Meeting Title: Catalyst Model Issue Discussion Date: 2026-02-02 Meeting participants: Robert Tseng, Zoran Selinger
WEBVTT
1 00:01:50.590 ⇒ 00:01:51.780 Robert Tseng: Hey, Zaron.
2 00:01:53.760 ⇒ 00:01:54.930 Robert Tseng: Can you hear me?
3 00:01:56.370 ⇒ 00:01:57.709 Robert Tseng: That’s interesting.
4 00:01:59.300 ⇒ 00:02:02.449 Robert Tseng: Testing, testing… can you hear me? Can you hear me?
5 00:02:04.380 ⇒ 00:02:06.950 Robert Tseng: Looks like my audio’s going through.
6 00:02:08.060 ⇒ 00:02:14.310 Robert Tseng: But you’re also… Muted.
7 00:02:15.590 ⇒ 00:02:17.949 Robert Tseng: Can you hear me? Testing, testing…
8 00:02:27.610 ⇒ 00:02:28.650 Robert Tseng: Okay.
9 00:02:39.040 ⇒ 00:02:47.330 Robert Tseng: Okay, great. Alright, well, I guess, like, I’m following the threads here,
10 00:02:50.230 ⇒ 00:02:56.339 Robert Tseng: Yeah, it just feels like what Greg is saying is, like, very similar to what we discussed last week already.
11 00:02:56.700 ⇒ 00:03:02.089 Robert Tseng: like, the problem isn’t sending duplicate orders, like, we already knew that was, like, we’ve already…
12 00:03:02.810 ⇒ 00:03:10.539 Robert Tseng: Catalyst isn’t double counting those. We don’t have a problem with double counting. So, I feel like, yeah, that’s… that seems like a…
13 00:03:11.060 ⇒ 00:03:14.609 Robert Tseng: Not really that important for him to address.
14 00:03:17.480 ⇒ 00:03:28.030 Robert Tseng: I’m trying to understand, like, what will moving to fully relying on the client side is going to change. So…
15 00:03:48.160 ⇒ 00:03:52.580 Robert Tseng: So it sounds like you made the 7-day criteria adjustment already.
16 00:03:56.970 ⇒ 00:03:58.220 Robert Tseng: So…
17 00:04:03.490 ⇒ 00:04:06.169 Robert Tseng: 7 days also not relevant, okay?
18 00:04:14.590 ⇒ 00:04:15.470 Robert Tseng: Yeah.
19 00:04:34.030 ⇒ 00:04:39.920 Robert Tseng: Okay, so are you waiting for Awash to update the model, and then you’re gonna basically backtest?
20 00:04:49.220 ⇒ 00:04:54.090 Robert Tseng: Like, I don’t… I guess I don’t need to, like, interrupt what you’re doing, I just… I just gotta, like, ELT…
21 00:04:54.490 ⇒ 00:05:08.789 Robert Tseng: times this morning already, and they’ve asked me, didn’t you and Zoran look into this last week already? Why is this still an issue? So, that’s the first one. It’s like, is this your guy’s fault? So, those are really the two things that I need to be able to answer.
22 00:05:08.840 ⇒ 00:05:22.669 Robert Tseng: Sounds like you’re working with Ryan to only use GTM, so maybe Awash is no longer… Awash and Greg are no longer part of… part of the solution for now? Or, like, I guess, like, I mean, if I can make a status update, like, I don’t know what I should say.
23 00:05:40.920 ⇒ 00:05:44.380 Robert Tseng: if we only use fully GTM, then, like.
24 00:05:44.790 ⇒ 00:05:57.810 Robert Tseng: does that discount some of the other work that we’ve done? Like, I… I thought the point of, like, having the model, sending it from segment into catalyst was that it’s, like, quote-unquote, server-to-server, so…
25 00:05:58.590 ⇒ 00:06:02.950 Robert Tseng: Like, I don’t know, that’s… But that’ll…
26 00:06:03.470 ⇒ 00:06:12.459 Robert Tseng: if I’m going back to them and saying, hey, actually, our server-to-server solution’s not great, there’s clearly issues with it, we’re just gonna go full client-side.
27 00:06:12.460 ⇒ 00:06:28.370 Robert Tseng: then, like, I can anticipate that they’re going to question, like, why did we go down this route in the first place? This edge layer still being used? Like, if we’re just going back to GTM, then, like, why did we even, like, kind of do all the stuff? So, that’s what I’m anticipating their reaction will be.
28 00:07:06.870 ⇒ 00:07:09.720 Robert Tseng: Okay, so maybe the way that I’ll put to say it is…
29 00:07:10.280 ⇒ 00:07:23.700 Robert Tseng: The client-side discrepancies we need to resolve first, so we just gotta switch over to a full GTM approach. That will be the fastest way to get the conversion rates back up to what the partners will expect.
30 00:07:23.940 ⇒ 00:07:38.029 Robert Tseng: Then, are we basically going to… then we’ll have more time to clean up the server… the server-to-server solution later on? Is that… is that fair to say? Or are we just gonna… are you suggesting that we should just not use server-to-server at all?
31 00:07:38.610 ⇒ 00:07:42.669 Robert Tseng: And we should just… we should just not go… we should just not use segment for this at all.
32 00:07:47.280 ⇒ 00:07:50.320 Robert Tseng: Okay, okay, that’s fair.
33 00:07:51.390 ⇒ 00:08:07.699 Robert Tseng: Yeah, I guess, like, there was just… there were some… if they’re gonna ask, okay, well, what parts of the model did we not agree on? And we may not have to get into that detail, but it sounds like attribution window was kind of, like, unclear. You know, you thought 14 days, someone else said 7 days.
34 00:08:08.170 ⇒ 00:08:12.989 Robert Tseng: And then… I guess, like, we were sending in…
35 00:08:13.190 ⇒ 00:08:17.730 Robert Tseng: Certain types of duplicates, or multiple records, just…
36 00:08:17.970 ⇒ 00:08:25.919 Robert Tseng: I mean, that’s more kind of art… I mean, that doesn’t really affect performance, so I don’t really think that matters, but I’m trying to… trying to think of, like.
37 00:08:26.180 ⇒ 00:08:32.620 Robert Tseng: What, like, what are we hung up on, and, like, what are we disagreeing on that we need to have more time to figure out?
38 00:08:38.590 ⇒ 00:08:43.100 Robert Tseng: maybe this is going back to, like, kind of the first order,
39 00:08:44.250 ⇒ 00:08:54.030 Robert Tseng: like, I mean, we added some, you know, conditional logic on, like, what is considered a successful catalyst order, so maybe it’s, like, we need to spend more time
40 00:08:54.590 ⇒ 00:08:58.190 Robert Tseng: Okay, we had a simple rules-based conversion, first time only.
41 00:09:08.260 ⇒ 00:09:17.899 Robert Tseng: Okay, so you feel like we changed the model a couple times already. Ryan was not that involved, so maybe, like, our model kind of has drifted from what his expectation was.
42 00:09:18.040 ⇒ 00:09:27.350 Robert Tseng: Or, like, I mean, do you understand what I’m trying to, like, get at? It’s like, why is our server-to-server model wrong? Didn’t we have so much… didn’t we have a lot of time to figure this out already?
43 00:10:06.030 ⇒ 00:10:16.550 Robert Tseng: It’s okay if you don’t know the answer right now. I guess, like, you know, you can just tell me that. Maybe we need to do a retro afterwards. I just… I’m just trying to cover my bases so I don’t have to come back to you.
44 00:10:17.100 ⇒ 00:10:23.769 Robert Tseng: What is wrong? Palm is that there’s a lot of complexity, technical group, etc. Okay, sure. Okay, fine, I can, I can, I can accept that answer.
45 00:10:23.960 ⇒ 00:10:25.850 Robert Tseng: Yeah, I’ll…
46 00:10:26.210 ⇒ 00:10:39.960 Robert Tseng: I’ll say that to them. I think I should probably signal to Awash and Greg to stop, too, then. Like, I feel like they’re just, like, kind of spinning their wheels. Okay, alright, that sounds good. Okay, thanks, Ron.
47 00:10:41.040 ⇒ 00:10:41.810 Robert Tseng: Bye.