0:00 If you ask Claude or Chat GPT a 0:02 question, 0:03 >> [music] 0:04 >> it can be incorrect, but it makes it's 0:06 it's so confident, right? It makes you 0:08 feel just 0:09 so confident in what it's saying, you 0:11 believe it. And there are and I know 0:14 it's giving me like incorrect answers 0:16 before, and it's like still makes you 0:19 feel like you almost second-guess 0:21 yourself. Well, maybe I'm wrong, but 0:23 like I know 2 + 2 = 4, but it's telling 0:26 me five in this. So, 0:28 I think a lot of people get caught 0:29 [music] up in that. 0:30 >> What we're going for with Wicked Reports 0:31 is what what we feel like we've enabled 0:34 is decision certainty. 0:36 That's the whole point of attribution. 0:38 I'm I'm you Why am I wading through all 0:40 this data, which you don't have to do 0:41 anymore? We'll get into that, but why am 0:43 I paying for data when I already have 0:45 all this other data? Well, we're trying 0:47 to get you decision certainty. And 0:48 instead [music] of watching trainings 0:50 and spending a couple hours a a week or 0:53 whatever it's going to be or not doing 0:55 that, um we're we're getting the 0:57 decision surface done on a platter now. 0:59 So, I'm real excited to to get into that 1:01 as our conversation unfolds. 1:03 >> All right, I have Scott with me again. 1:05 We just checked it's been a full year 1:08 since our last video we did together. Um 1:11 Scott, by the way, I've got uh you know, 1:13 if I take a drink, it's from a Wicked 1:15 Reports Yeti. Okay, so I I made sure to 1:18 bust this baby out. Yeah. I I This this 1:21 is my office. I I use this quite a bit, 1:23 so I do I do appreciate that. I think 1:24 you sent that after um our video last 1:27 time. So, I'll I'll take a sip of uh 1:28 water. 1:30 >> You know, that uh 1:31 we had done a customer survey and I had 1:34 offered consulting, which normally I get 1:36 a grand an hour, 1:38 and like a couple people cared. It 1:40 seemed like not some people, but it was 1:41 a handful. And then when I did uh and I 1:44 offered that for free for some sort of 1:46 promo, but then when we did the Yeti 1:47 mug, 1:49 overrun with 1:51 >> [laughter] 1:51 >> Really? 1:52 >> consumer impulses. Everyone loves 1:54 Everyone wanted the the Yeti mug, and uh 1:56 so I'm glad uh I'm glad you got it. 1:58 >> Forget Scott, they just give me the Yeti 2:00 mug, huh? Yeah. 2:02 >> I have no standing compared to that 2:04 beautiful mug. 2:05 >> [laughter] 2:06 >> Well, it is a great it's a great mug, so 2:09 I do I appreciate that. And I figured 2:12 I'd bust it out for this video, too. 2:14 Uh 2:15 you wanted to chat, and I thought it was 2:17 good like you've got some case studies 2:19 on sort of companies who are ROAS 2:22 healthy, 2:23 but they're not seeing new customer 2:25 growth. It's a topic that I love to talk 2:27 about because there's so many businesses 2:30 that just focus on ROAS ROAS ROAS, and 2:33 it's like hold on a second. Let's you 2:34 know, let's peel what this back. What 2:36 does all of this mean? And let's take a 2:38 look at your business cuz ROAS 2:40 is a guiding metric. It should not be 2:42 your North Star. 2:44 Uh for those who don't know, Scott owns 2:47 a company called Wicked Reports. I've 2:49 talked about Wicked Reports on this 2:50 YouTube channel, in my school group. 2:53 We use it for lots of different clients. 2:56 Uh third-party attribution tool. 2:58 It is my favorite from all the ones that 3:00 we've tested. 3:02 Uh the ease of use, uh Scott's team is 3:06 really good support on like onboarding, 3:08 get getting you integrated correctly. In 3:11 fact, I had one of my big clients we had 3:13 like somehow our Google Ads integration 3:15 sort of like expired or like I don't 3:17 know what happened. I had no idea. I I 3:20 wasn't paying attention, and then you 3:21 know, like they were following up with 3:23 me uh 3:25 every day to get that set up again, and 3:27 I was like, "Oh, I had no idea." So, 3:29 incredible support, easy to use 3:31 platform, and [clears throat] 3:34 um I trust the data as well that that 3:36 I'm being presented with Wicked Reports. 3:38 Some of the other tools out there, I 3:41 I don't know. Seems seems a tad iffy or 3:44 game-ified a little bit. But, you've got 3:47 you've got some cool case studies to 3:48 share, and then also some new features 3:50 with Wicked Reports that can help 3:53 with attribution, especially now with 3:55 this AI stuff that's all over the place. 3:58 So, yeah, tell tell me I guess what you 4:00 what you have going on. 4:01 >> Yeah, well, 4:03 thanks for the nice intro. And you know, 4:05 what we're trying to do, we've always 4:06 been attribution's always been a 4:08 verified source of truth. That's been 4:09 the point of it, at least in my point of 4:11 view. 4:12 And then with AI coming out, you know, 4:14 initially was like, "Oh, is this going 4:16 to be a threat? Is it going to be 4:17 helpful?" And fortunately, architected 4:20 the right way, the attribution data can 4:23 be the the layer of trust, along with 4:26 all our expertise layered in there as 4:28 well, 4:30 that can then be the layer of trust so 4:32 that you can get 4:34 AI that's actually accurate, not just 4:38 plausible, sounds great, but is 4:40 completely made up and going to run your 4:42 account into the ground. Which we've 4:44 seen a lot of that. I mean, AI it's you 4:46 know, I always have to tell it to stop 4:48 being just my cheerleader. Every I just 4:51 saw something that you forgot, and 4:52 you're like, "Oh, great idea." When I 4:54 thought I had a concrete plan. That 4:55 happens all the time. And I love using 4:57 Claude. It's very fun. It's a lot of 4:59 fun. The data analysis in it is is 5:02 suspect. If you don't have that 5:05 foundation of accurate data, 5:07 particularly in marketing. 5:09 And what we're going for at Wicked 5:10 Reports is what we've feel like we've 5:12 enabled is decision certainty. 5:15 That's the whole point of attribution. 5:17 I'm I'm why am I waiting through all 5:19 this data, which you don't have to do 5:20 anymore, we'll get into that, but why am 5:22 I paying for data when I already have 5:24 all this other data? Well, we're trying 5:26 to get you decision certainty. And 5:28 instead of watching trainings and 5:30 spending a couple hours a week or 5:32 whatever it's going to be, or not doing 5:34 that, we're we're getting the decision 5:37 surface on a platter now. So, I'm really 5:39 excited to to get into that as our 5:40 conversation unfolds. 5:42 >> Okay, amazing. 5:44 Yeah, I mean, we were just talking about 5:46 I know I'm skipping the into a bit of 5:48 the AI stuff already, but like I was 5:50 sharing and I I did a YouTube video on 5:52 this 5:53 uh that just actually came out this 5:55 week, but you know, we did have a case 5:57 where 5:58 um a client went rogue. There There was 6:01 two business partners there. One of the 6:02 business partners went rogue. They had 6:04 two accounts we were managing. So, he 6:06 took the one account, went rogue, and 6:09 look, I don't know, you know, the the 6:11 other business partner said he's using 6:13 He wants to see Kenny run this with AI 6:15 and talking to Claude. I don't I have no 6:18 proof. I don't know if the guy I can't 6:20 see his conversations with with Claude 6:22 or ChatGPT. 6:24 Um and then there's no It's not like, 6:26 "Oh, you can see in the change history 6:27 Claude, you know, was the the made 6:29 this." No, it was just a bunch of 6:31 changes that they were doing. Also, 6:33 applying a lot of the Google 6:34 recommendations, too. So, it's like 6:36 fully leaning into 6:39 I don't need any experts for this, you 6:41 know, this should all be automated at 6:42 this point. Google's going to tell me 6:44 what to do. And then I if I have 6:45 questions, I'll just go to Claude now. I 6:48 I can do all things in the universe uh 6:51 that I have no knowledge of, right? And 6:54 they wreck the account. I mean, you 6:56 know, leads are down by half uh and 6:59 they're paying double for those leads. 7:01 And it is 7:03 because 7:04 you know, if you ask Claude or ChatGPT a 7:08 question, 7:09 it can be incorrect, but it makes it's 7:11 it's so confident, right? It makes you 7:13 feel just 7:14 so confident in what it's saying, you 7:16 believe it. And there are and I know 7:19 it's giving me like incorrect answers 7:21 before, and it's like still makes you 7:25 feel like you almost second-guess 7:26 yourself. Well, maybe I'm wrong, but 7:28 like I know 2 + 2 = 4, but it's telling 7:31 me five in this. So, 7:33 I think a lot of people get caught up in 7:35 that. I mean, and I know like there are 7:37 certain things that I I now think AI 7:40 first, right? So, there are there are 7:42 decisions that I make within the 7:43 business where it's like well, where can 7:46 I use AI to sort of help here? Think 7:49 some people have taken this a little too 7:50 far and we've seen that now with this 7:54 specific account. 7:56 But with like 7:58 some of like, you know, your your guys' 8:00 clients or case studies, have you seen 8:02 any like crazy examples of that? 8:04 >> Um 8:05 well, fortunately 8:07 our particularly some of our longest 8:09 clients like we have 8:11 a lot of clients with us that have been 8:13 with us since pre-2025. 8:16 And those that have adopted our point of 8:18 view and now are using our deterministic 8:20 AI 8:21 help, which we can get into 8:23 deterministic versus generative. 8:25 >> Yeah. 8:25 >> Um 8:26 they now rely on it or then if they're 8:28 busy agency, they'll hand off the report 8:31 that it generates to the junior media 8:32 buyers and say, "Hey, go make these 8:34 changes." Because it's a it's an 8:36 interactive system with your goals and 8:39 it gets to know your historical data and 8:43 then there's 8:44 12 years of expertise we put into it and 8:46 then 20,000 lines of code to make sure 8:48 the attribution's accurate. And then 12 8:51 years of expertise to make sure the AI 8:53 isn't doing things it thinks is smart 8:55 that isn't. And that took mon- eight 8:57 months till we were like, "Now we don't 9:00 have to like baby sit it." It was a lot 9:02 of work. 9:03 >> Right. 9:04 >> So, I've 9:05 fortunately we haven't had people go off 9:07 the rails and then just say, "You know, 9:09 I'm just going to listen to what Meta's 9:10 MCP tells me." 9:12 >> Sure. 9:12 >> Because that thing is that I mean 9:15 uh 9:16 Think of how Meta's AI works as an 9:18 algorithm. You're higher you say you 9:20 want to do sales, you've excluded your 9:22 existing customer list cuz you want to 9:24 acquire new customers ideally 9:27 and it just ignores it because its job 9:30 is 9:31 find the fastest conversions and it 9:33 learns, "Oh hey, I know you said to 9:36 exclude these people, but they convert 9:38 way faster. 9:40 And then we see time and time again 9:42 where someone's prospecting and set up 9:44 cold trafficking, but the meta algorithm 9:47 is bringing more repeat customers and 9:50 the new visits from supposed cold 9:52 traffic 9:53 is not even half of the traffic. Not 9:56 even half of it's accurate according to 9:57 how you defined. 9:58 >> Right. 9:59 >> We can pull up IP logs, you know, all 10:01 these ways we do this identification, 10:04 but it's matched against your order 10:05 data, against your CRM, and against 10:07 historical traffic. So, it's not just us 10:10 putting a thumb in the air and trying to 10:12 guess. It's very logical and 10:14 and it makes sense. 10:15 >> Right. 10:16 >> And so, the out the the the core behind 10:18 their data is not conducive to 10:23 your true North Star metric as a 10:25 business, which is I need more new 10:26 customers to grow. 10:28 It's like I and Meta's is I need to show 10:30 more sales and ROAS so you'll keep 10:32 spending. And then you're allowing Meta 10:34 to grade its own homework and then tell 10:36 you what to scale. It's just going to 10:38 scale retargeting bottom of the funnel 10:41 almost every time. There's ways to get 10:43 around it, which we can talk about, but 10:44 they're 10:45 they're technical and you got to do them 10:47 or you're going to 10:48 you're going to wonder why your ROAS 10:49 looks at five five ROAS all the time and 10:51 you're break even. 10:52 >> Yeah. Yeah, and like good luck getting a 10:55 five X on a true cold prospecting 10:58 campaign on any channel, right? 11:00 >> to do a 180-day view and have 11:03 supplement repeat. 11:05 >> Yeah, that's why it it it yeah, it makes 11:07 people feel good, but I'm you know, like 11:09 what what hold on, let's look behind 11:10 what that actually is. I know I did a a 11:13 video um which probably been a year or 11:15 two ago now, but um where I was I I 11:18 showcased Wicked Reports uh because we 11:22 had that with a Performance Max campaign 11:23 with a client where like more than half 11:27 of the true conversion value was from 11:30 repeat customers. This was a supplement 11:32 brand, so it made sense, right? And PMax 11:35 was just sucking up a lot of the these 11:37 repeat customer orders. 11:40 So what looked great, what looked like, 11:42 "Wow, we made all this money." It was 11:44 like actually 11:46 those customers were more than likely 11:48 going to buy from you again already. And 11:50 Performance Max somehow scooped that up. 11:53 And so your true new customer 11:54 acquisition cost was like totally 11:56 unprofitable. And so it's like, "Stop 11:59 leaning into that. We have to kind of 12:01 lean out We got to get out of that." And 12:03 you know, we started pushing a little 12:04 further on standard shopping. And in the 12:06 data that improved their standard 12:08 shopping 12:09 had more new customer 12:11 acquisition as well as no percent of new 12:14 visitor of traffic. So we knew we were 12:16 actually hitting 12:17 a higher percent of colder audiences who 12:20 have never heard of this brand before, 12:21 which is what advertising should be 12:23 doing, right? And Weekly Reports was 12:25 great and gave us that clarity to make 12:29 those decisions. But I like what you 12:31 were just saying, so we'll we'll go into 12:32 like these new features. Your It sounds 12:35 like 12:36 cuz I honestly, I don't think I've even 12:38 played around Well, they're not released 12:40 yet, right? They're coming out next week 12:42 or 12:42 >> So they're getting wider released each 12:45 week, but we have our first brave souls 12:48 on it as of this week. So we onboarded 12:50 people this week. We had people stealth 12:53 testing the week before. 12:55 And then it'll be wide release. I don't 12:56 know when this will come out, but it'll 12:57 be This will be wide release July 15th. 13:00 I can certainly get you on it after the 13:01 call though. I'll just send you the 13:02 directions and you can 13:04 >> Yeah. Yeah. Excellent. 13:05 >> running on your production data already. 13:07 >> Excellent. Uh well, I like what you said 13:10 because 13:12 you guys are using We'll just call it AI 13:14 technology, right? But it's built with 13:17 your 12 years of expertise. So the 13:20 guardrails, the the context that it's 13:23 learned from, your you created that. And 13:26 yes, it's still like AI technology, 13:28 right? And that's where same with like 13:31 this client who would kind of went rogue 13:33 and decided, well, I you know, Claude is 13:35 God, so I'll just talk to Claude and get 13:38 all the answers of the universe. And it 13:41 feels that simple at times, but it's 13:43 not. And it's because Claude or ChatGPT, 13:47 any of these LLMs, 13:48 you're going to get I it's very good at 13:51 a quick search. How what's the 13:52 ingredients of an apple pie? It's it's 13:54 going to give you a, you know, the 13:55 ingredients to go 13:57 buy so you can go make an apple pie very 13:59 quickly, right? But 14:02 I would not use it for something that 14:05 you would need true expertise in. And 14:07 what you could use though, I mean, I 14:09 think what what's Is it Harvey that's 14:11 like this massive AI company for the 14:14 law industry? 14:16 >> Mhm. 14:16 >> Um you're not asking you may ask Claude 14:19 some generic and I actually have some 14:22 generic law advice. 14:25 >> Yeah, business and see how I'm getting 14:27 screwed. 14:27 >> But I'm still Yes, I am still paying a 14:31 lawyer at the end of the day cuz I I'm 14:34 not going to go to court with what 14:36 ChatGPT just, you know, gave me or a 14:38 Claude. And I that's how I 14:41 when it comes to like Google Ads or Meta 14:43 Ads, doesn't matter, any sort of 14:45 expertise, 14:46 my you know, we've trained our our own 14:49 brain, so the Grow My Ads brain, if you 14:51 will, our own internal IP is trained on 14:55 5,000 14:56 hours of Loom videos. And thankfully, 15:00 we've been recording Loom videos for 15:03 >> Yeah. 15:03 >> the last five or six years. Obviously, 15:05 all my YouTube content, I do Q&As about 15:09 once a week. And so all of this 15:13 information of our expertise is being 15:16 fed into our brain. And so when I ask my 15:19 Claude a Google Ads question, if I'm 15:22 like, you know, trying to decipher some 15:24 data with a client's account, it I'm 15:28 going to get a 100x better answer than 15:31 someone who just logs into their account 15:34 that they're paying $20 a month for and 15:35 they generically ask, right? So, it's 15:37 about the inputs you give it. And so, 15:41 what it you know, what it sounds like is 15:43 like, yes, you guys have this AI these 15:45 AI features, but it's it's technically 15:48 this is your, you know, internal IP that 15:51 you guys have built. It's just also 15:55 essentially AI technology now. 15:58 >> Yeah, AI 15:59 the AI models are becoming a commodity. 16:02 And they'll keep releasing them cuz they 16:04 want us to spend more in tokens and oh, 16:06 it's 8% better and oh, the government 16:08 doesn't want you to get it cuz it's 16:10 going to crack everyone's They're going 16:11 to keep hyping it up cuz they're going 16:12 to IPO. 16:14 But, at the end of the day, with us, we 16:17 have two modes. One's the data that's 16:19 really hard to collect and track 16:20 correctly. 16:22 And then the expertise combined with 16:24 that data, like we have a decision agent 16:26 which is based like what's the decision 16:29 here, which is based on a lot of rules. 16:30 Two two simple ones which aren't simple 16:34 to AI, but is like one is don't scale 16:37 certain branded search at first cuz 16:40 they'll always say, oh my god, it's 12x 16:42 or whatever. 16:43 >> Yeah. 16:43 >> And I was like, well, you need to go 16:44 spend more at the top of the funnel 16:46 that's feeding that. 16:47 >> Right. 16:47 >> And then or if you just make a change, 16:50 there's these, you know, you pay for 16:52 clicks and there's a lag when they 16:53 convert. I mean, hopefully they all 16:54 convert that day, but particularly the 16:57 higher AOV you have, the longer the lag 16:59 and the more you need top of funnel, 17:00 middle and bottom. Well, let's just say 17:03 you're spending at top of funnel and you 17:05 spend for a week and then you cut the 17:07 spend you the the 17:08 the client says, ah, it's not working, 17:11 cut it. And you cut it like 90% from 10 17:13 grand to 1 grand. Well, then over the 17:15 next couple days, some of those delayed 17:16 clicks start converting, and then the 17:19 return on ad spend looks great because 17:20 you have more sales and less spend. 17:22 Well, the AI will then say, "Raise the 17:24 spend." And you'll be like, "I just cut 17:26 the spend." 17:27 And it'll yo-yo you cuz it doesn't 17:28 understand observation windows. 17:30 >> Right. 17:31 >> Attribution models, there's a lot of 17:33 different ways to attribute. What what 17:34 should you pick? Depends on your goal, 17:36 depends on how long it generally takes 17:38 your lag clicks to convert. That's a 17:41 complex scientific thing. AI has no 17:44 concept of that cuz it's been trained on 17:46 a million hours of YouTube of people 17:48 that are in there blabbing away about 17:50 trying to hype up themselves on their 17:53 meta ROAS on retargeting. So, it learns 17:55 from bad info rather than expert 17:59 scientific people that aren't probably 18:01 sharing all that knowledge, or it isn't 18:03 a dominant input to the overall model. 18:07 So, and that's happening everywhere in 18:08 science, right? You chat in, it's 18:10 getting its science info, which this is 18:12 a marketing science measurement. 18:15 >> Yeah. 18:15 >> It's getting it from somewhere. You 18:16 don't know where, you don't know how it 18:17 knew which ones to keep or not, or why. 18:20 And was that person's viewpoint may have 18:22 been accurate 3 years ago, but is it 18:24 still accurate today? 18:25 >> Right. It may have been brilliant 3 18:26 years ago, and now that 18:28 everything's so different now. 18:30 >> So, all that factors in, and people 18:32 aren't aware cuz they get this b- 18:34 I'll, you know, uh em- not really 18:37 embarrassing, but like I did a whole 18:39 clawed marketing plan. I went back and 18:40 forth with it for 2 days. This was 3 18:42 months ago. I was like, "Oh my god, this 18:44 is amazing." 18:45 It kind of sucked. In reality, the 18:47 report the results were very so-so. At 18:51 first, it was a little spike, and then 18:52 it went downhill. And it looked 18:54 beautiful on paper. I was like, "Oh my 18:56 god, this is going to be great." And in 18:58 hindsight, I went and did a postmortem. 19:00 I got this postmortem skill. 19:03 And now I run pre-mortems, 19:04 pre-postmortems, I guess. I got it off a 19:06 guy on LinkedIn. 19:07 >> That's right. 19:08 >> Well, let's suppose this didn't happen. 19:10 What could it be? And it'll rip its own 19:11 plan to shreds, and I'll be like, 19:13 "Jesus." 19:13 >> Yeah, yeah. And it's like, 19:15 you wrote the plan. Yeah. 19:17 >> Yeah, so I mean, I I fell for it. This 19:19 was in uh like February. I was like, "Oh 19:21 my god, this is amazing." Chugged out 19:23 all this content. Then I was like, gave 19:25 it to someone, "Just go do all this. Oh 19:26 my god, this is amazing." And it was it 19:29 kept me busy, kept her busy, kept me 19:31 feeling good, didn't get us much 19:33 results. Uh so, that I had a real-world 19:36 experience doing that. 19:37 >> Oh, yeah, yeah. I'm Look, I like using 19:39 it. I like brain- I like brainstorming 19:41 with it 19:42 um just to kind of you know, go back and 19:45 forth. It doesn't mean I'm I'm buying 19:47 into everything, but sometimes it can 19:50 open some ideas up. Especially, I've 19:53 trained mine um on my my cloud code 19:56 references like a lot of experts that 19:59 I've just have like their material from. 20:02 And so, 20:03 um you know, 20:04 >> That's a good idea. 20:05 >> Yeah, it could be so like, "Hey, what 20:06 would so-and-so think about this?" 20:09 Because I've got you know, just an 20:10 insane folder of content from this 20:13 specific expert on this specific topic. 20:16 So, it's like 20:16 >> doing that more. I'll upload like I'll 20:19 cut and paste the blog post or link to 20:21 it saying 20:21 >> Yeah. 20:22 >> April Dunford for positioning. I don't 20:23 know many of your audience does 20:25 positioning, they might, but she's like, 20:26 in my opinion, the top. 20:29 I've read her read her books. I've 20:30 interacted with her, and I was like, 20:32 "Before I change positioning and with 20:34 this new approach we have, what does 20:35 April Dunford think?" 20:37 >> Yeah. 20:38 >> I don't know why I assume you don't know 20:39 her. 20:40 >> I'll look her up after this, but um 20:42 that's exactly So, mine will auto-fetch. 20:45 It's part of this like daily thing I 20:46 have it run. It will auto-fetch certain 20:49 experts' either email newsletters I'm 20:51 on, their YouTube channels if they drop 20:53 new videos, I'll just transcribe it and 20:55 save the the transcript. So, it's cool. 20:58 I'm able to then say like, "Hey, 21:00 so-and-so, you know, they they might be 21:01 an expert on 21:03 on, you know, Bitcoin or the stocks if 21:05 I'm like messing around with, you know, 21:06 finance." And it's just like what what 21:08 what what would they think right now? 21:10 Doesn't mean it's right. It could be a 21:11 made-up answer still, but it's like a 21:13 good brainstorm exercise there. 21:14 Obviously, it's look, it's very good 21:16 with like with pattern recognition and 21:19 and then like 21:20 workflows that we do that uh really are 21:23 just like very logical. 21:25 Um you know, like just yesterday I I 21:28 I've got this like really beautiful 21:29 report 21:31 that blends in data from various sources 21:35 and 21:36 Claude pretty much nails it. Like I just 21:38 give it the the sources. Um it's a 21:41 monthly thing that gets done. 21:43 And it like builds the sheet for me. 21:45 It's like, "Okay, cool. I didn't need to 21:47 go do that." Like yeah, it's awesome, 21:50 but it's it's doing it's just doing the 21:53 work for me. Um 21:55 it's not like literally telling me how 21:57 to run and operate things. Uh and so I I 22:01 I I do it for the brainstorm session 22:03 there, but it's not telling me the exact 22:05 steps on how to go achieve certain 22:07 results. 22:08 >> Because when we we have that it will 22:10 achieve certain results, and it took 22:12 eight months of, you know, this AI guru, 22:16 me with my brain on it, my product 22:18 manager, our database team. It took us 22:20 forever to get Eight months is a long 22:23 time 22:24 >> Yeah. 22:24 >> iterating on something until we're like, 22:26 "Okay, we don't need to sleep with one 22:28 eye open on this." And then that was on 22:30 like a 3.6 model. Then you're terrified 22:32 to go up because the AI will just 22:34 totally change its tune. So you got to 22:35 keep the model with your 22:38 uh let's one jargon thing I want to get 22:40 clear with the audience. So generative 22:42 AI is you open Claude, you start 22:44 blabbing away, it starts like, you know, 22:46 vomiting out stuff, often very helpful 22:48 stuff, sometimes not. 22:50 But it's probabilistic. 22:52 And then deterministic AI is you have 22:55 either agents or a very extensive prompt 22:58 or other guardrails 23:00 or guardrails on the source data, as 23:02 well, so that it's 23:05 it's a controlled output that you then 23:07 can be generative with because you know 23:09 you trust the quality of the initial 23:11 insight or analysis. That's what I've 23:13 come to learn That's what we ended up 23:14 building without and then we thought, 23:16 "Oh, it doesn't seem like it's AI 23:18 enough." And we were like, "Oh, crap." 23:20 Turns out it was the right thing to 23:22 build. 23:22 >> Yeah. 23:23 >> Cuz then you can still chat away and do 23:24 other AI cool things and AI can wade 23:27 between it to pluck out from a question 23:30 which deterministic pipe we built. Uh so 23:34 it ended up being the right move. It 23:36 wasn't strategically that way. It was 23:37 just like more we were so terrified at 23:38 what the results were when we didn't do 23:40 that. 23:41 >> Yeah. 23:41 >> That we had to do it. 23:42 >> Um well, so I guess then are you able to 23:46 share or 23:48 let's 23:49 let's let's get into it. I want to see 23:51 then, you know, 23:52 what exactly you guys have built and and 23:55 and how it can help and how it is 23:57 superior to someone who thinks they can 23:59 just do attribution themselves through 24:00 data dumping into a clawed chat. 24:04 >> Sure. And first of all, you had 24:05 mentioned some case studies. So we got 24:07 them on our site here. These are all 24:09 recent ones as well. 24:11 This one was probably the biggest 24:12 adopter. Now, 24:14 probably the most successful that 24:16 happened. So I can't say this is going 24:17 to this is abnormally amazing to drop 24:20 your end CAC 50 bucks was like 24:22 mind-blowing. 24:23 >> Wow. 24:23 >> Now, they also 24:25 and it's not just better data and 24:28 magically, you still got to be good at 24:29 marketing. Like I can't say, "Oh, I'm 24:32 going to fix your bad marketing." These 24:33 guys are good at marketing and they 24:35 really bought into this creative 24:37 protocol. Well, you know, they got the 24:38 all the different types of creative. 24:40 >> Sure. 24:41 >> But then the two other pieces was the 24:43 advanced signal, which is we've sent in 24:46 the signal of new customer purchases 24:49 through the meta CAPI custom event. 24:52 >> Mhm. 24:52 >> And then the ad set chooses that to 24:55 optimize on. 24:56 >> Yeah. 24:57 >> So this gives Meta blinders on repeat 24:59 sales so it'll learn. Like it'll be 25:01 like, "Great, I just converted another 25:03 repeat customer." Like it usually does 25:05 and then you'll be like, "It This makes 25:07 it not see that conversion for that ad 25:10 set. 25:10 >> So yeah, to 25:12 to simplify that for some people, 25:14 they're they they're not using just a 25:16 generic Facebook ad purchase pixel right 25:19 now. They're they're just importing in 25:22 new customer purchase data only and 25:24 that's what it feeds into their Meta ad 25:26 campaigns. 25:28 >> Correct. So they have their regular 25:29 signal copy, but then they we turn on 25:31 this one where we auto segment when the 25:33 order comes in, we just say was it new 25:36 or repeat and then we just send the new 25:38 signal up. But then that new signal you 25:40 have to tell the ad set, "I want you to 25:42 use this to optimize on." And then this 25:45 occurs where the new customers go up and 25:47 the end cap goes down. This is like the 25:50 dream scenario. 25:52 >> It's a dream scenario. I've had I I not 25:55 to speak like I think it's a smart 25:57 strategy. 25:58 >> Yeah. 25:59 >> But man, it's got to be done so 26:01 carefully because I've seen people just 26:03 blow. In fact, I have team my own team 26:06 has done this. Like they with Google Ads 26:08 accounts where they only work off of 26:12 and this wasn't this wasn't like 26:14 Wicked's 26:15 Wicked Reports 26:17 they this was used I think it was 26:19 Audience OS and then Blot out. Um 26:23 but they you know, switch to the first 26:27 or new customer only 26:30 conversion tag import. 26:33 And some accounts could not bounce back. 26:37 Like even after a like some of them did 26:39 a one-month transition, they let that 26:41 that tag live, collect data and then 26:43 they started this transition period 26:45 over. 26:46 And so I've seen it. I have seen it 26:48 backfire on a few and it's very odd. 26:50 It's like, "Why 26:51 Why can't you scale from that? 26:53 But it's when it works it's wonderful. 26:56 So it's amazing that like in that case 26:57 >> I said not a silver bullet. We because 27:00 otherwise I'd say let's just sell this 27:01 product. Who cares about anything else? 27:03 >> Yeah. 27:04 >> Because this is going to magically drop 27:05 in CAC every time just do that. But 27:07 that's not always the case. They had 27:09 good creative. They're in skin care. 27:11 Their their their in CAC wasn't horrific 27:13 at the beginning. And you can only scale 27:15 it so low. There's a you know a vent I 27:17 mean this is crazy. I mean you can't get 27:18 a lot lower than that um when you're 27:20 selling you know 27:21 >> And they were able so they were able to 27:22 do that while maintaining 27:25 growth which is incredible. 27:28 >> Yes. So yeah, sometimes it'll be like 27:31 and then it won't you know all of a 27:32 sudden it turns out maybe your ad set 27:34 can't find or your creative isn't 27:37 working on new customers. It worked 27:39 better on repeat. 27:41 So we have our or or the products you're 27:43 offering work better as a repeat. You 27:46 might say oh I got 5,000 sales a month 27:48 from this lips I mean I'm not I don't 27:50 know makeup right but 27:52 blush. 27:53 >> Yeah. 27:54 >> Maybe like lipstick is what people like 27:56 to try out a brand for. I don't know. So 27:58 you could be offering the wrong product. 28:00 But what I don't know the data will 28:01 know. You got to go back and look at 28:03 your data. We have a report that does 28:04 this. What is the most common product 28:07 that's bought by first time customers 28:10 that have the highest eventual LTV? 28:12 >> Yeah. 28:13 >> the product you should use. Then you got 28:15 to use all your new you know all your 28:17 creativity for the right hooks for that 28:19 product for people that haven't tried 28:21 the brand which is a different strategy 28:22 than hey you have success with my 28:25 eyeliner and now try this lipstick. You 28:27 know it's totally different. So that all 28:30 has to happen. So you still need good 28:32 marketing. Claude isn't going to sort 28:33 that out for you. You still got to have 28:34 that well maybe it will someday. 28:35 >> Abs well man I don't know. 28:39 >> So then the other piece I'll get to 28:40 these new screens later. Here's a it's a 28:43 different one. This is our this is the 28:45 five forces so this is our deterministic 28:47 AI report. Now, um this is a This is the 28:51 If you want to see all the data, but at 28:53 its heart, this advertiser isn't doing 28:55 much, but I know I can share his data 28:57 right now. 28:58 Um we we bucket the spend based on 29:00 scale, chill, killer, observing, and 29:02 then we we give you context as to why. 29:05 So, all this was hard-fought to get this 29:09 to be something that I'm not going to be 29:10 terrified to show. 29:12 >> [laughter] 29:12 >> Because it understands how long to look 29:15 at something, what date range to compare 29:17 it to. It looks at your 29:20 accounts benchmarks 29:22 to see whether something is exciting or 29:24 not. So, all this stuff in here, now we 29:27 baked into our agentic stuff, which I'll 29:29 show you. But, this is an agent running 29:32 that spits you out a report every day by 29:34 campaign. 29:35 So, you don't need to look at this every 29:37 day, but this particular guy, they have 29:39 an agency if they were busy or whatever, 29:41 they just kick this report to the junior 29:44 media buyer and say, "Hey, go do 29:45 whatever" whoop, wrong screen, "whatever 29:47 this report's telling you." 29:49 >> So, inside of Wicked Reports, you're 29:51 telling like a a specific campaign, 29:53 you're saying, "This is my goal for this 29:55 campaign." And then it knows kind of 29:57 what you're trying to achieve, and then 30:00 uh it reviews Yeah, there we go. Okay. 30:03 >> So, you pick the goal. So, this is the 30:05 feedback loop because yeah, otherwise 30:07 AI, you might chat back and forth, "I 30:08 want to acquire new customers." Okay, 30:10 well, your ROAS is this. Well, no, I 30:11 want to do NCAC. And then then your AI 30:13 If you were chatting with Meta AI, it 30:16 doesn't have new verse repeat in there. 30:18 So, it wouldn't know what to do. 30:20 Or if it was 30:21 Or in it could have been new, but it was 30:23 like some people with pure e-com 30:26 just having pure new and then going NCAC 30:28 totally can work. But, if you're a 30:30 bigger AOV or you're a multi-step funnel 30:34 you need to know your top of the funnel 30:36 uh campaign NCACs, and that's different, 30:38 different measurement model, all kinds 30:39 of different stuff there. 30:40 >> Yeah. 30:41 >> How this works is we have you pick the 30:43 intention and tell us the chill zone, 30:46 which is I will continue spending the 30:49 same amount if you can hit this range 30:52 rather than a target. So, if you have a 30:54 target, you're either above it or you're 30:55 below it, and there's these couple day 30:57 fluctuations that could cause you 30:59 unknown anxiety. You know, you're 31:01 chilling out, and then Sunday is low 31:03 traffic day, your in-cat spikes, 31:05 client's like, "What the hell?" 31:06 >> Yeah. 31:07 >> We do a range over a time period. So, 31:09 this comes from my I used to design 31:11 NASDAQ trading systems. 31:13 You need ranges, and you need time 31:15 frames so that you're not constantly 31:16 overreacting to short-term tiny 31:19 volatility. 31:20 >> Sure. Sure. 31:21 >> That's something that I, you know, an 31:22 average brand owner who's successful and 31:24 smart person, they may not know that 31:26 concept. So, you got to like train them 31:27 up on or just say, "Hey, you know, 31:28 steady the account. Just let me do it." 31:30 >> Yeah, drives me insane cuz it's like, I 31:32 mean, you know, you could look at Let's 31:34 say you can go if Why not day? Well, 31:35 let's go to day Let's go to hours. How 31:37 about that? Let's Let's go scrutinize 31:39 hours. Actually, let's scrutinize 31:40 seconds of hours, and then let's see 31:43 what the fluctuation of in-cat is. 31:45 >> I hate that people want to do that 31:47 because the data is is in the terms, 31:51 particularly with Google, they'll update 31:53 conversion data up to 72 hours to make 31:55 it accurate. 31:56 >> Sure. 31:57 >> So, you're acting on inaccurate data 31:59 because the conversion tracking, 32:01 lo and behold, you send a good email or 32:03 SMS push, and it seems to pop a quote in 32:06 the Klaviyo or whatever you're using, 32:08 Attentive, whatever. Oh, wow, I just 32:10 spiked that All of a sudden, my Google 32:13 and Meta conversion spiked. Well, that's 32:15 because it's picking up on those 32:16 conversions. 32:17 So, the whole premise is just not not 32:20 intelligent, in my opinion. 32:21 >> No, it's not, but 32:23 >> I I have I have people that still do it, 32:24 and they're smart and otherwise, "Oh, 32:26 it's how we always done it." I'm like, 32:27 "Well, 32:28 you're lucky." 32:29 >> [laughter] 32:29 >> Yeah. Yeah. 32:30 >> There's no way it's because of that. 32:32 There's no way. 32:33 >> You've got to educate people on it. I I 32:35 it also it might go back to the days of 32:37 like where you had full manual control 32:40 over advertising campaigns where you 32:42 could like go bid adjust hours inside 32:45 Google Ads and it actually made a 32:46 difference back then, right? And so I 32:49 don't know if that's just still 32:50 ingrained in people, but 32:52 I don't know. To operate a company, 32:54 look, I spend, you know, money myself 32:56 for for Grow My Ads for our own 32:58 marketing efforts and uh 33:00 yeah, like I look I I I know, you know, 33:03 you can have all the tools, right? You 33:05 could you you have in-app setup, you 33:07 could have you could be using Wicked 33:08 Reports, you could be doing 33:10 um 33:11 uh post-purchase surveys or form fill 33:13 surveys of source where they tell you 33:15 where they found you even though they 33:17 could even be wrong there and you could 33:19 blend all of that together and 33:21 ultimately you still aren't going to 33:22 have 100% truth. It's just not simple. 33:25 People don't like I I'm not wearing it 33:28 today, but you usually have a black 33:30 T-shirt on and I've got a lot of the 33:31 black T-shirts from one particular 33:32 company and I saw an Instagram ad once 33:36 and then 6 months later bought from 33:37 them. There there's no attribution there 33:40 of me ordering. I just didn't need a 33:41 black T-shirt at that time, but I 33:43 remembered them cuz I was like that that 33:45 looks like a 33:46 minimalistic athleisure sort of black 33:48 shirt I would like to try sometime next 33:50 time I need a re-up my or refresh my my 33:53 black tea collection. And so I went 33:56 direct to them. In their attribution it 33:58 probably looked like uh brand. I 34:01 probably I doubt I I clicked one of 34:03 their brand ads, but it's a it was a 34:05 direct search of their brand name, 34:08 directly went to them. That's probably 34:11 in any of their attribution what was 34:13 picked up unless maybe they were using 34:15 Wicked Reports, but 34:17 it's just not it's not simple. So yeah, 34:19 to try to scrutinize like days and and 34:21 and even weeks sometimes it's just like 34:23 you can't you can't. 34:25 >> That's why we have this sales length 34:26 here like you're allowed to edit it, but 34:28 we data set it from your data. This is 34:31 important because this allows us to know 34:34 observation window for top of the funnel 34:36 >> Mhm. 34:37 >> and what trend to use because sometimes 34:39 your stats aren't good, but you've made 34:41 some adjustments and the trend's good. 34:44 Well, knowing this sales cycle length 34:46 plus then to look at two week trends, 34:49 you can compare the trend, so it's not 34:51 just always black and white here. Like 34:52 cold traffic, we use first click 34:54 attributed. Like if you're if you're 34:56 actually a 1.7, 34:58 it would be like, oh, AI would say, oh, 35:00 you got to kill it. You didn't hit your 35:02 two. But if you were at one and now 35:04 you're up to 1.7 and we know it's a 35:05 14-day cycle. We're like, well, you've 35:07 almost 100% improved your first click 35:10 ROAS. We don't want you to kill it even 35:12 though it's technically what you said to 35:13 kill it because it's improved. 35:15 >> Right. 35:15 >> Those nuances are in there which, you 35:17 know, 35:19 I always knew I didn't realize how 35:20 complicated measurement was until I had 35:23 to actually 35:24 try to allow this to work without me. I 35:27 was like, oh, man, I have like all this. 35:30 Fortunately, I have all this expertise. 35:32 I mean, I've been doing it for a decade. 35:33 I mean, I'd be I'd be dumb if I didn't 35:35 have some expertise by now. 35:37 >> You have the expertise. Yes. 35:39 >> Putting that all in there was like, 35:40 geez, no wonder we have people that 35:42 sometimes churn with great data. Like 35:44 they don't have time to think of all 35:45 these things. 35:46 >> Right. 35:46 >> So, this is a whole new thing. I was 35:48 like, with AI, I was like, okay, once we 35:50 got it tight here, you still got to go 35:52 read. Then I was like, how do we make it 35:53 even easier and get decision certainty 35:56 that we can actually guarantee So, now 35:58 we can guarantee three times what you 36:00 pay us. We're you're going to make 36:02 decisions three times what you paid us 36:05 or you don't pay. 36:06 Uh which generally it's it's virtually 36:09 there's there's no way you're not going 36:11 to get that unless you maybe you're 36:12 spending 100 bucks a month, but then you 36:14 wouldn't be spending much with us, 36:15 either. 36:16 >> Right. 36:16 >> We wouldn't We should not 36:18 you if you're not spending at least 10 36:20 grand. We shouldn't sell you. 36:21 >> Yeah, yeah, yeah. 36:22 >> sneak in and buy, fair enough, but we we 36:24 don't want you to. 36:25 >> Right, right. 36:26 >> That's Yeah, this is kind of an idea. 36:28 I'll get into more of the like sexier 36:29 screens. This is like a lot to take in, 36:31 but this is the this is the minimized 36:33 version, which we then chop up into like 36:36 a diagnosis agent. So, like if if you're 36:39 again, if you're like you were trying to 36:41 cold traffic prospect, which I hope 36:42 everyone's doing. And it says 0.8. We 36:46 don't want to just say kill it and go 36:47 from scratch. We mine all the data 36:49 points based on this goal. Cuz we have 36:52 like, I don't know, 80 to 100 data 36:54 points and then all the different models 36:55 and all the trends and then the pacing. 36:58 So, and the forecast. So, it's a couple 37:01 hundred data points and we're like, 37:03 okay, which four matter to help you 37:05 improve this? And we 37:07 narrow down so your brain isn't like 37:10 tweaking on 85 elements. 37:12 >> Yeah. 37:13 >> We get it down. 37:15 Um so, here was initial Wicked Reports. 37:18 Uh you know, our UX, you know, we get 37:19 it's okay, but it's a little antiquated 37:21 and we like we got to make it a little 37:23 more sexy, but really what matters is 37:25 more that it's more easy to use and gets 37:28 decisions on a platter. 37:30 >> I think it's easy to use. The ones that 37:32 look, you know, 37:34 sexy, I suppose, it it's almost in two, 37:38 you're like, wait a second. What am I 37:40 like, give me just like, what am I 37:41 looking at here? And I I think most 37:44 users who would be using your tool 37:46 probably do appreciate the simplicity of 37:49 some of it. 37:50 >> So, we kept it simple. Here's the new 37:52 one. It's live in accounts now for 37:54 people that are early adopters and then 37:56 we, you know, about to be on for 37:57 everyone here and I don't know, but I 38:00 don't think we want to release anything 38:01 over July 4th week cuz that's when it 38:03 will be like 38:04 nothing left to do but just make it live 38:06 live for everyone. So, July like 7th, 38:09 this will be the new brand new homepage 38:11 for everyone. So, it's got some cool 38:13 stuff where uh some new functionality. 38:15 The biggest one that I kind of love is 38:18 that it's doing pacing. 38:20 Um so, we'll show stats for you 38:24 and your marketing performance stats, 38:25 but then we'll show pacing. Pacing is, 38:28 well, in the last 30 days, I've spent 38:31 this and I've gained this s- amount of 38:33 sales, but based on your history 38:37 and based on everything we know in 38:38 attribution, we are forecasting that 38:42 this ad spend is actually going to get 38:44 you in this case th- this is live data. 38:45 This isn't fake data for this customer. 38:47 They're pacing to get an extra 300 sales 38:50 from the spent. 38:52 So even though like they're already 38:53 doing pretty well, they're actually 38:55 pacing to do even better and we can pop 38:57 it up and see what they're pacing to. So 38:59 it's like as adding a forecast based on 39:02 how long their conversions take to bu- 39:05 people take to buy, 39:06 >> Mhm. 39:07 >> historically when they've stopped when 39:09 the when when we've tracked clicks after 39:11 this how long they take to convert, the 39:13 same thing. Conversion rate, 39:16 future LTV, we bake all that in and pace 39:19 out 39:20 where they're going, which really helps 39:21 cuz like you'd be like, "Oh, I got a 100 39:23 in CAC." Yeah, but it's pacing to 83 39:25 bucks. So if you're like, "Oh god, I I 39:28 needed to d- be below 90." Well, we 39:30 actually think what you the the clicks 39:32 you've bought, you're going to be safely 39:34 below 90. 39:35 >> Right, like if you just stopped 39:36 everything and then waited an additional 39:39 whatever the sales cycle period is, 39:41 you're going to populate 39:43 a projection of of that of an $80 in 39:47 CAC. 39:48 >> Yeah, and this would normally mean I'd 39:50 train you, I'd be like, "Go to the 39:52 benchmarks report. Look at how fast 39:54 they're converting. Go to your LTV and 39:56 see how many non-day zero what the 39:58 accrued value is." Okay, now do trend 40:01 versus trend, then snapshot it and then 40:03 come back and look and confirm. All that 40:05 stuff's done, we just calculate and give 40:07 you the number and more importantly our 40:09 AI is trained up on it. So if it knows 40:12 that your in CAC's 90 is your goal, you 40:14 know, 90 to 80 or whatever, it'll be 40:16 like, "Hey, you can keep spending here. 40:18 You're good You're good." Rather than 40:20 all that work and hoping you ask the 40:22 right questions in the right order, look 40:24 at the right data, have time to think 40:26 about it. It's already trained up on all 40:28 this stuff. 40:28 >> And by the way, to be able So, to be 40:30 able to view data like this 40:33 to make decisions, which it sounds like 40:35 now, you know, you guys inside Wicker 40:38 Reports help with even decision-making 40:40 in regards to the reporting. 40:42 >> Yeah. 40:42 >> This is why, you know, when I talk to 40:44 some companies 40:45 and they just are generically following 40:48 a ROAS metric, and that's it. That's 40:51 like the end-all, be-all for them is 40:52 just what their in-app ROAS is. 40:55 And 40:56 you know, it's like well, no wonder 40:58 you're being dominated by your 40:59 competition. 41:01 And they can't figure it out They're not 41:03 It's like Google 41:55 Ads is a matured advertising platform. 41:57 So, smart companies with good data that 42:00 they understand in their business, 42:01 especially those that understand in CAC 42:03 to LTV, and they know payback cash 42:06 periods and everything else, they got it 42:07 all figured out. They're They're able to 42:09 beat you because of all of that, and 42:11 that's why they're able to be super 42:13 aggressive in those auctions where you 42:14 are not. And that's how you're getting 42:16 beat. It's not like some superior thing 42:18 they hacked in Google Ads in most cases. 42:21 It's literally just smart business. And 42:24 they follow the data, they trust the 42:26 data, they work the data, and that's how 42:29 they're able to bring more new customers 42:30 into their business than yours. And so, 42:33 having the ability with a tool which is 42:35 very affordable for most businesses, 42:37 like if you're spending 10k a month on 42:38 ads, you go at like this is a no-brainer 42:41 to have. To be able to make then 42:43 decisions on this, this is how you beat 42:44 your competition in auctions. 42:47 >> 100% cuz you can spend more 42:50 >> [snorts] 42:50 >> intelligently in the right spots cuz you 42:53 know you're getting the right type of 42:55 traffic that converts at a spread over 42:58 time that you can afford. 43:00 And a lot of people never get there. 43:02 Because look at the And like if you're 43:03 doing the day So, look at the day-to-day 43:05 fluctuation here. Like his new 43:07 customers, it's all over the map. That's 43:08 against the day-to-day. 43:11 And then you're always getting this lag 43:12 of new ones coming in. 43:14 Um 43:15 so, exactly. Like the Like his revenue 43:19 and meta, you'd say, "Uh-oh, I better 43:21 kill it." But he knows, "Well, I'm 43:22 pacing to 70 I'm looking underwater. 43:25 Yeah, but I'm pacing to 1.9." 43:27 So, I actually can spend if that's fair 43:29 to him. Because this return on ad spend 43:32 is distributing credit equally among 43:34 touch points with no bias to one 43:38 platform. So, it's grow as you could use 43:41 or as CAC 311, he'd be horrified. Pacing 43:43 to 148, that's probably still too high 43:45 for him. But in Google 43:47 >> Also, 43:48 your Wicked Reports, and I know this 43:51 this is done now 43:53 in the reporting and and 43:56 so, you don't need to go do this. But I 43:58 like going in, and I've done this 44:01 multiple times, and I will just go for 44:03 the last like 30 days, and I'll go click 44:06 all the orders 44:08 and I'll see the journey. 44:11 And of cuz you have the the the customer 44:14 click journey, right? From Oh, they 44:16 started here, they ended up here. I do 44:18 that to go get a feel myself for some of 44:21 the actual data. And it's very 44:24 interesting to see what campaigns on 44:26 Meta or Google they're interacting with. 44:29 And I I actually then it's it's not 44:31 something most people would do. Uh I 44:33 might have just too much time on my 44:35 hands, I guess. But it is like for me to 44:38 recognize the patterns of the customer 44:40 journey. Um but I love being able to 44:43 look at that data because you guys are 44:45 able to capture that that that first 44:47 click to the last click of when they 44:48 make a purchase. 44:50 >> Yes, so you can go in here. I don't 44:52 know, I'll just click on some random 44:53 person here. 44:54 It'll go in and um show you the people, 44:57 new or repeat, then you can click on 44:59 them 45:00 and it'll show the whole um 45:02 history 45:04 of that person. So, this person started 45:06 we tracked for this 45:08 uh for this client an 8-year journey to 45:11 first purchase. Which I wanted him to 45:13 like, "Hey, can you 45:15 can you email that person and we buy 45:16 them lobsters so we can like show who it 45:20 was and find out, how did this happen?" 45:23 >> Yeah. 45:23 >> at this person. Here's like Here's a 45:25 common thing though. They bought 45:26 something in July and then now he's just 45:28 reactivated them nicely. So, you'd be 45:30 like, "Hey, yeah, to your point." 45:31 Because you can go find like a juicy new 45:33 customer that maybe was high value and 45:35 be like, "What was their path? Did 45:37 something new trigger what did this? And 45:41 are there other people like them?" Cuz 45:42 here was like, "Okay." 45:45 They had some welcome emails and then 45:47 they bought a year ago because of some 45:49 affiliate Digidip, whatever the heck 45:51 that is. And then 45:54 time goes by and then, lo and behold, 45:57 they struck with the They didn't really 45:59 email them much, which is crazy. I 46:00 wonder if he like 46:04 This is a Yeah, it's some snap. Oh, this 46:06 guy emails a lot. Somehow this 46:10 person got lost. 46:12 And then he finally sent a text randomly 46:14 9 months later, and bam, that was a good 46:17 text cuz I got 200 bucks. 46:18 >> reactivated him. Yeah. 46:19 >> And then they loved it, then they sent a 46:21 text again, they bought again. He got an 46:23 extra almost 600 bucks from just 46:25 starting to text them. 46:26 >> power of yeah, text messaging right 46:28 there for 46:29 >> But having the the the And being able to 46:31 text them because they were on the list. 46:32 I don't know, he probably did a cleanup 46:34 of his attentive. He's like, "Oh my god, 46:36 I wasn't texting these people." 46:38 >> Yeah. Yeah, yeah, no, that's But like 46:40 this is this right there, right? Oh, 46:42 that one data point. Now you you're 46:44 going to probably go check multiple like 46:46 that's not one person's not a good 46:48 sample size, but being able to then to 46:50 take that back to your marketing team, 46:52 whether it's internal or an agency or 46:54 whatever, and say, "Hey, hold on a 46:56 second. We just were reviewing the 46:58 customer journeys, and every looks like 47:01 look at the SMSs here that are resulting 47:03 in reactivating 47:05 clients or sorry, not clients, but 47:08 customers who have not purchased from us 47:11 in X amount of time. And so, 47:14 uh you know, this is this is all 47:15 available inside of Wicked Reports, and 47:18 I I do love it cuz I love looking at the 47:20 customer journeys by clicks. 47:22 >> Here's the new cuz the new customer 47:23 journey report's live. It's a little 47:25 sexier. 47:26 So, it'll show all the reports and all 47:28 the products bought. 47:31 And then it's just, you know, visually 47:32 really slick cuz you can filter it. Look 47:35 at this person. So, this was the beauty. 47:36 Meta pros- See, prospecting. 47:39 2020. 47:41 >> Wow. 47:42 >> Still buying off SMS 47:45 yesterday. And here's what they bought. 47:48 And started 6 years ago. 47:50 >> Yeah. 47:51 >> Yeah, so that's pretty cool. This person 47:53 >> I always love when 47:54 >> channels too. Perplexed, so we got AI 47:57 was chatting with this AI. 48:00 Yeah, this person is insane how 48:02 interesting all the different channels. 48:04 So that's pretty cool. 48:04 >> It is cool. Yeah, but that's you know 48:06 that's why I always love when like 48:07 someone's like oh you know well email 48:10 does this for us and it's like well you 48:12 can't go really acquire just from email. 48:14 Like how you got to get people on the 48:16 list, right? So you're still going to 48:17 have to go do cold prospecting through a 48:21 campaign of some sort. 48:23 >> So that brings me to one do I have a dev 48:25 area up because this one is in acquire. 48:26 So oh you can change the colors now 48:28 different thing that's just kind of like 48:29 whatever. But we're working on something 48:32 that I'll tease out. It's just we're 48:34 just verifying some of the math to your 48:36 point. 48:37 Uh halo modeling. So the idea of using 48:40 click data directionally it but if email 48:45 allowing you to weight top of the funnel 48:47 more 48:48 in terms of if you're direct if you have 48:51 a lot of YouTube or TikTok influencer 48:54 that you know isn't going to get a click 48:56 or a ton of email SMS that you know came 48:58 from somewhere else. We are doing the 49:01 first transparent incrementality {slash} 49:05 data driven {slash} halo modeling where 49:07 I'm exposing my model and allowing you 49:09 to tweak it. And so people then are our 49:12 power users can co-create and improve it 49:14 if they want. Um where most people have 49:16 a black box behind that hey this really 49:18 drove like more clicks due to the views 49:21 and you just got to trust it. 49:23 >> Trust it. 49:24 >> showing the model and how it works. So I 49:26 think that I think it's going to be kind 49:28 of cool. So basically the idea is you 49:30 give a prioritization between spend 49:32 versus awareness awareness being views 49:36 and then it's a zero-sum game. So I 49:38 can't just throw more credit at meta. I 49:40 have to take the revenue from somewhere 49:42 cuz if we're going to be 49:43 modeling anything it's got to be a 49:45 foundation of truth which is I want to 49:47 take my direct sales and give them to 49:49 paid. I I you should do that. I think 49:51 most of them direct cake How did they 49:53 hear about your brand that isn't a 49:54 household name? 49:55 >> Right. 49:56 Right. 49:57 >> So you can come in here and decide 50:00 I want to take my organic and give it 50:02 and you can decide how much to give. 50:05 I want to give my unattributed. That's a 50:07 no-brainer usually. I do I want to give 50:10 email and it'll take the credit from 50:12 there and give it to where you want. In 50:15 this case, they want to give a lot of it 50:16 to Tik Tok 50:18 and and a lot to YouTube and and a lot 50:21 to Meta. And and we have a analysis of 50:23 how to set these up. You don't have to 50:25 know that, but the power users you would 50:27 actually probably dig this. You can go 50:29 in cuz if your strategy is I'm going Tik 50:31 Tok influencers and heavy video to cold 50:34 traffic Meta, people aren't always going 50:36 to click on your call to action, but if 50:38 you're high if you're 50:40 um 50:41 if your overall North Star metric of 50:43 maybe it's just revenue, but maybe it's 50:44 new customers. If that's going up when 50:46 you spend on places that aren't likely 50:48 to generate as many clicks because 50:50 that's the way it works, we can adjust 50:52 for that here and use the click data as 50:55 the the foundation 50:57 and then the strategy input to then 50:59 distribute it without overcounting. 51:03 >> Yeah. It's pretty It's pretty wild that 51:05 you just have to 51:05 >> very rudimentary stuff here, but I 51:07 wanted to show it off for a minute. 51:08 >> Yeah. I I like I well I like how you you 51:10 you 51:12 you're open source almost with that 51:13 model cuz it's you know, that's that is 51:16 one thing, you know, whether you're 51:17 using 51:20 another plat Let's get like Triple Whale 51:22 or something. You're just you're 51:23 trusting their model. I mean it is it's 51:25 their model at the end of the day. So 51:27 it's like 51:28 you know, they're looking at attribution 51:30 just within their own modeling. So 51:33 having the ability to go in there and 51:35 customize that and and have it open this 51:37 really cool. 51:39 >> Yeah, cuz then you come in here and you 51:40 apply the changes. It immediately 51:42 applies the changes. 51:44 So you can see the impact and then it 51:45 will recompute. So if you're going to 51:47 use if you turn on Halo, then our then 51:50 our MCP server and our AI says, "Oh, 51:53 okay, they're using Halo, so we should 51:56 use these things instead of their 51:58 existing ones." So, it our AI, we went 52:02 uh and we got this 52:04 it reacts to the strategy. 52:06 So, because you're doing that type of 52:08 strategy, you would turn on Halo, and 52:10 then our AI is like, "Okay, we want to 52:12 use the Halo numbers, not the not the 52:15 the the the pure click numbers as and 52:18 and contrast them, and then educate the 52:20 person, "Well, the Halo numbers say this 52:22 because of your settings, and because if 52:24 your strategy is supporting this, then 52:28 well, let's look at one of them. Then 52:29 then uh then your meta is actually doing 52:31 your N CAC is here instead of whatever 52:33 it is when without this." 52:35 >> Yeah. 52:36 >> So, it 52:37 it kind of like 52:38 knows it it it translates it all into 52:42 spend more heat scale here or kill or 52:44 you got a problem or you don't without 52:46 you having to be like sit here and look 52:48 through like I mean, there's a lot of 52:49 columns here, and we have tons more. You 52:51 don't have to do that. It just knows 52:52 your strategy, and when you're saying, 52:54 "Hey, should I be scaling? How's my meta 52:56 doing?" Well, like, "Oh, you turned on 52:57 Halo. Well, here's how your meta's 52:59 doing. We have 90 click sales, and then 53:02 we have 40 more that are Halo because of 53:04 your settings, and then you can, you 53:06 know, chat back and forth from that 53:08 point from that educated foundation. 53:10 >> Yeah, amazing. 53:13 >> So, 53:14 now the the other piece here is our 53:16 analyst. So, decisions on a platter. How 53:19 What does it mean when we're decision 53:21 certainty? 53:22 Well, I asked this about top of funnel. 53:25 It went and found these 53:26 behind-the-scenes 53:28 slices of attribution. We have 54 of 53:30 them created. So, that we've taken your 53:33 most common questions, of which I've 53:35 listed them 53:39 Get my window. Get you out of the way. 53:41 So, we've got a set of questions that we 53:44 feel are things you should be asking 53:47 your attribution and yours might be 53:49 different. They might be like these. 53:52 The point is is that behind these are 53:53 all the different decisions to make that 53:56 the data is accurate. 53:58 That I'm looking at the right set of 54:00 columns and the right trends and I've 54:01 used the right attribution model and 54:03 I've used the right time period. All 54:05 that's been figured out with these 54 54:07 reviews. 54:08 So that we pick the right one whether 54:11 you're chatting with MCP or you're in 54:13 our tool. 54:15 If you're in an MCP, we have this genie 54:17 mode where you're like you you're asking 54:18 for a wish, you say ask wicked. And that 54:21 tells our MCP server 54:24 don't just let the AI fumble around with 54:26 the tools or maybe be accurate. Use our 54:30 expertise which is like I don't even 54:32 know like 6,000 lines of prompts. 54:36 All the 54 views, all this expertise, it 54:39 goes in and uses that to give you the 54:41 right answer which you then can chat 54:42 with it all you want. Um but the idea 54:45 here is which we're co-creating with our 54:47 customers right now. We have these from 54:49 what we know from doing this. 54:51 Our customers are going to think up all 54:52 kinds of other cool ones and we're just 54:54 going to put them in. 54:55 And then when you get the answer and 54:56 you're like oh I wish it kind of did 54:58 this or this, we can edit and iterate it 55:00 in like a day. It's not like oh I'll put 55:02 in the backlog and maybe in 6 months 55:04 you'll get it. Like we can make an 55:06 adjustment. We have custom prompts that 55:09 can override our global prompts specific 55:12 to your account if you have some unique 55:13 stuff. We can just override it in here. 55:16 So I'm really pumped about it because 55:18 all these things 55:21 are in here that we have all 55:22 >> cool. 55:23 >> All these like 55:25 what's my true profitable end cap? Well 55:26 all the stuff that has to go into that 55:28 is already in here. 55:30 >> Yeah, it's amazing. Yeah that's like I 55:32 mean to me it is the the continued 55:35 future of reporting is you're really 55:37 just it's like chatting to the data. 55:39 >> Yes. 55:40 And and then what we are is we're the 55:42 layer 55:44 before your model that makes sure you 55:47 can trust what the model is going to 55:49 tell you. 55:50 Or at least it's going to have its best 55:52 chance of doing that. 55:54 >> Right. 55:54 >> We have data checks too. Like if the 55:56 data suddenly like your tracking data 55:57 went wrong for 2 days or your budget was 56:00 weird for a couple days. The data check 56:03 agent will be the data 56:05 uh the accuracy or something we call it 56:07 uh 56:08 data accuracy or data verification 56:10 agent. We'd say, "Hey, just so you know, 56:13 blah blah blah about your data. Um do 56:15 you still want me to get it?" Rather 56:17 than you having to go wonder like cuz we 56:19 have all these indicators is the data 56:20 really accurate? It usually is, but I 56:22 mean it might be like, "Hey, four ad 56:24 sets didn't get tracking here. We better 56:26 not recommend anything." 56:28 >> Yeah. 56:29 >> Yeah. 56:30 Yeah. No, I this this piece is uh this 56:33 this piece is great right here. So 56:34 again, this is why this is like 56:37 far superior than you'd like someone 56:39 dumping in a bunch of random data from 56:42 random sources into Claude or ChatGPT 56:45 and then trying to get an answer about 56:47 it when this is all powered by this 56:50 inside of Wicked Reports. 56:52 >> Yeah. 56:53 Yeah, we we we show this prompt here. 56:56 Behind the scenes 56:58 50 to 700 to 2,000 lines of prompt. We 57:02 try to keep it brief so we're not 57:03 getting crushed on tokens. 57:05 >> We aren't uh right now we're letting our 57:06 customers do what they want with it and 57:08 just keeping an eye on it in the 57:09 background. 57:09 >> Wow. 57:10 >> Uh but if it's if it's in our app, if 57:12 it's your AI, then you know, you you're 57:14 going to be paying that for it. 57:15 >> Sure. 57:15 >> Our MCP narrows down the columns cuz you 57:18 can pay more. If you just let AI figure 57:21 it out, it brings back you know, 60 57:23 columns. You know, you're going to be 57:24 paying for 40 columns, 2/3 of that you 57:27 probably didn't need that you just 57:29 burned through money. 57:30 >> Sure. 57:31 >> we have our way to 57:32 be efficient on that. 57:33 >> Yeah. Yeah. Yeah. This is amazing. 57:37 >> Yep. Then you got them by category. So, 57:39 it's kind of cool. These ones are 57:41 coming. I didn't like what they were 57:42 saying yet, so we put them back to 57:43 coming soon. But for an agency, talk 57:46 tracks 57:48 and agency things. Uh I mean, you get an 57:51 agency 57:52 Give me which client am I most at risk 57:53 of losing? Which one needs my attention? 57:56 Prove this drill pipe. 57:59 Um I've got a couple more here that I've 58:01 You probably have a couple where you 58:02 like Jeez, I always have to go in and 58:03 like Well, you know, here's why we're 58:05 spending at top of the funnel. 58:07 So, like this one. 58:09 >> [laughter] 58:09 >> Yeah. 58:10 >> You know? I mean, but we got them in 58:11 there. Just see Just one less thing you 58:13 got to do. We've had to do this, you 58:14 know, for 10 years help our agencies 58:16 explain to clients why they're they're 58:19 not spending on the 12x brand search. 58:22 They're spending on the 1.3 58:25 top of funnel meta campaign. Well, or or 58:28 why this 58:30 Why am I spit Why am I paying more cost 58:32 per click for this set of keywords? 58:34 Well, it's cuz that's where all the LTV 58:35 customers are coming from. 58:37 >> Right. Right. Right. 58:38 >> costs more because they're more valuable 58:40 to acquire. 58:41 >> Yeah. No, but having these pre-packaged 58:44 and populated for people to use is 58:46 excellent. 58:47 >> Because then in the MCP, you can then 58:48 have a cloud code, you know, if you're 58:50 savvy with it, which I heard you were 58:52 for sure if you're doing that with the 58:53 podcast and the and the blogs. You could 58:56 have it cycle, hit the MCP, give it, and 58:59 create your weekly report or your slack 59:02 message you're going to send to the 59:03 client. It can be a data-driven one that 59:05 you're just approving and having it 59:07 sent. 59:07 >> Yeah. 59:08 >> Cuz all this thinking is done. You can 59:10 just override it with your own or, you 59:11 know, 59:12 customize it. 59:13 >> Yeah. Yeah. That's amazing. 59:16 >> Yep. So, 59:18 that's 59:19 uh that's kind of my my dog and pony 59:21 show today. 59:22 >> Incredible. 59:24 Uh some of the beta testers right now, 59:27 um you know, 59:28 what what's the feedback from them so 59:30 far? 59:31 >> They're quite happy with it so far. I 59:33 mean, this has been the 59:35 first Well, I mean, we used 59:37 when we were the first ones to really 59:39 track for SMBs, meaning under a million 59:41 dollar companies in revenue, 59:44 at all. I mean, our our sweet spot's 5 59:46 to 50 million, but when we first got 59:47 into this, 59:49 we went pretty viral cuz like Frank 59:51 Kern, I don't know if he's still even 59:53 doing anything, but he would be speaking 59:54 to us at the Digital Marketer events, 59:56 and then people would be just like, "Oh 59:57 my god, this And this is the first this 59:59 this stuff is where people we don't know 1:00:02 are that are that are customers are 1:00:04 like, "Hey, I heard this is available. 1:00:06 Can I get in on it?" Like that's the 1:00:07 type of 1:00:08 >> Yeah. 1:00:09 >> spread that's happening, and it's barely 1:00:11 out. It's been out, you know, it's been 1:00:12 in hands for like a couple days, and 1:00:14 we're getting to hit up already. 1:00:16 So, that's a good good sign. And then 1:00:18 the the first webinars were 100% 1:00:20 attendance, which 1:00:22 You can get 20% live attendance 1:00:23 nowadays, it's great. I mean, these were 1:00:24 small ones, you know, like 10, 20 1:00:27 cohorts. 1:00:27 >> Doesn't matter. 100% 1:00:29 >> showed?" I was like, "Oh my god, 1:00:31 I was like, "I just want to be talking 1:00:33 to my product manager and sending the 1:00:35 recording." 1:00:37 >> Yeah. 1:00:37 >> showed and was active. So, it's a good 1:00:39 sign that we're on the on the right 1:00:40 track. I'd 1:00:41 >> No, that's incredible. Um, you know, 1:00:43 this is why, you know, 1:00:45 again, 1:00:47 companies who are, you know, just stuck 1:00:50 looking at in app metrics, I mean, you 1:00:52 know, there are tools like Wicked 1:00:53 Reports out there, um, you know, that 1:00:57 can that have a lot of answers for you 1:00:59 in your data to make better uh, 1:01:02 decisions with your ad campaigns versus 1:01:05 just looking at, "Oh, here's ROAS. This 1:01:08 is ROAS is high, so let's just continue 1:01:10 to pour money into that campaign." Where 1:01:12 you, you know, this pulls all of that, 1:01:16 you know, all the layers back, and 1:01:18 you're able to really understand what's 1:01:20 going on. And even if you're not as 1:01:22 sophisticated as many some, you know, 1:01:24 nerds might be with all of this, 1:01:27 it Wicked Reports now is is 1:01:29 got all of these AI features baked into 1:01:32 it that make it very simple to go use to 1:01:35 go talk to the 1:01:36 data and be able to then make your 1:01:38 decisions for you. So, and 1:01:40 you know, listen, as a user, by the way, 1:01:42 folks, this is you know, I don't this is 1:01:44 not like some affiliate deal. Scott did 1:01:46 not pay me to to get on the YouTube 1:01:48 channel. I am just an advocate for 1:01:51 Wicked Reports and and we, you know, 1:01:54 have have had success with our clients 1:01:56 that 1:01:57 are using it. So, 1:01:59 I just I love what is being built at 1:02:02 Wicked Reports. 1:02:03 And the companies that get into it are 1:02:06 able to break away from this like ROAS 1:02:08 paradigm only and I think that's just 1:02:11 how companies need to 1:02:14 uh 1:02:15 look at at their data in order to 1:02:17 continue to actually scale and grow 1:02:20 profitably. And a the tool like Wicked 1:02:22 Reports makes that super simple. 1:02:24 >> Mhm. Yeah, I appreciate that, you know. 1:02:27 It's decisions, not dashboards. So, 1:02:30 getting to the the whole point of what 1:02:32 we're doing. 1:02:33 That get to the outcome that you need, 1:02:35 which is I need to 1:02:37 I don't need to be looking at I love 1:02:39 looking at data, but 1:02:41 but I don't you as if you're busy, you 1:02:43 don't need to look at data. Just tell me 1:02:45 the decisions I need to make 1:02:48 or the decisions you've already made for 1:02:49 me based on what I've told you is 1:02:51 important to my business and what my 1:02:53 marketing strategy is. Apply a 1:02:55 measurement strategy to that to give me 1:02:57 the insights so I can 1:02:59 improve. That's what we're doing now. 1:03:01 So, it's a great relief to me to see 1:03:03 that. 1:03:03 >> not data. 1:03:05 >> Mhm. 1:03:05 >> I like that. Decision Yeah, I like that 1:03:07 a lot. 1:03:08 Okay, so Scott, this will be out like if 1:03:11 someone comes and signs up for Wicked 1:03:12 Reports or maybe they're already a a 1:03:14 Wicked Report customer, this is going 1:03:17 going to be out sounds like to everyone 1:03:19 after July 4th then? 1:03:20 >> Yeah, so right now that that a couple of 1:03:24 those screens and the MCP server are 1:03:26 already in people's hands. 1:03:29 The Ask Wicked Genie, I feel like I'll 1:03:32 release July 1st, maybe sooner cuz it's 1:03:35 it's looking good, but I want to know if 1:03:37 I'm putting my neck out that it's it 1:03:39 works as good as that printed out report 1:03:41 does. 1:03:42 Um so that'll be July 1st. 1:03:45 And then the attribution report is 1:03:47 coming mid-July, but that already the 1:03:48 MCP will already have all that data in 1:03:50 there. 1:03:51 So by July 15th, everything there except 1:03:54 for maybe that attribution report cuz I 1:03:56 don't know um 1:03:58 it just depends on testing. It'll be in 1:04:00 testing. So people that are using Wicked 1:04:02 Reports now and get this yeah, the pulse 1:04:04 and the MCP we've been we've messaged a 1:04:06 few people, but it'll be wide release by 1:04:08 mid-July and people can get on it now if 1:04:10 they watch this. 1:04:11 >> Okay, amazing. Well, 1:04:13 I will have, you know, the link to 1:04:16 your site. It's very simple though, 1:04:17 wickedreports.com. 1:04:19 That will be in the description below. 1:04:21 Scott, I appreciate you coming back on. 1:04:23 I can't believe it's been a full year. 1:04:26 I didn't realize 1:04:27 >> Mind-blowing. 1:04:28 >> a year has gone by that quickly. It is 1:04:30 funny I was on a call yesterday with 1:04:32 another guy. 1:04:34 He he owns another agency. So someone I 1:04:36 met up with last last year, but we were 1:04:38 like, oh when when was that? And I in my 1:04:40 head it felt like 6 months ago. I was 1:04:41 like, oh that's been a full year. So you 1:04:44 know, time time just flies. But 1:04:47 yeah, 1:04:48 a lot a lot you've obviously been 1:04:50 working on over the last year. So 1:04:52 exciting to see the continued evolution 1:04:55 of Wicked Reports. And 1:04:58 I will absolutely let you know, more of 1:05:00 my clients know about it that are who 1:05:03 are not on Wicked Reports already. 1:05:05 >> That's awesome. Thanks. Always a good 1:05:07 time to be on here. 1:05:08 >> Absolutely, man. All right, until next 1:05:10 time. Hopefully it's not another full 1:05:12 year. I appreciate you on and everyone 1:05:14 go check out Wicked Reports.