Meeting Title: Eden Daily Standup Date: 2026-02-25 Meeting participants: Greg Stoutenburg, Robert Tseng, Zoran Selinger, Demilade Agboola, Amber Lin, Awaish Kumar
WEBVTT
1 00:00:07.490 ⇒ 00:00:08.580 Robert Tseng: Hey, Greg.
2 00:00:08.580 ⇒ 00:00:10.010 Greg Stoutenburg: Hey, how’s it going?
3 00:00:10.710 ⇒ 00:00:15.710 Robert Tseng: Good! Sorry, we got… we’re a little bit delayed on the other stand-up.
4 00:00:15.970 ⇒ 00:00:16.630 Greg Stoutenburg: Sure.
5 00:00:18.420 ⇒ 00:00:24.279 Robert Tseng: Cool. I mean, I guess, do you want to kind of go first, on kind of where you’re at with things?
6 00:00:24.620 ⇒ 00:00:43.409 Greg Stoutenburg: Yep, so I… I just posted an update about the migration, basically today’s QA, and get snapshot delivery set up. Everything that was in Tableau has been moved over, at least provisionally, and is either, looking good, and I just want to make sure that, you know, every cell matches every cell.
7 00:00:43.410 ⇒ 00:00:49.690 Greg Stoutenburg: And or, you know, or there’s some issue that’s been identified that’s already been tagged for follow-up, so…
8 00:00:50.040 ⇒ 00:01:01.810 Greg Stoutenburg: Feeling pretty good about that progress. I saw that Jonah asked a question about where to get data. Tableau is still on right now.
9 00:01:01.810 ⇒ 00:01:13.849 Greg Stoutenburg: And I’d rather not turn ELT over to Omni until I feel like it’s just, like, beautiful. So, I don’t know if anyone on ELT responded to the invitation to have office hours.
10 00:01:14.510 ⇒ 00:01:16.300 Greg Stoutenburg: Did they say anything to you?
11 00:01:17.040 ⇒ 00:01:21.780 Robert Tseng: No, but I’m talking to them in, like, an hour, or two hours, so I’ll let them know.
12 00:01:22.090 ⇒ 00:01:39.809 Greg Stoutenburg: Okay. And by then, I might be able to say, yeah, Omni’s got a bow on it, so, you know, dig in. So, yeah, that’s where that’s at. So I think… I think for now, maybe I just respond to Jonah’s message and say, you know, hey, putting the finishing touches on Omni, have a look at Tableau… for now.
13 00:01:39.810 ⇒ 00:01:43.380 Robert Tseng: There was already… how would the first training go? Like, kind of… yeah.
14 00:01:43.570 ⇒ 00:02:00.290 Greg Stoutenburg: Yeah, it was fine. I mean, it was… it was yesterday with, Judd, Ryan, and Matt, and, I mean, they took to it quickly. They found it pretty easy to use, so those are all good things. Blobby gave a wrong answer to Ryan’s first question, which was a little disheartening.
15 00:02:00.290 ⇒ 00:02:10.789 Greg Stoutenburg: Yeah, it, like, ignored a filter. Like, he said, you know, between this date and this date, you know, question. And it ignored the filter and gave the wrong answer. So,
16 00:02:10.850 ⇒ 00:02:22.100 Greg Stoutenburg: that’s not good, and so I need to get to the bottom of that one. I hope it’s just… I hope it’s just Blobby made a mistake, but, you know, we don’t want to rely on the stakeholder already knowing what the correct answer is.
17 00:02:22.320 ⇒ 00:02:26.229 Greg Stoutenburg: When we say, yeah, go for it, this is good for self-service.
18 00:02:26.540 ⇒ 00:02:27.470 Greg Stoutenburg: Yep.
19 00:02:28.470 ⇒ 00:02:30.079 Greg Stoutenburg: But it was good, I mean, you know, it’s…
20 00:02:30.230 ⇒ 00:02:43.059 Greg Stoutenburg: once they… they very quickly took to the revised interface and get where you can ask AI questions, and that’s… that’s kind of the whole thing. I mean, that’s… that’s what we wanted them to see. Like, this is Tableau, but you can ask questions using AI and get your answers faster.
21 00:02:43.450 ⇒ 00:02:50.370 Robert Tseng: Can you add, like, a link or two or something? Like, maybe it’s your, like, a screenshot from your demo or something? Like, I think this… obviously, this is just the chat interface, but…
22 00:02:50.370 ⇒ 00:02:50.750 Greg Stoutenburg: Yeah.
23 00:02:50.750 ⇒ 00:03:03.209 Robert Tseng: I mean, I’m gonna put this… already shared this deck out with ELT, but, like, they’re gonna… like, they look at my slides, so, like, I would like at least to just show them… just keep showing them Omni from every different angle, and…
24 00:03:03.210 ⇒ 00:03:03.770 Greg Stoutenburg: Doing stuff.
25 00:03:03.770 ⇒ 00:03:04.390 Robert Tseng: Yeah.
26 00:03:04.630 ⇒ 00:03:13.209 Greg Stoutenburg: Sure, yeah, I’ll just pull the Zoom clip, and yeah, well, okay, it’s in here now. After this, after this call, I’ll pull a Zoom clip and put a screenshot in there.
27 00:03:13.410 ⇒ 00:03:15.070 Robert Tseng: Okay, great. Thanks.
28 00:03:16.010 ⇒ 00:03:23.110 Robert Tseng: Yeah, and then… I guess anything else on the Omni thing? Otherwise, I’ll kind of move on. I’ll jump to a couple others.
29 00:03:23.110 ⇒ 00:03:25.260 Greg Stoutenburg: That’s it. Just, yep.
30 00:03:26.380 ⇒ 00:03:31.939 Robert Tseng: Yeah, heard you on this, Zoran, so I think all good here, we can keep it as is.
31 00:03:32.070 ⇒ 00:03:37.399 Robert Tseng: I’m not entirely sure, like, what your conversations with Mitesh are. I did look at, like.
32 00:03:38.590 ⇒ 00:03:43.289 Robert Tseng: Yeah. Like, I don’t know how Mitesh is using this, frankly, and I also… is this?
33 00:03:43.650 ⇒ 00:03:48.500 Robert Tseng: I was looking through what Ryan and Mitesh shared today, which is.
34 00:03:48.940 ⇒ 00:03:51.469 Robert Tseng: I think this is the right idea, is what we’re thinking.
35 00:03:51.970 ⇒ 00:03:54.599 Robert Tseng: Different model by funnel, yeah,
36 00:03:54.950 ⇒ 00:04:13.820 Zoran Selinger: Yeah, we were in a call, about this, earlier. We were looking at trying to figure out what exactly means clicks plus deterministic views in NordBeam, and, because numbers are vastly different when we, when we started using deterministic or,
37 00:04:13.820 ⇒ 00:04:15.940 Zoran Selinger: Or modeled views, right?
38 00:04:15.940 ⇒ 00:04:20.489 Zoran Selinger: Yeah. And that conversation is ongoing at the moment.
39 00:04:21.660 ⇒ 00:04:29.440 Zoran Selinger: So this is essentially… Mintage doesn’t care what they use internally, and what they, what they,
40 00:04:30.350 ⇒ 00:04:38.910 Zoran Selinger: What they, basically, kind of optimized towards, we are going to use clicks only in…
41 00:04:39.260 ⇒ 00:04:42.840 Zoran Selinger: in Norbeam to evaluate everyone.
42 00:04:43.050 ⇒ 00:04:48.319 Zoran Selinger: So that column F, that’s the… that’s the main… that’s the main point.
43 00:04:48.480 ⇒ 00:04:57.980 Zoran Selinger: And… and the column D is what we would propose, sorry, not the column D, but the column, C.
44 00:04:58.500 ⇒ 00:05:04.380 Zoran Selinger: Is what we would kind of propose to them to use. I’m…
45 00:05:05.140 ⇒ 00:05:17.320 Zoran Selinger: And that’s basically it. There is one outstanding conversation that we did not get to earlier today, and that is the models that we use to propagate
46 00:05:17.690 ⇒ 00:05:18.940 Zoran Selinger: Purchase events.
47 00:05:19.270 ⇒ 00:05:19.890 Robert Tseng: Yeah.
48 00:05:20.380 ⇒ 00:05:39.420 Zoran Selinger: So this is still an outstanding conversation that we… we still don’t have an answer of how we would approach this exactly. It’s not perfectly clear yet. We haven’t had a chance to catch up on that in the meeting, but we acknowledge that this is a conversation we need to have.
49 00:05:41.040 ⇒ 00:05:44.470 Robert Tseng: I’m in view only, I couldn’t leave a comment here. Alright, whatever.
50 00:05:44.740 ⇒ 00:05:48.069 Robert Tseng: Okay, sure. I’m assuming you’ll have that conversation then.
51 00:05:48.070 ⇒ 00:05:48.770 Zoran Selinger: Yeah, yeah.
52 00:05:48.770 ⇒ 00:05:49.460 Robert Tseng: I’m just gonna make a comment.
53 00:05:49.460 ⇒ 00:05:57.270 Zoran Selinger: We need to, we need to, because initially, initially, Mitesh said, kind of, we don’t, we don’t care what they use.
54 00:05:57.510 ⇒ 00:06:06.139 Zoran Selinger: for… to optimize, and I said, but we… we do make an impact on how they… what they use to optimize, because we are sending the purchase event.
55 00:06:06.290 ⇒ 00:06:11.559 Zoran Selinger: And we are using an attribution model that we decided on. Yeah.
56 00:06:11.790 ⇒ 00:06:15.330 Zoran Selinger: And how to send it. And you saw the message from Ghost?
57 00:06:16.320 ⇒ 00:06:18.880 Robert Tseng: I have not yet, but yeah.
58 00:06:18.880 ⇒ 00:06:31.420 Zoran Selinger: it seems super low to them, it’s just weird. I know it’s weird. I agree with them that it’s weird, that kind of performance is super low, and I don’t know what to do about it, because…
59 00:06:33.510 ⇒ 00:06:38.329 Zoran Selinger: He just… When we used other models, attribution models, it just…
60 00:06:38.520 ⇒ 00:06:43.510 Zoran Selinger: It looks super, super unnaturally low. I don’t, I don’t know.
61 00:06:44.080 ⇒ 00:06:48.709 Zoran Selinger: I don’t know. I would be skeptical too, but we checked it.
62 00:06:48.840 ⇒ 00:06:58.820 Zoran Selinger: Actually, I don’t know what the latest model, is exactly. What did you guys implement on Friday? I was just about to go into the dbt model and try to figure it out.
63 00:06:59.740 ⇒ 00:07:04.090 Robert Tseng: Yeah, okay, well, I laid out…
64 00:07:04.090 ⇒ 00:07:08.799 Zoran Selinger: I mean, if you don’t remember, if you don’t have it on top of your head, fine, I’ll have a look.
65 00:07:09.160 ⇒ 00:07:11.720 Awaish Kumar: Yeah, we can… I can tell.
66 00:07:12.130 ⇒ 00:07:13.890 Zoran Selinger: Okay, sure, go ahead.
67 00:07:14.120 ⇒ 00:07:20.160 Awaish Kumar: Yeah, what we did was mainly, we start… I use the same identity stitching that you use.
68 00:07:20.350 ⇒ 00:07:24.610 Awaish Kumar: on, like, user ID, anonymous ID, or whatever, like, IDs there are.
69 00:07:24.610 ⇒ 00:07:25.240 Zoran Selinger: Yup.
70 00:07:25.240 ⇒ 00:07:32.730 Awaish Kumar: Yeah. And then, are you… we have this 30 days period where, like, is there… if there is any session.
71 00:07:32.880 ⇒ 00:07:38.219 Awaish Kumar: in, thank you page or in Edge Layer,
72 00:07:38.440 ⇒ 00:07:55.719 Awaish Kumar: in the last 30 days, then we just selected, selected order. But I think what made it more restrictive was because now we are not relying on UTM medium and source. We removed that filter. We don’t check for Facebook or anything.
73 00:07:55.990 ⇒ 00:07:57.039 Robert Tseng: Yeah, it should be any.
74 00:07:57.040 ⇒ 00:07:58.890 Awaish Kumar: SPCL ID.
75 00:07:59.180 ⇒ 00:08:13.230 Awaish Kumar: So, we are using FBCL ID, which basically, if that is null, then we don’t consider it, right? We are saying it should be not null, and maybe that is something that made it more restrictive.
76 00:08:14.380 ⇒ 00:08:19.570 Robert Tseng: Yeah, so we removed the UTM filter, so that should have made it broader, but then we’re still using the FBE QuickID.
77 00:08:19.570 ⇒ 00:08:20.230 Zoran Selinger: It’s like, give it.
78 00:08:20.230 ⇒ 00:08:21.480 Robert Tseng: Still… yeah.
79 00:08:21.780 ⇒ 00:08:23.120 Robert Tseng: If it’s artificially depressed.
80 00:08:23.120 ⇒ 00:08:24.819 Zoran Selinger: It’s because of any touch here.
81 00:08:25.070 ⇒ 00:08:28.470 Zoran Selinger: I think the biggest impact is any touch conversion.
82 00:08:29.540 ⇒ 00:08:36.119 Zoran Selinger: Here, because previously, we were on… first, we were on the first touch, then we were on the last touch.
83 00:08:36.610 ⇒ 00:08:42.150 Zoran Selinger: And now they’re any touch. That’s the biggest impact, I think, from the change here.
84 00:08:45.090 ⇒ 00:08:45.600 Awaish Kumar: Hey, yeah.
85 00:08:45.610 ⇒ 00:08:47.360 Zoran Selinger: I think this… let’s see, the curr…
86 00:08:47.360 ⇒ 00:08:48.040 Awaish Kumar: in Australia.
87 00:08:48.040 ⇒ 00:08:49.599 Zoran Selinger: Instead of undercurrent.
88 00:08:50.440 ⇒ 00:08:53.780 Zoran Selinger: Thank you, Paige And Magic Session Image.
89 00:08:54.870 ⇒ 00:08:56.390 Robert Tseng: That’s what we had before, right? So…
90 00:08:56.390 ⇒ 00:09:02.220 Zoran Selinger: No, no. Sorry, that was… that is wrong. We were much more restricted than this.
91 00:09:02.390 ⇒ 00:09:05.920 Zoran Selinger: What you said in current is not the case.
92 00:09:07.100 ⇒ 00:09:13.700 Zoran Selinger: we… the model that was active was… first, I implemented the first touch.
93 00:09:14.080 ⇒ 00:09:16.159 Zoran Selinger: So it had to be first touch.
94 00:09:17.730 ⇒ 00:09:27.440 Zoran Selinger: So, essentially, the order… The order summary table has, a first UTM
95 00:09:27.870 ⇒ 00:09:30.379 Zoran Selinger: field. I was looking into that.
96 00:09:30.500 ⇒ 00:09:35.600 Zoran Selinger: And if it’s Meta, they get the conversion, okay?
97 00:09:35.600 ⇒ 00:09:47.320 Robert Tseng: I think we need to dig into the segment model, because when I pulled it with the wish, and we looked at what was live, I looked at that model, and that’s just what it was. We should set up more time to discuss this.
98 00:09:48.030 ⇒ 00:09:57.229 Zoran Selinger: Because, by the way, I just logged in earlier today, and I saw both Meta First Time Last Touch was…
99 00:09:58.310 ⇒ 00:10:02.880 Robert Tseng: Yeah, they’re both on, because, like, I needed to, like, wait 24 hours. I mean, I never turned it off on Monday.
100 00:10:02.880 ⇒ 00:10:08.310 Zoran Selinger: Okay, okay. And, now metacopy broad conversions is…
101 00:10:08.310 ⇒ 00:10:09.390 Robert Tseng: Yeah, they’re both on, yeah.
102 00:10:09.390 ⇒ 00:10:10.800 Zoran Selinger: And a lot, and I…
103 00:10:11.010 ⇒ 00:10:17.139 Zoran Selinger: I compared the numbers, and yeah, it’s 3 to 4 times more with your broad model.
104 00:10:17.140 ⇒ 00:10:18.749 Robert Tseng: Yeah, I mean, I did say that, and I was like…
105 00:10:18.750 ⇒ 00:10:20.640 Zoran Selinger: Look, it’s 2.5 times more.
106 00:10:20.640 ⇒ 00:10:21.500 Robert Tseng: More, yeah.
107 00:10:21.500 ⇒ 00:10:22.030 Zoran Selinger: Yeah.
108 00:10:22.260 ⇒ 00:10:27.810 Zoran Selinger: And it’s still… the numbers are, like… there’s only 45 conversions in that table at the moment.
109 00:10:28.810 ⇒ 00:10:30.230 Zoran Selinger: In the broad model.
110 00:10:30.390 ⇒ 00:10:33.710 Zoran Selinger: And they spent 25K.
111 00:10:34.910 ⇒ 00:10:38.440 Zoran Selinger: So far. So it’s just…
112 00:10:38.660 ⇒ 00:10:47.359 Zoran Selinger: I don’t know, I don’t know what to think, and that’s… we probably almost definitely did not make any mistake in the… in the modeling.
113 00:10:47.540 ⇒ 00:10:49.599 Zoran Selinger: It’s just… I don’t know.
114 00:10:49.600 ⇒ 00:11:05.260 Robert Tseng: I mean, nothing is looking as surprised to me. I literally said, this is what we’re pushing, and it’s gonna have a 4X impact, and it’s what… it’s exactly what it has done. And they’re saying that it’s just, like, way too low. Whatever. Like, we… well, it’s not whatever. We have to have an answer. They’re gonna randomly start calling me in, like, an hour.
115 00:11:05.450 ⇒ 00:11:07.879 Robert Tseng: So, I would like us to get ahead of that.
116 00:11:07.880 ⇒ 00:11:08.230 Zoran Selinger: Okay.
117 00:11:08.230 ⇒ 00:11:22.649 Robert Tseng: But let’s table this. Let’s just go through anything else on Eden that needs to get pushed out. Deck-wise, like, I made some changes. Greg’s deck, I just want to call out. I kind of took it in a different direction. Like, I think we had too much going on in one slide. It’s not like…
118 00:11:22.940 ⇒ 00:11:31.090 Robert Tseng: So… Yeah, I… I mean, I can… I’ll probably record myself talking through it later, but…
119 00:11:31.290 ⇒ 00:11:36.609 Robert Tseng: I mean, I think it would be helpful if you kind of, like, looked through it and tried to, like, compare it to what you had before.
120 00:11:38.160 ⇒ 00:11:40.729 Robert Tseng: Yeah, I mean, in short, like.
121 00:11:41.980 ⇒ 00:11:47.970 Robert Tseng: the… I think, yeah, Mixpanel, VWO, Tableau, plus, like, BigQuery, like, whatever, like, I thought there was just…
122 00:11:48.290 ⇒ 00:12:03.109 Robert Tseng: I think it was more helpful to kind of focus on different, like, domain areas, or, like, topics, or whatever you want to call it. So, like, here’s one for CRO, here’s one for commercial, like, here’s one for supply. So, I think, like, kind of…
123 00:12:03.110 ⇒ 00:12:10.840 Robert Tseng: kind of combining everything into one narrative is, like, too hard to do, like, and ultimately, they have different objectives, like.
124 00:12:10.840 ⇒ 00:12:11.310 Greg Stoutenburg: Yep.
125 00:12:11.310 ⇒ 00:12:29.029 Robert Tseng: like, from the CRO one, like, I mean, I basically took your goals. I think these are the right goals, but, like, I felt like there was just other noise in there when we were trying to weave in Basque and Tableau. So, I… I think that’s… that’s just, like, something that… I made that intentional decision to split into two slides.
126 00:12:29.030 ⇒ 00:12:31.539 Robert Tseng: it’s kind of saying the same thing. It’s like.
127 00:12:31.610 ⇒ 00:12:48.119 Robert Tseng: This is what it was before. We have HealthOS, we’re basically able to, like, enrich it with more data that we have control over. It’s all gonna be, like, through our warehouse, and… but, like, the goal state is going to be different for each of these initiatives. So, yeah, I guess, you know, no need to, like, kind of, like.
128 00:12:48.350 ⇒ 00:12:53.120 Robert Tseng: change anything on the slide at this point, but I just wanted to let you know that’s what I… that’s what I did differently.
129 00:12:53.610 ⇒ 00:12:54.210 Greg Stoutenburg: Tricked.
130 00:12:55.120 ⇒ 00:13:03.969 Robert Tseng: And then I asked the AI team to kind of review this. I don’t think they have, but this is another proposal that, Danny put together. I basically tried to push…
131 00:13:04.130 ⇒ 00:13:20.980 Robert Tseng: I’ll open another, like, AI work stream with them. So, I mean, between this and… yeah, well, it’s just a lot… there’s probably too much to cover today, like, with them, so… but, I… anyway, I already sent this deck to them, so we’ll see how it goes in a couple hours.
132 00:13:25.440 ⇒ 00:13:28.520 Robert Tseng: Cool. Yeah, I guess,
133 00:13:29.170 ⇒ 00:13:33.709 Robert Tseng: Anything else that’s pressing that we should just kind of address now?
134 00:13:34.750 ⇒ 00:13:36.599 Robert Tseng: I can stay on for a couple more minutes.
135 00:13:40.650 ⇒ 00:13:45.939 Robert Tseng: Okay, if not, then maybe Zoran will stay on. We’ll do… let’s just kind of try to get into the…
136 00:13:45.940 ⇒ 00:13:47.929 Zoran Selinger: Let’s figure it out.
137 00:13:48.680 ⇒ 00:13:50.779 Robert Tseng: I’ll skip that, I’ll skip that call.
138 00:13:50.890 ⇒ 00:13:52.300 Robert Tseng: dealt.
139 00:13:54.740 ⇒ 00:13:55.540 Robert Tseng: Okay.
140 00:13:55.770 ⇒ 00:13:56.670 Robert Tseng: So…
141 00:13:58.640 ⇒ 00:14:01.880 Zoran Selinger: First thing I wanna do, I just wanna check…
142 00:14:02.410 ⇒ 00:14:04.169 Zoran Selinger: And I’m gonna do that now.
143 00:14:04.450 ⇒ 00:14:12.749 Zoran Selinger: I wanna check… in our… So, your model…
144 00:14:16.000 ⇒ 00:14:23.680 Zoran Selinger: when we, when we’re looking at the, at the Facebook Click ID, Which table is that in?
145 00:14:27.110 ⇒ 00:14:28.400 Robert Tseng: In, in BigQuery?
146 00:14:31.370 ⇒ 00:14:32.080 Zoran Selinger: Oh, yeah.
147 00:14:32.200 ⇒ 00:14:36.089 Zoran Selinger: Are we looking at the edge, or something else?
148 00:14:39.720 ⇒ 00:14:44.240 Zoran Selinger: or just looking at the… at the thank you page, Facebook link ID.
149 00:14:44.500 ⇒ 00:14:46.770 Robert Tseng: Yeah, let me, let me see. I didn’t know there was a difference.
150 00:14:49.190 ⇒ 00:14:51.219 Zoran Selinger: I mean, there…
151 00:14:51.360 ⇒ 00:15:02.499 Zoran Selinger: it’s there for… in each table, because they are keeping the UTMs going from… from the… essentially, from the session start…
152 00:15:02.700 ⇒ 00:15:08.579 Zoran Selinger: to the thank you page. So they’re keeping the UTMs in most cases. I just want to make sure
153 00:15:10.910 ⇒ 00:15:15.389 Zoran Selinger: That… that’s 100% the case. I wanna… basically, I wanna count.
154 00:15:15.710 ⇒ 00:15:19.589 Zoran Selinger: I wanna count…
155 00:15:29.230 ⇒ 00:15:36.050 Robert Tseng: Okay, so there are the attribution thank you page visits, there’s the edge layer raw data,
156 00:15:36.820 ⇒ 00:15:41.830 Robert Tseng: And you’re saying that… But obviously, Azure Raw Data is going to have.
157 00:15:42.970 ⇒ 00:15:44.340 Zoran Selinger: Those are the sessions, yeah.
158 00:15:44.340 ⇒ 00:15:48.280 Robert Tseng: Yeah, those are… it’s about the session level, so it’s gonna be more.
159 00:15:49.250 ⇒ 00:15:57.829 Zoran Selinger: basically, I just wanna… I just wanna make sure, that basically anytime we have UTMs.
160 00:15:58.790 ⇒ 00:16:00.980 Zoran Selinger: we also have the Facebook link ID.
161 00:16:01.820 ⇒ 00:16:09.960 Zoran Selinger: Because I would… I would credit them, I would credit them with anything that is either-or, right?
162 00:16:10.150 ⇒ 00:16:19.479 Zoran Selinger: either we see the UTMs, or we see the existence of Facebook ClickID. In most cases, all of those should be true.
163 00:16:19.630 ⇒ 00:16:20.520 Zoran Selinger: Ow.
164 00:16:26.540 ⇒ 00:16:31.239 Robert Tseng: I’m not finding the right… SB CAPI model…
165 00:16:33.290 ⇒ 00:16:37.970 Robert Tseng: I forgot when Awash named it in dbt, I called it something different in the segment.
166 00:17:00.960 ⇒ 00:17:04.900 Robert Tseng: Yeah, meta, Cappy, broad conversions, and pay.
167 00:17:04.900 ⇒ 00:17:07.129 Zoran Selinger: Interesting. Am I really…
168 00:17:19.000 ⇒ 00:17:22.310 Robert Tseng: order summary, which comes from…
169 00:17:22.490 ⇒ 00:17:28.680 Robert Tseng: Well, yeah, so order summary, it’s like… it’s… yeah, it’s either that, or it comes from the edge layer data.
170 00:17:29.560 ⇒ 00:17:32.960 Robert Tseng: that would be such preferred.
171 00:17:35.200 ⇒ 00:17:40.569 Zoran Selinger: So I’m hearing… I have, currently, 15.5K
172 00:17:45.290 ⇒ 00:17:51.469 Zoran Selinger: distinct Facebook link IDs in the… in the session table. Edge layer of data.
173 00:17:54.400 ⇒ 00:17:58.250 Zoran Selinger: And if I count everything…
174 00:18:05.450 ⇒ 00:18:07.780 Zoran Selinger: where UTM source.
175 00:18:07.780 ⇒ 00:18:09.510 Robert Tseng: Just for the past 7 days?
176 00:18:10.760 ⇒ 00:18:12.970 Zoran Selinger: No, no, that’s total.
177 00:18:12.970 ⇒ 00:18:15.119 Robert Tseng: I mean, you know what I mean, do you…
178 00:18:15.410 ⇒ 00:18:20.000 Zoran Selinger: Equals, paid social.
179 00:18:28.710 ⇒ 00:18:30.110 Zoran Selinger: Interesting.
180 00:18:31.000 ⇒ 00:18:39.510 Zoran Selinger: So we have 75.5 rose, where…
181 00:18:40.060 ⇒ 00:18:45.400 Zoran Selinger: It’s coming from Facebook Paid Social. Basically, Meta Paid Social.
182 00:18:46.550 ⇒ 00:18:54.989 Zoran Selinger: And we only have 15.5, Okay, distinct Facebook Click IDs.
183 00:19:02.340 ⇒ 00:19:04.720 Zoran Selinger: And let’s go click IDE.
184 00:19:04.930 ⇒ 00:19:06.520 Zoran Selinger: is now.
185 00:19:13.000 ⇒ 00:19:13.800 Zoran Selinger: There you go.
186 00:19:14.220 ⇒ 00:19:25.770 Zoran Selinger: Yeah, so we have 63.7K, sessions with Meta, Paid social.
187 00:19:26.730 ⇒ 00:19:29.600 Zoran Selinger: with no Facebook Click ID.
188 00:19:37.850 ⇒ 00:19:39.060 Zoran Selinger: In the same role.
189 00:19:41.450 ⇒ 00:19:45.450 Zoran Selinger: So, in most cases, we won’t have Facebook link ID.
190 00:19:48.750 ⇒ 00:19:49.400 Robert Tseng: Okay.
191 00:19:53.880 ⇒ 00:20:01.839 Zoran Selinger: So we should rely on the UTMs to determine if the Facebook was a touchpoint.
192 00:20:25.690 ⇒ 00:20:27.080 Zoran Selinger: That’s old time.
193 00:20:38.290 ⇒ 00:20:43.729 Robert Tseng: So, we should not be using FB ClickIDs, and we should only be using UTMs.
194 00:20:45.780 ⇒ 00:20:50.780 Robert Tseng: why were we even considering FBQuickID in the first place? That just… it was just misleading for me.
195 00:20:50.780 ⇒ 00:20:57.639 Zoran Selinger: I’m not sure, I’m not even sure why there’s no Facebook Click ID for almost every visit.
196 00:20:57.990 ⇒ 00:21:01.740 Zoran Selinger: via… via MetaPaid Social.
197 00:21:01.850 ⇒ 00:21:17.869 Zoran Selinger: it is possible that most of their campaigns are not tagged with a template, with a tracking template that will give us Facebook link ID in the landing page URL. It is possible, I’m not sure if that’s the case.
198 00:21:18.020 ⇒ 00:21:21.950 Zoran Selinger: In any case, we should use the UTMs.
199 00:21:22.380 ⇒ 00:21:24.640 Zoran Selinger: To determine the source.
200 00:21:25.210 ⇒ 00:21:32.239 Robert Tseng: This is for all time. I mean, can we… can we show that there’s been, like, a drop in the quality of the QuickID? I mean, they haven’t run…
201 00:21:32.390 ⇒ 00:21:43.149 Robert Tseng: Facebook in a while, so if we could look at before Ghost, and then after Ghost to see if, like, the ratio of FB ClickIDs to Meta, has… has changed.
202 00:21:43.410 ⇒ 00:21:48.909 Zoran Selinger: The problem is we… I mean, we have essentially no meta-paid social before Ghost.
203 00:21:49.160 ⇒ 00:21:50.899 Zoran Selinger: That channel wasn’t active.
204 00:21:51.670 ⇒ 00:21:58.719 Zoran Selinger: So, I would guess that, essentially, vast majority of this data is very recent.
205 00:21:58.720 ⇒ 00:22:00.800 Robert Tseng: It’s all their… it’s all their data, yeah.
206 00:22:01.000 ⇒ 00:22:01.540 Zoran Selinger: Yeah.
207 00:22:07.830 ⇒ 00:22:08.450 Robert Tseng: Okay.
208 00:22:10.370 ⇒ 00:22:14.890 Robert Tseng: Okay, so when we have a conversation with them, that’s one possibility that we’ll talk about.
209 00:22:15.100 ⇒ 00:22:16.360 Robert Tseng: that…
210 00:22:17.180 ⇒ 00:22:24.740 Robert Tseng: yeah, for whatever reason, they’re not passing… the way that they set it up, the quick IDs are not coming through. Looks like UTMs are more…
211 00:22:24.970 ⇒ 00:22:32.260 Robert Tseng: Well, we’re saying there’s 63,000 sessions, but… 75,000 UTMs.
212 00:22:32.570 ⇒ 00:22:36.960 Robert Tseng: So, UTM’s still off, but it’s, like, off by 10%.
213 00:22:38.020 ⇒ 00:22:41.799 Robert Tseng: Like, how do you explain the 75 to 63 discrepancy?
214 00:22:47.050 ⇒ 00:22:48.799 Zoran Selinger: Right, right.
215 00:22:50.250 ⇒ 00:22:51.550 Zoran Selinger: Interesting.
216 00:22:54.720 ⇒ 00:23:05.279 Zoran Selinger: Oh, right, well, there’s gonna be… so sometimes Facebook, so maybe some of the Facebook link IDs are there,
217 00:23:07.790 ⇒ 00:23:10.270 Zoran Selinger: Wait, so, sorry, let me see…
218 00:23:10.570 ⇒ 00:23:14.860 Zoran Selinger: Yeah, yeah, yeah. So, for some sessions,
219 00:23:15.470 ⇒ 00:23:18.869 Zoran Selinger: some Facebook link IDs are there where there’s…
220 00:23:19.100 ⇒ 00:23:21.760 Zoran Selinger: The tagging is not meta-paid social.
221 00:23:22.340 ⇒ 00:23:38.329 Zoran Selinger: But it’s something else, right? So I know that, for example, they mistagged a few campaigns with paid, not underscore social, but literally a space, social. So that would explain some of the discrepancy.
222 00:23:38.650 ⇒ 00:23:48.109 Zoran Selinger: And I’m not sure if they… if Facebook also puts a click ID on organic clicks. That is also possible.
223 00:23:48.580 ⇒ 00:23:54.109 Zoran Selinger: So some of those are organic, some of those Facebook link IDs are from organic posts.
224 00:23:54.300 ⇒ 00:23:56.590 Robert Tseng: No way it’s that much, though.
225 00:23:58.880 ⇒ 00:24:02.999 Zoran Selinger: But how much, what, this is, what, 3K?
226 00:24:03.550 ⇒ 00:24:05.160 Zoran Selinger: Is that a lot?
227 00:24:07.070 ⇒ 00:24:08.600 Robert Tseng: Maybe it is, yeah, maybe.
228 00:24:09.440 ⇒ 00:24:11.539 Robert Tseng: Well, it’s 12, it’s 13K, right?
229 00:24:12.250 ⇒ 00:24:19.519 Zoran Selinger: In any case, I think… I think it’s clear that we should use UTMs to determine if it’s
230 00:24:19.630 ⇒ 00:24:26.440 Zoran Selinger: can we ask for that change right away, and see what the impact would be? I mean, I can investigate this myself.
231 00:24:29.080 ⇒ 00:24:30.289 Zoran Selinger: I could try to run…
232 00:24:30.290 ⇒ 00:24:34.479 Robert Tseng: We would… we would make the change in our model, or I guess you would make the change in the model, and…
233 00:24:34.750 ⇒ 00:24:36.040 Robert Tseng: Just tell them that.
234 00:24:36.680 ⇒ 00:24:38.120 Robert Tseng: We’re gonna make that change.
235 00:24:39.520 ⇒ 00:24:47.620 Robert Tseng: I don’t want to just keep… do you think that’s… like, what’s… I don’t know what to believe. So, I mean, if you think the UTMs are more accurate, then sure, but, like.
236 00:24:48.450 ⇒ 00:24:52.250 Zoran Selinger: I mean, according to this, it’s… they’re way more accurate.
237 00:24:52.710 ⇒ 00:24:54.530 Zoran Selinger: No, it’s not even close.
238 00:24:56.520 ⇒ 00:25:00.580 Zoran Selinger: I mean, obviously, Robert, I would… I would do the… the OR.
239 00:25:01.090 ⇒ 00:25:11.139 Zoran Selinger: I will do the OR, and I’ve done OR before. Like, if Facebook link ID is there, consider it Facebook, or if the UTMs are this and that, consider it Facebook.
240 00:25:11.640 ⇒ 00:25:21.699 Zoran Selinger: Okay. There’s no reason to use one or the, or the other, just use… Everything’s the criteria there.
241 00:25:21.700 ⇒ 00:25:30.330 Robert Tseng: Okay, yeah, let’s just make that change then. And, like, we can… I mean, there’s still got a lot of explanation or whatever, so we have to share what we’ve learned, so I’ll put that together.
242 00:25:31.830 ⇒ 00:25:49.639 Zoran Selinger: Okay, cool. I’ll let you know what the impact is. I have a call with Jasmine right now. After that call, I’ll go through that exercise and see what the impact is, and then I’ll let you know if you say, okay, push that live, I’m gonna…
243 00:25:49.640 ⇒ 00:25:50.190 Robert Tseng: Yeah.
244 00:25:50.190 ⇒ 00:25:50.780 Zoran Selinger: Oh, cool.
245 00:25:50.970 ⇒ 00:25:56.479 Zoran Selinger: I’m gonna make a PR… being a wish to have a look and…
246 00:25:57.210 ⇒ 00:26:09.389 Zoran Selinger: We’ll change the model. In any case, in any case, they will retroactively get everything. Facebook will complain, Meta will complain, because they want to get events almost immediately after they happen, and not.
247 00:26:09.390 ⇒ 00:26:09.980 Robert Tseng: Yeah.
248 00:26:09.980 ⇒ 00:26:16.060 Zoran Selinger: Like, a week after. But, that should still be fine.
249 00:26:16.740 ⇒ 00:26:17.310 Robert Tseng: Okay.
250 00:26:17.520 ⇒ 00:26:18.050 Zoran Selinger: Yeah.
251 00:26:18.400 ⇒ 00:26:18.930 Zoran Selinger: Okay.
252 00:26:18.930 ⇒ 00:26:25.539 Robert Tseng: Okay, sounds good. Cool. What core did you use to find this? Can I, can I, can I just, like, look at your… let me, let me look into it a bit more, too.
253 00:26:25.700 ⇒ 00:26:27.350 Zoran Selinger: Sorry? Right, dear.
254 00:26:28.200 ⇒ 00:26:30.620 Robert Tseng: They… this message that you sent me with the… with the.
255 00:26:30.620 ⇒ 00:26:38.980 Zoran Selinger: Oh, yeah, yeah, yeah, so, okay. So, for the last one… So here’s the query.
256 00:26:40.210 ⇒ 00:26:48.100 Zoran Selinger: I was just counting, everything from the table where I have meta paid social, and where Facebook ClickID is null.
257 00:26:48.600 ⇒ 00:26:52.119 Zoran Selinger: Okay. I’m getting 64 4K.
258 00:26:52.630 ⇒ 00:26:54.050 Zoran Selinger: of rows.
259 00:26:55.220 ⇒ 00:26:55.770 Robert Tseng: Okay.
260 00:26:56.520 ⇒ 00:26:58.570 Zoran Selinger: Okay, cool. Alright, we’ll talk later.
261 00:26:59.210 ⇒ 00:26:59.880 Zoran Selinger: Yeah. Cheers.