48 is that it's not directly associated 9:12:49 with performance except on its fringes. 9:12:52 So if CPCs are super high, yeah, it's an 9:12:54 issue. If CPCs are super low, yeah, it's 9:12:56 an issue. But CPCs can go up and as long 9:12:59 as conversion rates go up as well, we're 9:13:02 all good because CPA stays the same. The 9:13:04 issue becomes when CPCs go up and 9:13:06 conversion rates stay the same, then 9:13:08 it's like, oh, okay, well, it might be a 9:13:10 quality score or competition or a 9:13:11 premium auction issue. and more 9:13:13 specifically, we're entering premium 9:13:14 auctions, but it's not paying off. We're 9:13:17 not getting higher quality clicks and 9:13:19 higher quality users to the website, and 9:13:20 so it isn't worth it. Now, on conversion 9:13:23 rate being down, the first thing that 9:13:24 you want to check is just tracking. Make 9:13:26 sure has tracking stopped working. Are 9:13:29 you looking at the last short time 9:13:30 period, but you're not looking at 9:13:32 conversion by time? Those are going to 9:13:34 make conversion rate look bad, but it 9:13:35 actually has nothing to do with anything 9:13:36 in the platform. Number two, has the 9:13:38 landing page materially changed? Now, 9:13:41 number one, have you actually changed to 9:13:43 a different URL? Number two, have you 9:13:45 made changes materially to the URL? Has 9:13:47 the price changed? Is there some kind of 9:13:49 redirect in place that's now pushing to 9:13:50 a different website? Has there been 9:13:52 upsells or crossells added? Has the 9:13:54 merchandising on the website changed? 9:13:56 Ultimately, if conversion rate is 9:13:57 decreasing, this is probably the number 9:13:59 one reason as to why. Number three, has 9:14:01 the audience changed? You can have the 9:14:03 same landing page, same website, but if 9:14:05 you have higher quality users coming 9:14:06 from more premium auctions, your 9:14:08 conversion rate will be better. vice 9:14:09 versa. It can be worse if you have worse 9:14:11 quality audiences. So, you want to look 9:14:13 into search term reports there. Have the 9:14:15 search terms that we're driving people 9:14:16 through changed? Are they coming from 9:14:18 different keywords that might not be 9:14:19 converting as well? Because maybe 9:14:20 there's not congruency there with the 9:14:22 landing page. Have we materially changed 9:14:23 the bidding strategy that might be 9:14:25 pushing towards and optimizing towards a 9:14:26 different audience? What's going on 9:14:28 here? Number four is unfortunately the 9:14:30 learning phase is a thing on Google Ads. 9:14:33 It's actually a very big thing and a 9:14:34 very annoying part of the platform. And 9:14:35 so, has the learning phase reset in some 9:14:37 capacity? Have we made some large 9:14:39 structural change to the account that's 9:14:41 pushed it back into learning phase which 9:14:42 will always decrease conversion rates 9:14:44 because you're going to be entering into 9:14:45 the wrong auctions and be driving poor 9:14:47 quality traffic to the website. Number 9:14:49 five, and this is off really the back of 9:14:51 the landing page, but has the offer 9:14:53 changed? Has a promo ended? Has a 9:14:55 discount being removed? Has a free 9:14:56 shipping threshold changed on the 9:14:58 website? Because this is also going to 9:15:00 impact conversion rates. Now, let's say 9:15:02 it's not a CPA issue. And so, our 9:15:03 efficiency is the same. It hasn't 9:15:05 changed, but instead volume has changed. 9:15:07 We're now either not spending as much or 9:15:09 we're not getting as much revenue. Well, 9:15:11 that is usually a function of clicks 9:15:13 falling off. We're driving less volume 9:15:15 out of Google Ads to the website and as 9:15:16 a product of that, we're doing less 9:15:18 volume. So, the subset of clicks, the 9:15:20 submetrics is impressions and 9:15:22 click-through rate. On impressions, 9:15:23 there's two types of reasons as to why 9:15:25 you lose impressions or you lose 9:15:27 impression share. Number one is 9:15:29 impression share lost to budget. You 9:15:30 just don't have the budget and so you 9:15:32 can't get more impressions cuz you're 9:15:33 maxed out. there's no more spend to be 9:15:35 able to get you more impressions. Number 9:15:36 two is impression share loss to the bid 9:15:39 and so your bid just isn't high enough 9:15:40 to enter enough auctions to actually 9:15:42 spend the money. You want to identify 9:15:45 which one it is and the way to identify 9:15:47 it is pretty easy. You're not losing 9:15:49 impression share to budget. Sorry, you 9:15:51 are losing impression share to budget if 9:15:53 you're hitting the budget. So if you 9:15:54 have a $100 a day budget and you're 9:15:56 hitting it, well the reason why you're 9:15:57 not getting more impressions is likely 9:15:59 because you just need to increase spend. 9:16:00 Now you might be thinking our spend has 9:16:02 always been the same. Why has our 9:16:03 impressions changed? The impressions 9:16:05 might have changed because now you have 9:16:06 higher quality users. So CTR has gone up 9:16:09 which is offsetting click volume. Now if 9:16:10 click volume is down and impressions are 9:16:13 down, I can almost guarantee it has 9:16:15 nothing to do with your budget because 9:16:16 your impressions and click volume is now 9:16:18 getting throttled likely due to the bid. 9:16:21 Now the bid is a product of all three of 9:16:23 these things. It's a product of quality 9:16:25 score, new competitors, and how 9:16:27 aggressively you're entering auctions. 9:16:28 You want to go back and troubleshoot all 9:16:30 of these to be able to fix impressions I 9:16:32 lost a bit. On CTR decreasing, it's 9:16:35 usually one of two things. It's ad 9:16:37 fatigue, which is pretty rare on Google 9:16:39 to be completely honest. And then it's 9:16:41 an audience mismatch. Now, why would the 9:16:43 audience be correct historically and now 9:16:45 be bad now? Like why did we used to have 9:16:47 high CTRs but now CTRs are declining? 9:16:50 Well, it is because of over here the 9:16:52 audience changes. So, you can see all of 9:16:54 these are somewhat related. Anytime you 9:16:56 troubleshoot a submetric, it is going to 9:16:58 be related to a similar process that you 9:16:59 do elsewhere. If conversion rates 9:17:01 decrease, it's likely due to due to an 9:17:02 audience change. If CTRs decrease, it's 9:17:05 likely due to an audience change because 9:17:06 you're taking the ad that used to 9:17:08 resonate with people very well on 9:17:10 specific search terms with specific 9:17:11 people and now it's not and CTR is 9:17:13 lower. And so that is probably a product 9:17:15 of the audience that you've been 9:17:17 targeting materially changing. It could 9:17:18 also be fatigue on the ad. Now, this is 9:17:20 super unlikely in something like 9:17:22 shopping ads. In search ads, it can 9:17:24 potentially be the case. Um, you could 9:17:25 just have incorrect seasonal language on 9:17:27 the ad. So, you could be talking about 9:17:28 some kind of summer sale, but it's no 9:17:30 longer a summer sale. So, you want to be 9:17:31 just looking at the ad copy. Is there 9:17:33 something here that lacks congruency to 9:17:35 this moment in time? Um, another reason 9:17:37 why CTRs will decrease quite heavily on 9:17:39 Google Shopping is due to competitor 9:17:41 pricing. And so, you want to look at 9:17:42 competitive pricing on the products. And 9:17:44 that will normally tell you one for one 9:17:46 CTRs have decreased because one of your 9:17:48 competitors has gone on to a 20% off 9:17:50 sale. And so, now anytime someone is 9:17:51 searching for Nike shoes, they're not 9:17:53 going to click on you. they're going to 9:17:54 click on the competitor cuz they sell 9:17:55 the same product for way cheaper than 9:17:56 you. In fact, we actually found this 9:17:58 with uh one of our clients that we work 9:18:00 with where they are a large wholesaler 9:18:02 of sunglasses and every time performance 9:18:06 decreases at a product level, it's a 9:18:07 product of their biggest competitor 9:18:09 taking that product to sale on Google 9:18:10 Shopping. And so we set up a real-time 9:18:12 script to be able to understand price 9:18:14 changes on the competitor website so 9:18:16 that in real time as a competitor would 9:18:18 drop prices, we would just pull the 9:18:20 product or depp prioritize the product 9:18:22 within the campaign structure. And the 9:18:23 reason being is we could continue 9:18:25 running it and continue forcing spend to 9:18:26 it, but it would never be profitable 9:18:27 because row would just drop off a cliff 9:18:29 the second the competitor goes in and 9:18:31 just wipes price by 20 to 30%. So rather 9:18:34 than playing that game and meeting them 9:18:35 on discounts, which would have eroded 9:18:37 gross margin to almost nothing, instead 9:18:39 we're just not playing the game at all 9:18:40 and we just pull the products out of the 9:18:42 auction the second the competitor starts 9:18:43 discounting to no margin. Also, as a 9:18:45 side note before we move on, you can 9:18:47 actually check this as a metric in 9:18:49 Google. It's called search impression 9:18:51 share lost bracket rank. So I recommend 9:18:53 pulling this out and adding it to your 9:18:54 columns. Now on to level eight, which is 9:18:56 the root cause. The default way to think 9:18:59 through a next step here on Google 9:19:00 specifically is that you always want to 9:19:02 avoid a rebuild or a restructure because 9:19:04 anytime you rebuild it triggers a 9:19:05 learning reset which is going to just 9:19:07 put you in an even worse position. And 9:19:09 this is where you can get yourself a 9:19:10 really risky position on Google which is 9:19:12 that performance declines. So you make a 9:19:14 big change which causes performance to 9:19:16 decline even more. So you make a big 9:19:17 change which causes performance to 9:19:19 decline even more and you get into this 9:19:20 feedback loop until you're in a terrible 9:19:22 position. And so you want to be very 9:19:23 careful and the first thing that you 9:19:24 want to do is just apply a fix. So let's 9:19:27 say we drilled all the way down to the 9:19:29 submetric and we found out that 9:19:30 conversion rate on the landing page was 9:19:33 the issue that caused all of these 9:19:35 upstream issues. And so as a product of 9:19:37 that we apply a fix to the landing page. 9:19:40 We make a material change. Then the key 9:19:43 here is that we need to be confident in 9:19:45 allowing enough time to then see if the 9:19:48 fix worked and then potentially go into 9:19:50 another fix. We also need to have the 9:19:52 understanding of whether we can group 9:19:54 multiple fixes at once and that it won't 9:19:57 confound the outcome in being able to 9:20:00 draw what actually worked. And so this 9:20:01 is where you have to be incredibly 9:20:03 strategical. And this is probably the 9:20:04 most strategical part of this entire 9:20:06 process because we might find that 9:20:08 conversion rates are down on the 9:20:09 landing. There's a million things that 9:20:10 we can do to improve conversion rates, 9:20:12 right? We can change the offer. We can 9:20:14 improve the landing page design. We can 9:20:16 put more continuity into the ad traffic 9:20:17 that's actually driving here. There's 9:20:19 all this stuff that we can do. But if we 9:20:20 do it all and it works, we don't 9:20:22 actually know what worked, what 9:20:24 materially changed if we did nine 9:20:26 things. And so there actually is value 9:20:28 in stripping back and only doing a few 9:20:29 things so that we know what actually 9:20:31 made the impact. Or else we're just 9:20:32 throwing the kitchen sink at a problem 9:20:34 and then the problem solves and we're 9:20:35 going nice. But then when the problem 9:20:37 occurs again, we don't know how to solve 9:20:39 it without throwing the kitchen sink at 9:20:41 it again. So this is where you have to 9:20:43 assess the severity of the situation and 9:20:46 whether we want a strong conclusion off 9:20:48 the back end of the fix. Sometimes we 9:20:50 might be in a situation that is so dire 9:20:51 and so bad that we do just need to throw 9:20:53 the kitchen sink and we don't care what 9:20:55 works as long as it's fixed. Sometimes 9:20:56 it's just a little bit of a performance 9:20:58 decrease and we actually want to 9:21:00 understand what the fix is so that in 9:21:01 future it doesn't happen again. And so 9:21:03 you need to think through the 9:21:04 prioritization of the fixes that you're 9:21:06 going to go and apply. Now, I can't 9:21:07 really give you the fixes in this video 9:21:09 because there is like 20 different 9:21:11 submetrics that could be impacted. And 9:21:13 off the back of those 20 different 9:21:14 submetrics, there's probably a hundred 9:21:16 different potential fixes for each 9:21:18 individual one contextual to the 9:21:20 particular business. Which is why your 9:21:22 ability to do root cause analysis and 9:21:24 then actually apply fixes um becomes 9:21:26 probably the most important skill in 9:21:28 performance marketing and media buying 9:21:29 because there is such a large decision 9:21:30 tree in all of the different things that 9:21:32 you can do that really good media buyers 9:21:34 and really good performance marketers 9:21:35 are good at being able to identify the 9:21:37 ones that will actually fix the problem. 9:21:38 The main thing that I would recommend 9:21:39 just thinking through when you're 9:21:40 thinking through the fix here to be able 9:21:42 to solve the submetric is how much of an 9:21:45 impact is this change truly going to 9:21:48 make? And you need to be unbelievably 9:21:50 honest with your answer to that 9:21:52 question. Is changing the headline on 9:21:54 the website really going to change 9:21:57 conversion rates and fix this entire 9:21:58 business? Probably not, unless they had 9:22:01 a different headline previously and 9:22:02 there was a headline tweak which 9:22:03 actually caused a decrease in 9:22:04 performance. Okay? Okay. And so we need 9:22:06 to think about the severity of the 9:22:08 expected impact of the fix that we're 9:22:10 applying. And if it's not enough to fix 9:22:12 the decline in performance, then we need 9:22:14 to continue to rethink the fix until we 9:22:16 believe it will be strong enough to get 9:22:18 us out of the problem area that we're 9:22:21 in. So if there's three things to take 9:22:22 away from this root cause analysis 9:22:25 process on Google, it's number one, 9:22:27 don't change anything before you 9:22:29 understand it. You want to check change 9:22:31 history first. You want to make sure 9:22:32 that you're checking latency in the way 9:22:34 the conversions are getting attributed 9:22:36 into the account so you're accommodating 9:22:38 for conversion time. You want to do a 9:22:40 cross channel sanity check first. So you 9:22:42 don't just want to rush in and try to 9:22:43 fix Google when it might be something 9:22:45 else. Because the reality is is that 9:22:47 most accounts that appear broken aren't 9:22:49 actually broken at all. You're probably 9:22:50 just misreading them or looking at the 9:22:52 data wrong. Number two is you want to 9:22:54 drill from the top down. You never want 9:22:56 to start at the bottom and try to work 9:22:57 your way up. And this is really the core 9:22:59 reason why most people are not good at 9:23:02 root cause analysis is people go 9:23:04 straight to the bottom and try to solve 9:23:06 hair either as a time-saving exercise. I 9:23:08 honestly don't know why people do this 9:23:09 but you need to start at the top. What 9:23:11 is materially changed in the business? 9:23:13 Because you might have a client come to 9:23:14 you and the client goes the rorowaz on 9:23:16 this campaign is down. It's like okay 9:23:18 rather than going there to start zoom 9:23:20 out go to the top. Has anything at a 9:23:22 business level changed materially at 9:23:24 all? No. Everything's actually up. Okay. 9:23:26 If everything's up, does it matter that 9:23:28 this particular campaign has a 10% 9:23:31 decline in rorowaz in the Google account 9:23:33 and it's the fifth biggest spender? 9:23:34 Probably not. It probably doesn't matter 9:23:36 at all. Now, should we still 9:23:37 troubleshoot why it was down? Sure. But 9:23:40 let's start at the top first to 9:23:41 understand the severity. Then let's go 9:23:43 to a channel level and go, well, has 9:23:44 anything materially happened on the 9:23:46 other channels that could have impacted 9:23:47 this campaign? Because this might be a 9:23:48 brand search campaign which is just 9:23:50 completely dependent on metas-pan. Then 9:23:51 let's go into the campaign type. Then 9:23:53 let's go into the specific campaign. And 9:23:54 then let's finally get to the submetric 9:23:56 that the client pointed out and see now 9:23:58 that we have all of this larger context 9:24:00 within the business whether it matters 9:24:01 at all. And we might find out that it 9:24:02 does actually matter and it does connect 9:24:03 to the top and therefore we need to go 9:24:05 into a root cause analysis and make a 9:24:07 material change. But a lot of the time a 9:24:09 submetric will get called out to you and 9:24:11 like oh why CPC's down in this 9:24:13 particular ad. It's like well hold a 9:24:16 second zoom out go back to the top. Is 9:24:18 this impacting anything at all? Does 9:24:20 this actually matter? Is this worth me 9:24:22 spending an hour of my time on? No. All 9:24:24 right. Well, then let's not spend an 9:24:25 hour of time troubleshooting this one 9:24:27 particular item. My CPCs are down when 9:24:29 it's absolutely meaningless to the 9:24:30 growth of the business. Or we go to the 9:24:32 top, we work our way through for 5 10 9:24:34 minutes and we find actually yeah, this 9:24:35 does matter. This is a great call out. 9:24:37 All right, let's try to figure out why 9:24:38 did CPCs decline on this particular ad 9:24:40 and how do we fix it? So, always go back 9:24:42 to the top. Don't let yourself get drawn 9:24:43 straight into the bottom. And then 9:24:44 number three takeaway is that once you 9:24:46 go through this whole process and you 9:24:48 get to the bottom, you want to make sure 9:24:49 that you're defaulting to a fix rather 9:24:52 than a rebuild. Rebuilding ends up 9:24:54 costing you the learning phase and will 9:24:56 likely just make things worse. Rebuild 9:24:58 is the last thing you really want to do. 9:24:59 You always want to start at what are the 9:25:01 fixes that we can apply? What are the 9:25:03 highest impact fixes? Will it make the 9:25:04 impact that's required to change us back 9:25:06 to baseline? And only if the fixes don't 9:25:08 work or if the fixes won't make a 9:25:10 material enough difference to be able to 9:25:12 bridge the gap to the historical 9:25:14 performance, only then do we go to a 9:25:17 rebuild. One of the most dangerous 9:25:18 things that performance marketers do is 9:25:20 that they tweak all the time. They see a 9:25:22 number drop and they reach for a 9:25:24 setting, but most of the time it just 9:25:26 makes the problem worse because the 9:25:28 setting was not the cause and the change 9:25:31 further moves things in the wrong 9:25:32 direction. The skill that ultimately 9:25:34 separates a seriously good performance 9:25:36 marketer from someone who's just 9:25:38 tweaking against metrics all day is the 9:25:40 order of investigation. They start 9:25:41 broad. They drill narrower. They ask at 9:25:44 every single level why. And then they 9:25:46 only act when they can actually finish 9:25:48 the sentence. This campaign is 9:25:50 underperforming because this metric drop 9:25:52 driven by this submetric which was 9:25:54 caused by this trigger which itself was 9:25:56 the root cause. If you can't put that 9:25:58 sentence together after this process, 9:26:01 you do not touch the account. You need 9:26:03 to be able to say which metric dropped, 9:26:05 what was the submetric, why did that 9:26:07 submetric change, and therefore what is 9:26:09 the root cause. If you're a performance 9:26:10 marketer and you found this video 9:26:12 helpful, please reach out to us at 9:26:13 hiring at bluedigital.com.au. 9:26:16 We are always hiring for more talented 9:26:18 performance marketers. And if you're a 9:26:19 brand doing over $5 million a year in 9:26:21 revenue, click the link in the 9:26:22 description. It will take you to a short 9:26:23 two to three minute video which runs you 9:26:25 through how our audit process works.