Meeting Title: Attribution Solution Pitch with Joseph Date: 2026-01-13 Meeting participants: Robert Tseng, Zoran Selinger, Luke’s Notetaker, Luke Scorziell, Joseph
WEBVTT
1 00:00:21.270 ⇒ 00:00:25.460 Robert Tseng: Yeah, I’m assuming this goes well, like, I want to basically start,
2 00:00:25.720 ⇒ 00:00:30.270 Robert Tseng: You know, once we start actively pushing the service line on…
3 00:00:30.480 ⇒ 00:00:36.059 Robert Tseng: Kind of in content, or, like, with our network as well, like, you know, future calls where, you know, this type of
4 00:00:36.220 ⇒ 00:00:41.509 Robert Tseng: discovery… call comes in, would…
5 00:00:41.740 ⇒ 00:00:43.689 Robert Tseng: And can start to leap you guys in.
6 00:00:46.280 ⇒ 00:00:50.720 Luke Scorziell: So with, like, kind of asking about his ownership, if he owns…
7 00:00:52.580 ⇒ 00:00:56.900 Luke Scorziell: Yeah, and then I guess authority, if he’s the one that actually makes the decision, or if someone else is.
8 00:00:56.900 ⇒ 00:01:11.289 Robert Tseng: Yeah, well, I mean, those all… all those things, I think, are already… we know that already. Like, this is… this is just his business. Yeah, he’s, like, he, like, sells books online, right? And, like, I don’t know, he spends 200 grand a month. That’s… that’s what he’s given us so far.
9 00:01:11.620 ⇒ 00:01:13.170 Luke Scorziell: Yeah, okay. Yeah.
10 00:01:14.030 ⇒ 00:01:16.380 Luke Scorziell: Are there other things that you’d want to qualify with him?
11 00:01:19.760 ⇒ 00:01:26.520 Robert Tseng: Okay, well, I mean, if I were to be, kind of, Walked into the conversation, like.
12 00:01:27.410 ⇒ 00:01:30.350 Robert Tseng: I’d probably ask him kind of things like.
13 00:01:35.050 ⇒ 00:01:50.720 Robert Tseng: growth trajectory, like, is he spending more… is he expecting to spend more? Is he expecting to spend less? Like, kind of, like, what’s… you know, just try to understand his business a bit more. I think, you know, we know that it’s an e-com, like, book… e-books business, but if he has new product lines coming out, like, we’re just trying to, like.
14 00:01:52.150 ⇒ 00:01:59.389 Robert Tseng: We’re basically trying to size this business, without directly asking, like, what’s your revenue? I think that’s, like, a big piece of it.
15 00:01:59.510 ⇒ 00:02:02.359 Robert Tseng: And then, like, if he has any other kind of, like.
16 00:02:03.010 ⇒ 00:02:11.970 Robert Tseng: who’s on his technical staff? Like, you know, what tech… what stack does he use? Does he use? Like, is this a purely Shopify, like, website? Like.
17 00:02:12.260 ⇒ 00:02:17.120 Robert Tseng: Trying to understand, are there any other entry points into, like, the…
18 00:02:17.260 ⇒ 00:02:23.699 Robert Tseng: the tooling for, like, from the tooling that he uses. Yeah, I think those are probably the main things I’d ask.
19 00:02:33.330 ⇒ 00:02:34.460 Zoran Selinger: Welcome to…
20 00:03:05.080 ⇒ 00:03:13.080 Robert Tseng: Okay, if he doesn’t show up in, like, 2 minutes, then we’ll just kind of run through it, like, as is. This video’s recorded, so I can just send him the link.
21 00:04:36.410 ⇒ 00:04:42.920 Robert Tseng: I’m just on the website right now. I’m actually gonna just go and figure out, like, what is this deck? I think I could probably figure it out.
22 00:04:46.500 ⇒ 00:04:48.169 Luke Scorziell: Yeah, looks like he has a couple different books.
23 00:04:48.650 ⇒ 00:04:49.470 Luke Scorziell: For now.
24 00:04:50.050 ⇒ 00:04:50.870 Luke Scorziell: So…
25 00:04:58.540 ⇒ 00:05:00.529 Robert Tseng: in the body of Cat B.
26 00:05:01.650 ⇒ 00:05:05.846 Robert Tseng: It’s available… It works.
27 00:05:11.920 ⇒ 00:05:13.390 Robert Tseng: Yeah, it’s a Shopify.
28 00:05:15.400 ⇒ 00:05:22.529 Robert Tseng: That’s interesting. And he also blocks you from looking at his source code, which… Same, probably uses some CDN.
29 00:05:23.160 ⇒ 00:05:27.039 Robert Tseng: Goodbye, what is this?
30 00:05:28.090 ⇒ 00:05:29.190 Robert Tseng: two tools.
31 00:05:31.320 ⇒ 00:05:34.730 Robert Tseng: Yeah, he has Klaviyo, he has,
32 00:05:36.480 ⇒ 00:05:44.059 Robert Tseng: GTM, some basic GTM setup, a few different, like, payment things set up. What is rebuy?
33 00:05:46.570 ⇒ 00:05:52.939 Robert Tseng: Some personalization tool. I’m assuming it’s probably, like, Yapo or something.
34 00:05:53.500 ⇒ 00:05:59.750 Robert Tseng: Yeah, some, like, checkout flow tool that may, like, customize pricing based on the person that comes on.
35 00:06:00.710 ⇒ 00:06:08.889 Robert Tseng: And… Okay, alright, yeah, I think I understand his… his stack. It’s not… it’s not that complicated.
36 00:06:09.310 ⇒ 00:06:16.590 Robert Tseng: Okay, yeah. It’s a purely Shopify store, he has Klaviyo, he has Rebuy, he has… he has something set up on Google Tag Manager.
37 00:06:17.060 ⇒ 00:06:18.980 Robert Tseng: Okay, yeah, seems pretty straightforward to me.
38 00:06:19.300 ⇒ 00:06:23.600 Robert Tseng: Alright, well, yeah, I guess, like, we can… we can just get…
39 00:06:23.930 ⇒ 00:06:28.219 Robert Tseng: started. Yeah, let’s just kind of, like, time it so…
40 00:06:28.600 ⇒ 00:06:35.509 Robert Tseng: yeah, you know, I want to be able to just kind of repurpose this, and then we can maybe… yeah, I can share this pitch, to…
41 00:06:35.890 ⇒ 00:06:46.980 Robert Tseng: more than just this guy, so, we can just view it as, like, you guys are recording for, like, an async video that I might send over to other, other leads.
42 00:06:47.590 ⇒ 00:06:48.939 Luke Scorziell: Oh, there he is.
43 00:06:49.760 ⇒ 00:06:51.379 Joseph: Hey, Robert. Hey, Luke. My apologies.
44 00:06:51.380 ⇒ 00:06:52.130 Robert Tseng: Hey, Joseph.
45 00:06:52.530 ⇒ 00:06:53.510 Luke Scorziell: Hey, how’s it going?
46 00:06:53.510 ⇒ 00:06:57.450 Joseph: I just got back with my… My daughter from a walk.
47 00:06:57.730 ⇒ 00:06:58.630 Robert Tseng: Oh, no worries.
48 00:06:58.760 ⇒ 00:07:00.339 Robert Tseng: Where are you based out of?
49 00:07:00.340 ⇒ 00:07:01.800 Joseph: I’m in Orlando, you?
50 00:07:01.990 ⇒ 00:07:05.220 Robert Tseng: Okay, great. We’re kind of all over. I’m New York.
51 00:07:06.660 ⇒ 00:07:08.800 Luke Scorziell: In LA, and Los Angeles.
52 00:07:08.800 ⇒ 00:07:13.890 Joseph: Nice. I go to both those places pretty frequently, so that’s cool. I was in…
53 00:07:14.200 ⇒ 00:07:18.700 Joseph: LA maybe about a month or so ago, and then New York,
54 00:07:18.970 ⇒ 00:07:21.529 Joseph: Maybe about 3 months ago, or something like that.
55 00:07:21.830 ⇒ 00:07:30.809 Robert Tseng: Very cool. Yeah, I mean, I’m from California, and I mean, a lot of New Yorkers flee to Florida during the winter, so…
56 00:07:30.910 ⇒ 00:07:33.769 Robert Tseng: I’ll actually be going to Orlando in a few weeks, so…
57 00:07:34.030 ⇒ 00:07:34.680 Joseph: Nice.
58 00:07:34.680 ⇒ 00:07:35.200 Robert Tseng: Yeah.
59 00:07:35.700 ⇒ 00:07:43.890 Robert Tseng: I guess, yeah, I have Zord here on my team as well, so I guess, kind of, we were thinking, and we have a deck put together, he’d love to, like, just kind of walk you through
60 00:07:44.060 ⇒ 00:07:52.890 Robert Tseng: you know, just the details behind this solution that we’ve built. So we’re thinking we could probably spend, like, 10 minutes there, and then…
61 00:07:53.010 ⇒ 00:08:00.539 Robert Tseng: yeah, obviously stop if you have any questions, and I guess I’ll just pretty much let Luke and Zoran do the talking of…
62 00:08:00.990 ⇒ 00:08:02.829 Robert Tseng: Yeah, just trying to…
63 00:08:03.050 ⇒ 00:08:18.280 Robert Tseng: educate you on what we do, and yeah, see if there’s any way we can help adjacently, or, you know, if you have any questions for your learning, like, they can take it as well. So, yeah, if that sounds good, then Zoran, I’ll hand it over to you, you can kind of jump into it.
64 00:08:18.730 ⇒ 00:08:33.829 Zoran Selinger: Sure, sure. Hi, Joseph, nice to meet you. Yeah, so let’s go through this a little bit. We won’t get super technical at this point, but obviously, we want to answer any questions that you have.
65 00:08:34.220 ⇒ 00:08:52.080 Zoran Selinger: So, you know this, in today’s online world, we do deal with a lot of, a lot of, tracking prevention, which really decreases the, the quality of, of the data that we can, that we can typically collect, and what every… we went from
66 00:08:52.110 ⇒ 00:09:11.610 Zoran Selinger: you know, 5% of things that we cannot track 5 years ago, to now maybe jumping to 40-50% for some businesses, which is obviously a big problem. That kind of inbuilt error in our data is really, really hard to work with. So…
67 00:09:11.960 ⇒ 00:09:17.260 Zoran Selinger: We recently developed, something really interesting, where we are
68 00:09:17.870 ⇒ 00:09:22.539 Zoran Selinger: Successfully recovering the signal for attribution.
69 00:09:22.720 ⇒ 00:09:32.640 Zoran Selinger: We have 3 levels, there. One is… one is basically we can confirm the tools that you are already using.
70 00:09:32.790 ⇒ 00:09:34.200 Zoran Selinger: We…
71 00:09:34.640 ⇒ 00:09:47.469 Zoran Selinger: basically, it’s almost… almost 100% there when it comes to, the volume of events and sessions that you’re getting. We can… we can really reliably,
72 00:09:47.590 ⇒ 00:09:56.230 Zoran Selinger: Essentially, verify the tools that you already have, but then this gives us also the opportunity to do more.
73 00:09:56.230 ⇒ 00:10:02.260 Zoran Selinger: And we can start building some business logic, at the same moment.
74 00:10:02.260 ⇒ 00:10:25.140 Zoran Selinger: and kind of introduce sessionization and user identifiers, so we start doing identity stitching and all those type of things that are crucial for attribution. There’s also, on the, on the third level, we start integrating the same solution
75 00:10:25.180 ⇒ 00:10:27.849 Zoran Selinger: to your existing tools. So…
76 00:10:28.050 ⇒ 00:10:40.629 Zoran Selinger: And they… basically, what happens there is… is a… is a synergy, because we are improving, for example, on the edge, we are improving identity resolution, while also
77 00:10:40.630 ⇒ 00:10:55.180 Zoran Selinger: while also improving the accuracy of your existing tools that you have. This is the effect we got from a few of our case studies so far. We are very, very happy about it and excited about it.
78 00:10:55.220 ⇒ 00:11:02.689 Zoran Selinger: So the… basically, the goal is to… to capture the data before
79 00:11:03.170 ⇒ 00:11:09.680 Zoran Selinger: We get into a customer’s… we get to a client layer, meaning your browser.
80 00:11:09.850 ⇒ 00:11:14.899 Zoran Selinger: And all the other, you know, all the tracking prevention stuff comes, comes into place.
81 00:11:15.100 ⇒ 00:11:25.789 Zoran Selinger: And obviously, very important, you don’t have to abandon any of your current tools that you’re… that you’re using, that you’re used to, you have… you have built your…
82 00:11:28.750 ⇒ 00:11:34.230 Zoran Selinger: to be replaced at all. These can fit really nicely into that.
83 00:11:34.590 ⇒ 00:11:40.119 Zoran Selinger: Looks like. Let me know if you see, that I have connection problems.
84 00:11:40.120 ⇒ 00:11:40.719 Joseph: You’re good, you’re back.
85 00:11:40.720 ⇒ 00:11:46.790 Zoran Selinger: it’s feeding me that my internet is unstable at the moment.
86 00:11:47.230 ⇒ 00:11:53.080 Zoran Selinger: So, you’ve seen this, that we, that…
87 00:11:53.790 ⇒ 00:12:02.500 Zoran Selinger: In particular, we have these problems with third parties. Even first-party cookies are, in some browsers, very limited in terms of
88 00:12:02.810 ⇒ 00:12:07.299 Zoran Selinger: In terms of, duration and,
89 00:12:07.750 ⇒ 00:12:18.290 Zoran Selinger: ClientScript did get blocked GTM. Like I said before, we were dealing with 5% inaccuracies with GTM and Google Analytics. Now we are…
90 00:12:26.040 ⇒ 00:12:31.629 Robert Tseng: All right, well, I’ll give him, like, a minute. If not, I can… I can take over his presentation.
91 00:12:35.330 ⇒ 00:12:39.599 Luke Scorziell: I guess, Joe, do you have any questions or anything at this point?
92 00:12:40.320 ⇒ 00:12:41.540 Zoran Selinger: Apologies.
93 00:12:41.960 ⇒ 00:12:43.570 Zoran Selinger: I don’t know what’s happening.
94 00:12:43.900 ⇒ 00:12:44.470 Zoran Selinger: I’ve been…
95 00:12:44.470 ⇒ 00:12:45.150 Luke Scorziell: Okay.
96 00:12:45.150 ⇒ 00:12:48.650 Zoran Selinger: It’s been a while since I’ve had unstable connection.
97 00:12:49.210 ⇒ 00:12:55.739 Zoran Selinger: Or… And it’s… I don’t know. It’s happening now.
98 00:12:56.060 ⇒ 00:12:56.889 Zoran Selinger: And it’s really…
99 00:12:56.890 ⇒ 00:12:57.710 Robert Tseng: All good, all good.
100 00:12:57.970 ⇒ 00:12:58.770 Robert Tseng: Okay.
101 00:12:59.140 ⇒ 00:12:59.700 Zoran Selinger: Wow.
102 00:13:00.050 ⇒ 00:13:02.900 Zoran Selinger: So…
103 00:13:03.320 ⇒ 00:13:10.369 Zoran Selinger: The errors we make when we… when we have that kind of inaccuracy on the client side is…
104 00:13:10.370 ⇒ 00:13:28.740 Zoran Selinger: we are often pausing winning channels, because we don’t really see the full picture of what they, what they bring to the table. So that happens a lot. Instead of scaling particular channels, we do the opposite. We pause them, because we don’t actually know, what exactly,
105 00:13:29.450 ⇒ 00:13:35.730 Zoran Selinger: And they brought for the business, which is obviously a big problem. Retargeting gets…
106 00:13:36.110 ⇒ 00:13:52.259 Zoran Selinger: really, really tricky. We do lose… we do see multiple, multiple, you know, identifiers as being different people, but they are actually the same a lot of the time, so we… we can… we can start treating,
107 00:13:52.300 ⇒ 00:13:59.819 Zoran Selinger: existing customers, like, potential new customers in our campaigns, which gets really annoying. I know this happened to you.
108 00:13:59.820 ⇒ 00:14:13.719 Zoran Selinger: A lot of the time, some. You purchase a product, and you start seeing, ads, like, like you’re not already a customer, so this happens a lot. When we do identity stitching properly, this shouldn’t happen.
109 00:14:15.010 ⇒ 00:14:30.810 Zoran Selinger: the messaging should be tailored and precise. Obviously, when you’re in a bigger team, this really, really causes a lot of discussions about, are these numbers correct or not?
110 00:14:31.070 ⇒ 00:14:39.819 Zoran Selinger: We have multiple systems, multiple people that are trying to check them, all getting different, different numbers. We don’t really have a good source of truth.
111 00:14:40.340 ⇒ 00:14:58.730 Zoran Selinger: looking at actual server logs is basically impossible these days. So, we don’t really have an actual source of truth. With this, we are getting almost to, you know, very, very high level of confidence, and close to the source of truth.
112 00:14:58.880 ⇒ 00:15:00.850 Zoran Selinger: So…
113 00:15:01.550 ⇒ 00:15:11.740 Zoran Selinger: one of the case studies that we have, is, is… is this. So, we noticed around 70%,
114 00:15:13.020 ⇒ 00:15:22.320 Zoran Selinger: Transactions were completely invisible to a customer that we have, and this is a $6 million company a month.
115 00:15:22.660 ⇒ 00:15:26.649 Zoran Selinger: So it’s, it’s very, very significant.
116 00:15:27.130 ⇒ 00:15:32.109 Zoran Selinger: We also… they also had a system where they would, when they would,
117 00:15:33.150 ⇒ 00:15:42.980 Zoran Selinger: credit partners, affiliate partners, with conversions, and we just… they couldn’t check exactly, whether they… they are.
118 00:15:43.390 ⇒ 00:15:56.549 Zoran Selinger: they are crediting them correctly or not, and they were overpaying significantly. And using this, we were able to, we were able to confirm that this is what’s happening, and actually calculate
119 00:15:56.770 ⇒ 00:16:08.319 Zoran Selinger: how much, how much is it? So, like I said, like I’m saying, this is… this is the source of truth, a really reliable source of truth that we now have.
120 00:16:08.480 ⇒ 00:16:12.789 Zoran Selinger: Really good thing about it, we can test it pretty quickly.
121 00:16:13.250 ⇒ 00:16:22.510 Zoran Selinger: in, you know, in 30 days. So the setup isn’t that… isn’t that complicated at all, and pretty quickly, we can… we can start
122 00:16:23.180 ⇒ 00:16:29.670 Zoran Selinger: We can start seeing some… some real data, and start validating the numbers you have.
123 00:16:30.590 ⇒ 00:16:40.100 Zoran Selinger: kind of that discrepancy between how much you see and how much the edge layer sees starts to become really clear pretty early on. And then.
124 00:16:40.190 ⇒ 00:16:55.119 Zoran Selinger: Once… once that is done, once you’re confident this is… this is something to… to look into, then we start building business logic and really starting to, to power up your, your data, and obviously downstream
125 00:16:55.270 ⇒ 00:16:57.360 Zoran Selinger: All your systems that you have.
126 00:16:59.760 ⇒ 00:17:01.530 Zoran Selinger: Yeah, so that’s…
127 00:17:04.000 ⇒ 00:17:10.870 Zoran Selinger: That’s it. So your questions, and then I’m sure Luke will also have some questions for you.
128 00:17:12.210 ⇒ 00:17:19.630 Joseph: Yeah, for sure. Yeah, thanks for sharing a lot of this. So we’ll get, we’ll dive into, like, some of the details a little bit more.
129 00:17:19.750 ⇒ 00:17:22.709 Joseph: So the first few questions are, like.
130 00:17:23.109 ⇒ 00:17:29.819 Joseph: you know, I know it from a super high level, based on what you said, like, how it works, but there’s not really, like, a…
131 00:17:30.040 ⇒ 00:17:33.959 Joseph: Like, a clear, like, here’s exactly the tech stack that we use.
132 00:17:34.920 ⇒ 00:17:43.480 Joseph: why it works. Like, if you’re, for example, is it server-side tracking? Is it not? Is it something different? Are you using STAPE? Like…
133 00:17:43.480 ⇒ 00:17:47.690 Zoran Selinger: even before server-side tracking, so this is at the request level.
134 00:17:48.130 ⇒ 00:18:01.180 Zoran Selinger: So when some… it’s at the request level. This is why it works. It’s at the request level. At that point, we have all the data we can collect, and it’s enough to do really good attribution.
135 00:18:01.560 ⇒ 00:18:07.930 Zoran Selinger: On the request level, you have access to, you know, sources…
136 00:18:08.460 ⇒ 00:18:15.560 Zoran Selinger: potentially cookies and everything that you would need to do that kind of work. So, it works really nicely.
137 00:18:16.200 ⇒ 00:18:17.790 Joseph: Okay, interesting.
138 00:18:17.790 ⇒ 00:18:27.650 Zoran Selinger: You know, for some platforms, like Cloudflare, it’s actually not ex… even… it’s not expensive either, so it’s a really nice,
139 00:18:28.110 ⇒ 00:18:31.579 Zoran Selinger: you know, a key addition to the system.
140 00:18:31.580 ⇒ 00:18:31.970 Joseph: you’re talking.
141 00:18:31.970 ⇒ 00:18:33.819 Zoran Selinger: Right there, to use.
142 00:18:33.820 ⇒ 00:18:40.779 Joseph: At the request level, is that what Cloudflare sees, or, like, who is seeing the information first that you’re accessing?
143 00:18:41.300 ⇒ 00:18:48.360 Zoran Selinger: So that would be… yeah, that would be your… basically, your DNS provider sees, sees the request.
144 00:18:48.470 ⇒ 00:19:07.949 Zoran Selinger: I mean, on a really basic level, you could maybe do a lot of that with, you know, just server logs, which in modern world doesn’t really exist, no one really deals with that anymore. But yes, so request level would be at your, basically, your DNS.
145 00:19:08.140 ⇒ 00:19:20.889 Zoran Selinger: provider. So that’s the first touchpoint. And then, at that point, we can do… you know, business logic means you can claim PIIs, you can… you can do a little bit… obviously, Cloudflare does a lot of…
146 00:19:20.890 ⇒ 00:19:29.799 Zoran Selinger: like, bot removals, then they do it themselves, but if you need more customized logic, you can do that in there.
147 00:19:29.910 ⇒ 00:19:32.569 Zoran Selinger: That is… What is your oystern there?
148 00:19:33.390 ⇒ 00:19:37.939 Joseph: Okay, nice. So then, at the request level, you get that, then where does it go? Is that… is that when it hits.
149 00:19:37.940 ⇒ 00:19:40.660 Zoran Selinger: So you just… what’s really not…
150 00:19:40.660 ⇒ 00:19:41.029 Joseph: I should buy.
151 00:19:41.030 ⇒ 00:19:53.720 Zoran Selinger: about it is that you don’t have to wait for anything to finish. Your users just see your page loading normally, and all the kind of backend work just works.
152 00:19:53.850 ⇒ 00:20:09.940 Zoran Selinger: asynchronously in the background. So there’s no… well, there’s absolutely no performance impact on the website here, yeah, which is really, really nice. Because, I mean, we are doing some asynchronous work at that point, so, yeah.
153 00:20:10.450 ⇒ 00:20:15.810 Joseph: Okay, nice. And then, so once it… once you’re able to get that data from the, you know, at a request level, then…
154 00:20:16.750 ⇒ 00:20:25.819 Zoran Selinger: like, where are you getting the information next? Is that… Yeah, so that… that lands… that will land in your data warehouse, in most cases, right?
155 00:20:25.820 ⇒ 00:20:37.989 Zoran Selinger: Yeah, just whatever you have sent… where you have centralized data, it lands there. We… I mean, we are not limited by… we are only limited by the API that we use, right?
156 00:20:38.030 ⇒ 00:20:39.050 Zoran Selinger: Yeah.
157 00:20:39.050 ⇒ 00:20:44.570 Joseph: So does that mean that you’re only working on a… on that request level and not even using server-side at all, or…
158 00:20:44.850 ⇒ 00:20:45.849 Joseph: Anything like that?
159 00:20:46.110 ⇒ 00:20:54.549 Zoran Selinger: Yes. I mean, obviously, a part of our work, we still want to have client tracking as much as possible, and part of having client tracking
160 00:20:54.790 ⇒ 00:20:55.650 Zoran Selinger: is…
161 00:20:56.700 ⇒ 00:21:15.230 Zoran Selinger: to push it to where it can go, we would have server-side setup anyway, right? Try to be in a first-party context as much as possible. That is all part of this. I mean, we see it as a separate part, but like I said in the presentation, we are not expecting anyone to remove
162 00:21:15.300 ⇒ 00:21:20.660 Zoran Selinger: and what they have currently. And we would love to go in and also fix
163 00:21:20.790 ⇒ 00:21:30.019 Zoran Selinger: the client-side stuff as well. Your Google Tag Manager, for example, and if you see holes in… if that can be more precise, right?
164 00:21:31.020 ⇒ 00:21:41.640 Zoran Selinger: We would also like to do that, because those things, when you get to the level 3 of this implementation, they really do, they really do benefit each other.
165 00:21:41.930 ⇒ 00:21:51.799 Joseph: Okay, like, so you say that server-side and the rest of it, like, browser, cookies, that just, like, enriches the request level stuff that you’re doing, or how do you usually.
166 00:21:51.800 ⇒ 00:21:59.800 Zoran Selinger: It’s mostly the opposite, but the way it enriches, we would use identifiers from
167 00:22:00.250 ⇒ 00:22:03.590 Zoran Selinger: All those other… all those other systems that you have.
168 00:22:03.590 ⇒ 00:22:05.169 Joseph: And that’s the identity that you’re talking about.
169 00:22:05.170 ⇒ 00:22:22.970 Zoran Selinger: Yeah, and that’s the identity station that is… it’s improving it a little bit, right? The Edge by itself is very good at it. It just improves it a little bit with all those other systems. So that’s the way it can, you know, they enrich each other.
170 00:22:23.450 ⇒ 00:22:24.570 Joseph: Okay, nice.
171 00:22:26.090 ⇒ 00:22:35.399 Joseph: Other things… I guess, like, one other bigger question is more so, attribution and how you guys handle that.
172 00:22:35.580 ⇒ 00:22:45.169 Joseph: Yeah, it’s good to know where you’re overspending, underspending, which ads are actually producing results. You mentioned that you’re doing a full funnel type of…
173 00:22:45.350 ⇒ 00:23:00.459 Joseph: analysis on where people are coming from. One of the hardest things to do in attribution is to attribute it and weight it properly. So, I’m curious, with you guys, how are you thinking about that, to not overweight certain things or not? For example, most platforms…
174 00:23:00.790 ⇒ 00:23:01.240 Joseph: use…
175 00:23:01.240 ⇒ 00:23:07.149 Zoran Selinger: Yeah, that will mostly come from the client themselves. We obviously do have our preferred stack.
176 00:23:08.680 ⇒ 00:23:17.680 Zoran Selinger: of, you know, third-party tools that… that do that. But we do have, also, we do have capability to, to build
177 00:23:17.910 ⇒ 00:23:21.869 Zoran Selinger: attribution models ourselves, in the house, so that, yeah.
178 00:23:22.620 ⇒ 00:23:23.440 Joseph: Okay, nice.
179 00:23:24.100 ⇒ 00:23:24.450 Joseph: Yeah, I…
180 00:23:24.450 ⇒ 00:23:39.830 Robert Tseng: Joseph, to that point, it’s like, yeah, we’ve done enough of this that we’re opinionated about how you should handle attribution at different stages, but ultimately, like, every organization ends up doing attribution differently, and.
181 00:23:39.830 ⇒ 00:23:40.159 Joseph: I know.
182 00:23:40.160 ⇒ 00:23:47.070 Robert Tseng: us, like, kind of pointing to an out-of-the-box solution has actually not worked out well, because people are like, well, I don’t really want to do it that way.
183 00:23:47.370 ⇒ 00:23:48.750 Joseph: I wanna do it this way.
184 00:23:48.750 ⇒ 00:23:51.700 Robert Tseng: And I think we’ve had better…
185 00:23:51.850 ⇒ 00:23:59.160 Robert Tseng: results when we’re more willing to just be like, okay, we can take what you want and build it, build it.
186 00:23:59.160 ⇒ 00:24:04.650 Joseph: And you can always just offer your own attribution model, like, on the side, it’s like, hey, you can compare it to this, see if it helps or not.
187 00:24:04.650 ⇒ 00:24:05.140 Robert Tseng: Yeah.
188 00:24:05.140 ⇒ 00:24:07.899 Joseph: Whatever, if it doesn’t, cool.
189 00:24:07.900 ⇒ 00:24:08.440 Robert Tseng: Yeah.
190 00:24:08.720 ⇒ 00:24:15.369 Joseph: Yeah, cause… Yeah, I don’t think most people are thinking about it correctly. I don’t have the answer, but…
191 00:24:15.470 ⇒ 00:24:16.900 Joseph: I think, just…
192 00:24:17.990 ⇒ 00:24:30.909 Joseph: what I’ve noticed is it’s definitely more of an ecosystem than an isolated event, so it’s like, you just look at last click, but it’s like, okay, cool, then you can scale that last click ad, but then a lot of times you hit a, you know, a diminishing return point.
193 00:24:30.910 ⇒ 00:24:31.440 Robert Tseng: Yep.
194 00:24:31.440 ⇒ 00:24:49.319 Joseph: Or you can’t really scale that, but what if you scaled other parts instead of that one, and does that actually lift ROAS? You know, at the end of the day, and like, so, I don’t think people are understanding how things influence each other very well. I don’t know the answer, but what you guys do would probably give you guys good insights on
195 00:24:49.580 ⇒ 00:24:56.539 Joseph: On what you should actually be moving in order to actually scale, but that’s… Kind of a side tangent.
196 00:24:57.010 ⇒ 00:25:06.850 Robert Tseng: Yeah, no, I think that’s… that’s the right intuition. We’ve optimized for one part, and then once it kind of hits that diminishing return, we look for other ways to kind of continue to look for incremental lift.
197 00:25:07.020 ⇒ 00:25:13.359 Joseph: Yeah, cool. Yeah, those are the biggest questions I have for now. Any… anything you wanna…
198 00:25:13.790 ⇒ 00:25:15.760 Joseph: Dive into, specifically?
199 00:25:16.250 ⇒ 00:25:22.739 Luke Scorziell: Yeah, I guess, like, on our end, maybe it’d be interesting to know, like, are you working with a team of people, too, that are going, like.
200 00:25:23.610 ⇒ 00:25:26.070 Luke Scorziell: technical side with your stack, or what…
201 00:25:26.070 ⇒ 00:25:40.179 Joseph: But right now, I have a… just, like, freelancers that are working on implementing server-side tracking, things like that, just pushing it to, BigQuery, so, like, a data warehouse, and then, you know, pushing back the conversion data, like.
202 00:25:40.400 ⇒ 00:25:47.520 Joseph: to Meta, TikTok, Google, all that stuff. There is GA4, there is Stape that they’re using,
203 00:25:47.700 ⇒ 00:25:49.069 Joseph: I think those are the…
204 00:25:49.430 ⇒ 00:25:56.049 Joseph: main things at the moment, and then maybe, like, one ETL. I think we’re using…
205 00:25:56.950 ⇒ 00:26:01.229 Joseph: I forgot what they’re called specifically, but just a normal ETL to…
206 00:26:01.390 ⇒ 00:26:06.770 Joseph: to help connect… act as a connector to all the data points, so I can see, like, how much ad spend and…
207 00:26:06.890 ⇒ 00:26:10.660 Joseph: return and all that stuff. So…
208 00:26:10.940 ⇒ 00:26:15.709 Joseph: You know, not at a request level, but pretty much right after that is where
209 00:26:16.100 ⇒ 00:26:18.550 Joseph: Which is why we’re still here implementing things.
210 00:26:19.670 ⇒ 00:26:24.800 Luke Scorziell: Yeah, got it. And so with the ETL, is that kind of pulling everything just into one?
211 00:26:24.800 ⇒ 00:26:30.420 Joseph: Once, okay. Yeah, exactly. It’s all connecting into BigQuery, is where… you know.
212 00:26:31.060 ⇒ 00:26:33.250 Joseph: Where we’re housing all the data.
213 00:26:34.340 ⇒ 00:26:35.650 Luke Scorziell: Got it.
214 00:26:36.410 ⇒ 00:26:49.810 Joseph: And we’re probably using a, some sort of dashboard software to be able to make sense of it and make it so that it’s easy for me to manipulate the data to be able to create my own dashboards of whatever I want, versus needing to learn, like, you know, SQL or whatever.
215 00:26:50.210 ⇒ 00:26:56.090 Joseph: That’s, like, the main thing is, like, making it easy to use, because most people make dashboards that just can’t configure it properly and just don’t use it.
216 00:26:56.800 ⇒ 00:26:59.960 Luke Scorziell: Yeah, and then I know, Robert mentioned that you’re…
217 00:27:00.220 ⇒ 00:27:01.100 Joseph: Around, like.
218 00:27:01.100 ⇒ 00:27:04.150 Luke Scorziell: $200,000 a month or so? Is that an ad spend?
219 00:27:04.500 ⇒ 00:27:07.880 Joseph: In ad spend… no, that’s in revenue. Ad spend’s probably, like.
220 00:27:08.100 ⇒ 00:27:14.159 Joseph: I mean, if you count Amazon, it’s different, but, right now, between Meta and TikTok, it’s about 60…
221 00:27:14.360 ⇒ 00:27:15.650 Joseph: $70K a month.
222 00:27:16.430 ⇒ 00:27:17.100 Luke Scorziell: Okay.
223 00:27:17.290 ⇒ 00:27:20.929 Luke Scorziell: And most of that’s… Meta TikTok and Amazon.
224 00:27:21.310 ⇒ 00:27:24.129 Joseph: Just Meta and TikTok is the 60 to 70K.
225 00:27:24.640 ⇒ 00:27:25.240 Luke Scorziell: Okay.
226 00:27:29.640 ⇒ 00:27:34.430 Luke Scorziell: And have you noticed attribution issues so far on, in terms of the numbers lining up with what.
227 00:27:34.430 ⇒ 00:27:36.320 Joseph: For sure, I mean, I’m using, like, I just…
228 00:27:36.320 ⇒ 00:27:36.740 Luke Scorziell: Shopify book.
229 00:27:36.750 ⇒ 00:27:41.250 Joseph: I’ve been using Triplewell for a year, it’s decent, it’s better than…
230 00:27:41.490 ⇒ 00:27:49.470 Joseph: Anything else right now, I just also installed Hyros, so I’m split testing between those two to see the accuracy.
231 00:27:50.390 ⇒ 00:27:55.260 Joseph: it seems like Hydos is a little bit more robust in terms of, like, tracking, you know, individual,
232 00:27:55.380 ⇒ 00:27:58.420 Joseph: Like, every single individual action on the website, and…
233 00:27:58.690 ⇒ 00:28:02.139 Joseph: Making sure that there is some sort of stitching that’s happening,
234 00:28:02.700 ⇒ 00:28:11.680 Joseph: But I still have to test it more rigorously to tell if it is exactly what I’m looking for. The main reason why I’m pursuing all this stuff is, one, yes.
235 00:28:12.080 ⇒ 00:28:21.179 Joseph: accurate, data is one, and then attribution is a different question. And then third is more so, like, calculating, contribution margin
236 00:28:21.590 ⇒ 00:28:37.659 Joseph: per channel, and on an individual ad basis, like, an actual ad basis, which I don’t think any platform does right now, unless you’re getting into, like, enterprise stuff, and you have to build all your own stuff, but I’m talking about, like, out-of-the-box solution. I have not seen, like, a…
237 00:28:38.120 ⇒ 00:28:44.650 Joseph: Like, a fully accurate contribution margin on an ad level, Basis.
238 00:28:46.770 ⇒ 00:28:52.079 Luke Scorziell: Yeah, I got it. And, Zoran, feel free to hop in on that, too, but, I mean, those…
239 00:28:52.240 ⇒ 00:28:53.520 Luke Scorziell: I think the…
240 00:28:53.920 ⇒ 00:28:58.489 Luke Scorziell: Are those the tools that you’ve been experimenting with in your A-B testing, those are all kind of on the client side?
241 00:28:58.640 ⇒ 00:28:59.750 Joseph: that you’re…
242 00:29:00.010 ⇒ 00:29:05.539 Luke Scorziell: Yeah, so I think, I mean, I think the difference, too, with this is that… and I, like, we’ve already talked about, just that it’s…
243 00:29:05.730 ⇒ 00:29:11.279 Luke Scorziell: like… you know, and yeah, Zoran, feel free to hop in, too, on that, but, like, if…
244 00:29:12.050 ⇒ 00:29:19.720 Luke Scorziell: I think that the accuracy on the client side is only ever going to be so high, because we’re kind of going through and able to track before,
245 00:29:20.280 ⇒ 00:29:23.339 Luke Scorziell: Like, an ad blocker or whatnot kind of hits.
246 00:29:23.520 ⇒ 00:29:25.739 Luke Scorziell: And so, yeah, at least…
247 00:29:26.740 ⇒ 00:29:33.540 Luke Scorziell: I would imagine that then the accuracy that you’re gonna get, even between two client-side tools, is still gonna be a lot lower.
248 00:29:34.420 ⇒ 00:29:40.820 Zoran Selinger: Yeah, I mean, it obviously starts with good data, if you don’t have good data.
249 00:29:41.690 ⇒ 00:29:50.290 Zoran Selinger: there’s no tool that… I mean, they can do smart algorithmic stuff, and they can do, they can do really smart estimates, and…
250 00:29:50.420 ⇒ 00:30:00.260 Zoran Selinger: and all those things, but it’s never gonna be the truth. However, I wanted to ask, since you’re talking to us about this, have you calculated,
251 00:30:00.690 ⇒ 00:30:11.310 Zoran Selinger: Have you, kind of checked how many transactions you, for example, don’t see in the triple whale, as opposed to what you see in… are you on Shopify?
252 00:30:11.600 ⇒ 00:30:13.450 Joseph: Yeah, sure. Shopify, right, yeah.
253 00:30:14.510 ⇒ 00:30:15.750 Zoran Selinger: populated those.
254 00:30:17.360 ⇒ 00:30:22.719 Joseph: I mean, on a… it’s difficult to calculate on an ad platform level, because…
255 00:30:23.460 ⇒ 00:30:36.370 Joseph: there is no, really, source of truth with that. Like, I can see what the ad platform says, I can see what Triple Whale says, I can see what Shopify says, but they all attribute it differently.
256 00:30:36.800 ⇒ 00:30:39.799 Zoran Selinger: Oh, yeah, so I’m talking even more basic than.
257 00:30:39.800 ⇒ 00:30:41.629 Joseph: Like, there’s, like, total volume, you’re talking about?
258 00:30:41.630 ⇒ 00:30:48.529 Zoran Selinger: total volume of what Triple Whale sees, for example, and what Shopify knows that is the truth, right?
259 00:30:48.780 ⇒ 00:30:51.580 Joseph: Yeah, I don’t… I think I’ve…
260 00:30:52.010 ⇒ 00:30:54.949 Joseph: Yeah, based on, based on what I see, yeah, like, Triple W can do that.
261 00:30:58.150 ⇒ 00:31:01.419 Zoran Selinger: What you… so I’m, I’m asking about,
262 00:31:01.580 ⇒ 00:31:09.329 Zoran Selinger: The discrepancy between those. So, are you saying, Triple Yel sees absolutely every order that you… Yes. …that you see on the website?
263 00:31:09.330 ⇒ 00:31:11.780 Joseph: Yes. That’s, like, a very simple thing, yeah.
264 00:31:14.350 ⇒ 00:31:22.219 Luke Scorziell: And then, I guess, Zoran, are you trying to see if there’s a discrepancy between what AAAL is seeing, and then maybe what kind of attribution you’re getting through other platforms?
265 00:31:22.760 ⇒ 00:31:26.290 Zoran Selinger: No, no, like I said, even more basic than that.
266 00:31:26.780 ⇒ 00:31:34.600 Zoran Selinger: Just what, just what… how many transactions Triple Whale sees on the client side. Yeah.
267 00:31:34.600 ⇒ 00:31:35.669 Joseph: I see everything, it’s good.
268 00:31:35.670 ⇒ 00:31:36.360 Zoran Selinger: And what they…
269 00:31:36.360 ⇒ 00:31:37.460 Joseph: your new safety zone.
270 00:31:37.460 ⇒ 00:31:39.220 Zoran Selinger: actually happens in Shopify.
271 00:31:39.360 ⇒ 00:31:42.149 Joseph: Yeah, yeah, it’s mirrored, so, I mean, it’s like…
272 00:31:42.520 ⇒ 00:31:43.699 Joseph: It’s pretty hard to be wrong there.
273 00:31:43.700 ⇒ 00:32:01.290 Zoran Selinger: Oh, yeah, yeah, Triple Whale is, is… yeah, so Triple Whale is also… it links to your, your Shopify, so it’s not… yeah, it’s not client data directly, they are actually sharing the data via API, so it’s… what you see, yeah, in TripleWell, is…
274 00:32:01.290 ⇒ 00:32:14.610 Zoran Selinger: Yeah, basically imported, imported data. So, it would be interested, interesting to see another platform that is just, like, just client-side, even though if implemented with a, with, like, a…
275 00:32:14.710 ⇒ 00:32:21.999 Zoran Selinger: let’s say, server-side GTM, and that would be really interesting to see as well, if you have that.
276 00:32:22.260 ⇒ 00:32:34.759 Zoran Selinger: Yeah, but yeah, Triplewell is not a good example here, because they do… they do just import a lot of the data, and they try to do smart stuff with AI, with models.
277 00:32:41.820 ⇒ 00:32:43.690 Luke Scorziell: Yeah, and then in terms of, like, business…
278 00:32:44.360 ⇒ 00:32:48.359 Luke Scorziell: I’m gonna be zooming out a little bit, so that’s 200,
279 00:32:48.680 ⇒ 00:32:51.119 Luke Scorziell: Sorry, was that a year or a month in revenue?
280 00:32:51.120 ⇒ 00:33:01.799 Joseph: I mean, it just swings quite a bit, because the platforms, but yeah, anywhere between, I think, like, 130 to 200, I mean…
281 00:33:02.310 ⇒ 00:33:03.860 Joseph: Depending on the season.
282 00:33:04.420 ⇒ 00:33:05.830 Luke Scorziell: Yeah,
283 00:33:07.550 ⇒ 00:33:14.029 Luke Scorziell: And… are you doing… what’s kind of the goal for you? Are you… it looked like you had
284 00:33:14.320 ⇒ 00:33:17.990 Luke Scorziell: Maybe, like, under a dozen books? How many books are you selling right now?
285 00:33:19.160 ⇒ 00:33:24.189 Joseph: It’s probably around… 8 to 10, or something like that.
286 00:33:24.940 ⇒ 00:33:28.310 Luke Scorziell: Yeah. And do you have more that you’re kind of planning on launching?
287 00:33:29.350 ⇒ 00:33:33.330 Joseph: Not soon, but… Other… other products, yeah.
288 00:33:33.700 ⇒ 00:33:35.870 Luke Scorziell: Okay, like courses or.
289 00:33:35.870 ⇒ 00:33:39.850 Joseph: No, I mean, it’s an adjacent business, it’s an app, but…
290 00:33:40.800 ⇒ 00:33:43.909 Joseph: most of the stuff will funnel there, but it’s not gonna be tracked on Shopify.
291 00:33:46.690 ⇒ 00:33:53.049 Luke Scorziell: And is the hope just to, like, do you have growth, targets that you’re trying to reach with it, like, this year or next year?
292 00:33:53.440 ⇒ 00:33:56.100 Joseph: I mean, arbitrarily, I mean, it’s…
293 00:33:56.390 ⇒ 00:33:59.490 Luke Scorziell: Yeah. It doesn’t really matter to me that much.
294 00:33:59.590 ⇒ 00:34:01.820 Joseph: It’s more so play for me.
295 00:34:01.980 ⇒ 00:34:04.869 Joseph: But, you know, if you…
296 00:34:05.090 ⇒ 00:34:09.050 Joseph: If I had to put a number just to organize thought, then it’d be…
297 00:34:09.570 ⇒ 00:34:14.429 Joseph: you know, scaling from, I think right now, we’re about… 2 million.
298 00:34:14.600 ⇒ 00:34:15.610 Joseph: A year?
299 00:34:16.139 ⇒ 00:34:17.369 Joseph: To about 10.
300 00:34:20.460 ⇒ 00:34:23.559 Joseph: At about a 20% contribution margin, specifically.
301 00:34:24.290 ⇒ 00:34:30.769 Joseph: For… on, like, an ad level, so… That’s what we’re going for.
302 00:34:31.920 ⇒ 00:34:34.869 Luke Scorziell: Yeah, and do you feel like your current stack has you set up for that, or…
303 00:34:35.500 ⇒ 00:34:36.699 Joseph: No, that’s why I’m building it.
304 00:34:36.940 ⇒ 00:34:37.830 Luke Scorziell: Yeah, yeah.
305 00:34:47.770 ⇒ 00:34:49.749 Luke Scorziell: Yeah, I think… Robert, if you want to…
306 00:34:49.750 ⇒ 00:35:00.210 Robert Tseng: Oh yeah, I was just gonna say, I got a drop, so, it’s good to meet you, Joseph. Let me know if you have any other questions. We can send over, kind of, this deck to you, and,
307 00:35:01.100 ⇒ 00:35:09.359 Robert Tseng: Yeah, I mean, I guess I don’t… it doesn’t seem like we have, like, a path forward to work together right now, but, I mean, we can help put together a scope of work if you’re interested in…
308 00:35:09.360 ⇒ 00:35:10.340 Joseph: That’d be great.
309 00:35:10.340 ⇒ 00:35:11.370 Robert Tseng: to… yeah.
310 00:35:11.370 ⇒ 00:35:13.399 Joseph: What do you guys do, how,
311 00:35:13.630 ⇒ 00:35:16.949 Joseph: Yeah, pricing for, you know, this level.
312 00:35:16.950 ⇒ 00:35:17.710 Robert Tseng: Sure.
313 00:35:17.930 ⇒ 00:35:22.609 Joseph: Whatever else you have as an explanation of, like, the… how the, you know.
314 00:35:23.150 ⇒ 00:35:25.069 Joseph: on a request level works, like, that would be…
315 00:35:25.070 ⇒ 00:35:25.570 Robert Tseng: Yeah.
316 00:35:25.570 ⇒ 00:35:28.480 Joseph: Super helpful, just because it’s still super high level, but…
317 00:35:28.690 ⇒ 00:35:44.610 Robert Tseng: Yeah, I think what ends up, kind of, for us is, like, you know, if you feel like server-side tracking, I mean, you just kind of… it basically ends up being, like, an ROI kind of decision for you. It’s like, okay, does, like, you know, 15%… 15% to 20% lift in… in tracking, you know.
318 00:35:45.380 ⇒ 00:35:48.530 Robert Tseng: You know how much… you should know, like.
319 00:35:48.770 ⇒ 00:36:05.300 Robert Tseng: how much… how many more sales that’s gonna drive, and if that’s… if that’s kind of worth it for you right now, then, like, maybe, like, our… our service would kind of… would work for you. That’s why I was, like, kind of saying over message, like, it only really makes sense when you’re spending enough.
320 00:36:05.370 ⇒ 00:36:25.169 Robert Tseng: Because, yeah, I mean, it’s definitely not cheap for us to do this, but, I mean, for bigger brands, like, that are, you know, spending millions a month, it totally makes sense, because even just, like, a 5% lift, is, like, kind of game-changing for the business. I mean, your size is, like, kind of… I mean, I think there’s still a case for it, and I don’t want to diminish that, so…
321 00:36:25.210 ⇒ 00:36:32.130 Robert Tseng: Yeah, I think, like, maybe there’s still, opportunity for us to kind of work with you, on this.
322 00:36:32.940 ⇒ 00:36:37.880 Joseph: Yeah, for sure. Yeah, just send over whatever you’re having, and… I’ll definitely run some numbers.
323 00:36:38.400 ⇒ 00:36:40.709 Robert Tseng: Sure. All right. Thanks, Joseph. Thanks, Adam.
324 00:36:40.710 ⇒ 00:36:41.950 Joseph: Awesome, thanks so much, guys.
325 00:36:41.950 ⇒ 00:36:43.100 Zoran Selinger: Let me jump by that?