Meeting Title: Segment Integration Overview for Default Date: 2026-02-26 Meeting participants: Robert Tseng, Greg Stoutenburg


WEBVTT

1 00:00:24.820 00:00:25.680 Greg Stoutenburg: Hey, Robert.

2 00:00:26.680 00:00:27.670 Robert Tseng: Hey, how’s it going?

3 00:00:28.010 00:00:30.039 Greg Stoutenburg: Good, thanks for hopping on for this.

4 00:00:30.390 00:00:31.000 Robert Tseng: Yeah.

5 00:00:41.260 00:00:41.950 Greg Stoutenburg: Yeah.

6 00:00:47.420 00:00:50.720 Robert Tseng: Yeah, we’re gonna… we’re gonna poke around and segment.

7 00:00:50.780 00:00:52.629 Greg Stoutenburg: Or has anyone else? Okay.

8 00:00:52.890 00:00:56.379 Greg Stoutenburg: Yeah, so here’s Default’s segment,

9 00:00:56.530 00:01:09.689 Greg Stoutenburg: Environment. So, basically, like, I mean, first of all, just, like, a broad overview of segment would be great. I mean, I mean, I think I know basically what it does, but it’s probably good to hear from someone else. Yeah. And then, like, the reason I’m asking this question is because

10 00:01:10.180 00:01:13.780 Greg Stoutenburg: I know that they’re already sending stuff into Amplitude.

11 00:01:13.830 00:01:32.160 Greg Stoutenburg: And, for that matter, a variety of other places from Segment. They’ve decided that they want to do Post Hog for product analytics. Why they’re gonna have Amplitude sending stuff into Polytomic, and then Post Hog for product analytics, I don’t know, but that seems to be the decision. But basically, what I…

12 00:01:32.270 00:01:53.319 Greg Stoutenburg: what I want to get an understanding of is how Segment works with product analytics tools to do things like stitch user IDs together, like we were talking about Mixpanel the other day. Yeah. Because I don’t really understand how that works. So, since we’re moving into the user properties phase of the implementation for default, I feel like now is the time to be inquiring about this.

13 00:01:54.260 00:01:57.790 Robert Tseng: Sure, I wish I could.

14 00:01:58.880 00:02:04.410 Robert Tseng: access this, and then go click around. But, I mean, basically,

15 00:02:05.330 00:02:11.569 Robert Tseng: Yeah, you have, like, your protocols in UniFi on the left, so that’s pretty much,

16 00:02:11.760 00:02:18.690 Robert Tseng: better be, like, protocols is, like, their tracking pad. Unify is really, like, kind of their user… user profiles, kind of, like,

17 00:02:20.220 00:02:26.709 Robert Tseng: product. It doesn’t look like they’re even using it currently. Seems like it’s, like, not an add-on that they’ve turned on.

18 00:02:27.100 00:02:27.470 Greg Stoutenburg: Yeah.

19 00:02:29.800 00:02:32.979 Robert Tseng: Yeah, so I don’t even have that turned on, but…

20 00:02:33.170 00:02:36.390 Greg Stoutenburg: The idea is if you have multiple source.

21 00:02:36.800 00:02:44.820 Robert Tseng: data. Because, like, as you… as you route data into segment, it will… it will, by default, kind of create

22 00:02:45.020 00:02:59.470 Robert Tseng: a segment, like, segment ID, right? It’s like an anonymized ID. They try to stitch it across sessions, so, like, it’s a way for you to, like, aggregate multiple anonymous IDs into a single segment ID.

23 00:02:59.610 00:03:06.330 Robert Tseng: And they do that mostly, through… I mean, like…

24 00:03:07.090 00:03:13.220 Robert Tseng: there’s all this, like, logic that’s built into it. So, for example, if you have,

25 00:03:14.160 00:03:18.940 Robert Tseng: I guess default is… I don’t know if they have multiple,

26 00:03:21.520 00:03:29.209 Robert Tseng: domains for, like, one for the Phoenix product and one for the main one, but if… assuming that there were multiple.

27 00:03:30.150 00:03:32.939 Greg Stoutenburg: Domains, it’s the same person going to each of them.

28 00:03:32.940 00:03:49.279 Robert Tseng: you know, with… with the segment kind of, like, tracker installed, like, you… you would be able to say, like, hey, is this the same person that clicked on, like, all… on, like, two or other… or, like, two or three domains, and… and that would be stitched automatically in segment.

29 00:03:49.450 00:03:57.640 Robert Tseng: What we found is that the stitching segment is not complete, which is why, for Eden, we do it further in BigQuery, so…

30 00:03:58.340 00:04:03.170 Robert Tseng: not every data source gets put into segment, at… for Eden, and so…

31 00:04:03.490 00:04:11.700 Robert Tseng: But everything does end up in BigQuery. So we actually manage, like, an enriched customer profile table or meeting separately in the data warehouse.

32 00:04:11.940 00:04:16.990 Robert Tseng: And that’s… And then from there, we use segment to…

33 00:04:17.200 00:04:29.529 Robert Tseng: use reverse ETL to push our data in the warehouse into different destinations. So we don’t really rely on, segment through UniFi at Eden, and I think that’s generally

34 00:04:29.980 00:04:37.320 Robert Tseng: I wouldn’t say default needs to get there anytime soon, because it doesn’t look like they have that much data, and…

35 00:04:37.480 00:04:50.740 Robert Tseng: But I think it’s… it’s easier to get some, like, baseline level of identity stitching out of the boss’s segment, rather than having to custom build it, by building in all your own logic in BigQuery.

36 00:04:51.130 00:04:52.429 Greg Stoutenburg: Okay, okay.

37 00:04:53.160 00:04:54.939 Greg Stoutenburg: Okay, that’s helpful.

38 00:04:56.020 00:05:02.070 Robert Tseng: Yeah, so, I mean, from here, I’m seeing it’s just… it’s just a Node.js app and the website. So, I mean, everything is.

39 00:05:02.070 00:05:02.510 Greg Stoutenburg: Right.

40 00:05:02.510 00:05:07.149 Robert Tseng: You know, it’s a single-page application, and…

41 00:05:07.390 00:05:13.800 Robert Tseng: I mean, I don’t know if we click into… if we just click into what one of the live sources look at, like, default production or something.

42 00:05:15.170 00:05:17.040 Robert Tseng: See what they have coming in here.

43 00:05:17.230 00:05:21.080 Robert Tseng: Can I just scroll down, get a sense of what events that they’re firing in here.

44 00:05:21.190 00:05:26.599 Robert Tseng: Okay, they’re all identified track events. Can we look at event name? Just want to get a sense of what they’re showing here.

45 00:05:27.290 00:05:31.660 Robert Tseng: Okay, yeah, so these are mostly just, like,

46 00:05:32.120 00:05:43.050 Robert Tseng: user actions on their website, right? Booking meetings, logins, like, yeah, they have some level of, like, event creation already here. Looks like 50% of the events have no names, so I don’t even know what the heck that is.

47 00:05:43.380 00:05:46.859 Robert Tseng: That’s pretty noisy, so that’s definitely a red flag to me.

48 00:05:47.020 00:05:52.140 Robert Tseng: But other than that, like, it’s mostly, like, meetings booked, okay? That makes sense to me.

49 00:05:57.000 00:05:59.040 Robert Tseng: Yeah, and then…

50 00:05:59.200 00:06:05.549 Robert Tseng: Where are they pushing this data into? If you scroll down, they’re pushing it into… okay, amplitude.

51 00:06:05.690 00:06:12.859 Robert Tseng: Okay, that all… that all makes sense. I don’t know why it’s failing at Intercom, though. Looks like Intercom’s only receiving half of them.

52 00:06:15.660 00:06:21.190 Robert Tseng: But that’s okay. What about their other source? Like, I don’t know what they have. They have two sources?

53 00:06:23.650 00:06:26.510 Greg Stoutenburg: So, they have the Node.js stuff, and then the other one just said website.

54 00:06:26.860 00:06:32.630 Robert Tseng: His website, but it’s also no reasonable data, so let’s see. Maybe there isn’t actually anything coming in.

55 00:06:32.780 00:06:36.270 Robert Tseng: Yeah, so it’s purely just off a Node.js website.

56 00:06:37.540 00:06:39.770 Robert Tseng: Yeah, we’re not really sending anything else from here.

57 00:06:40.480 00:06:41.240 Greg Stoutenburg: Okay.

58 00:06:41.970 00:06:48.380 Robert Tseng: Yeah. So, if it’s only one source, I don’t frankly know why they need segments for one source, but… Yeah, okay.

59 00:06:48.380 00:06:52.260 Greg Stoutenburg: Okay, that helps, yeah. I’m like, well, this is what I’m trying to understand, I’m like, alright, I…

60 00:06:52.260 00:07:09.240 Greg Stoutenburg: because there were these earlier conversations where they were like, oh yeah, you know, we’ve been using Segment to push stuff to Amplitude, and I thought, like, alright, you know, sometimes it’s hard to tell, am I the person who doesn’t know what’s going on, or are they the person who doesn’t know what’s going on? Because I thought, well, why don’t you just send it to Amplitude?

61 00:07:09.240 00:07:09.830 Robert Tseng: Yeah.

62 00:07:09.830 00:07:12.839 Greg Stoutenburg: And there really wasn’t an answer to that, and now it’s… I think you’ve.

63 00:07:12.840 00:07:25.760 Robert Tseng: Now it’s pretty clear, yeah. Well, they probably just use it for, like, the destinations brought. It looks like… Sure. Well, from… Amplitude doesn’t have as many destinations it can send to, natively, so that’s probably why.

64 00:07:25.760 00:07:26.340 Greg Stoutenburg: Yeah.

65 00:07:28.100 00:07:29.280 Robert Tseng: Yeah…

66 00:07:29.770 00:07:30.520 Greg Stoutenburg: Okay.

67 00:07:32.620 00:07:38.570 Greg Stoutenburg: Alright, well then… Maybe it would benefit me to just get

68 00:07:38.700 00:07:43.140 Greg Stoutenburg: clear on… well, I mean, first of all, we can’t instrument this in Phoenix right now anyway, because it’s not locked.

69 00:07:43.140 00:07:43.490 Robert Tseng: Yeah.

70 00:07:43.490 00:07:51.709 Greg Stoutenburg: But as we… as we figure out what… if they want to continue collecting the same events through segment,

71 00:07:52.340 00:07:57.790 Greg Stoutenburg: we can at least raise the question, do you want to just add post hog as a destination for that set of events?

72 00:07:57.790 00:07:58.670 Robert Tseng: Yeah.

73 00:07:58.710 00:07:59.600 Greg Stoutenburg: Yeah.

74 00:07:59.790 00:08:01.429 Greg Stoutenburg: This is the one I clicked before, right?

75 00:08:02.250 00:08:12.860 Robert Tseng: Yeah, it is easier because it’s just, like, one… I mean, they already have the tracking code in there, and so all the destinations, you don’t have to… you don’t have to add any more code to do it, so that’s probably why they set it up this way.

76 00:08:13.080 00:08:13.780 Greg Stoutenburg: Yeah.

77 00:08:14.080 00:08:14.840 Greg Stoutenburg: Okay.

78 00:08:15.800 00:08:16.670 Greg Stoutenburg: Okay.

79 00:08:16.830 00:08:19.070 Greg Stoutenburg: Yeah. Alright, well, alright.

80 00:08:19.070 00:08:34.119 Robert Tseng: But segment to PostHog kind of just defeats the purpose of posthog. PostHog is trying to, like, be its own standalone thing, like, it must be code-friendly, like, the whole no-code approach for instrumenting posthog seems kind of like an anti-pattern, what they’re trying to do.

81 00:08:34.710 00:08:40.990 Greg Stoutenburg: Yeah, I… I think… I think that it does, and then, you know, especially if you combine that with…

82 00:08:41.140 00:08:44.880 Greg Stoutenburg: what we see here in the connections page, if they’re just…

83 00:08:45.320 00:08:50.469 Greg Stoutenburg: collecting data from the one source and sending it to all these places. Like, any product…

84 00:08:50.710 00:08:54.859 Greg Stoutenburg: I wouldn’t be surprised if PostHog has all of these connections already.

85 00:08:55.130 00:08:55.790 Robert Tseng: Yeah.

86 00:08:58.370 00:09:00.440 Greg Stoutenburg: So, they’re probably just paying for segment for nothing.

87 00:09:00.780 00:09:04.040 Robert Tseng: I wonder how much they’re paying for segment. Can we see it? Are we an admin?

88 00:09:04.440 00:09:05.230 Greg Stoutenburg: Let’s see.

89 00:09:06.410 00:09:12.020 Robert Tseng: Going… Under user preferences, yeah.

90 00:09:12.600 00:09:13.470 Robert Tseng: Billing.

91 00:09:14.240 00:09:15.479 Robert Tseng: Usage and billing.

92 00:09:15.780 00:09:16.800 Greg Stoutenburg: Oh, yeah.

93 00:09:17.980 00:09:21.719 Robert Tseng: Ability, 10,000 entities…

94 00:09:23.030 00:09:24.070 Greg Stoutenburg: 120 a month.

95 00:09:24.640 00:09:27.769 Robert Tseng: Yeah, I mean, they’re barely paying anything, it’s probably fine.

96 00:09:28.180 00:09:28.850 Greg Stoutenburg: Yeah.

97 00:09:30.200 00:09:38.030 Robert Tseng: Yeah, I mean, like, the Eden bill is, like… when I first there, it was, like, 60 grand a year, and then we cut it down to maybe, like, 20 grand.

98 00:09:38.100 00:09:39.170 Greg Stoutenburg: Oh, God.

99 00:09:39.170 00:09:43.250 Robert Tseng: So, then that probably makes more of a difference. But, I mean, we’re paying, like.

100 00:09:44.160 00:09:44.999 Greg Stoutenburg: Like, you know.

101 00:09:45.150 00:09:48.640 Robert Tseng: You know, not much, so it should be… if that makes sense.

102 00:09:48.970 00:09:51.620 Greg Stoutenburg: Yeah. If we look here, they’re also using, like.

103 00:09:52.050 00:09:54.110 Greg Stoutenburg: You know, 15% of their allowance.

104 00:09:54.760 00:09:55.300 Greg Stoutenburg: In line.

105 00:09:55.300 00:09:56.220 Robert Tseng: Yeah.

106 00:09:57.780 00:10:03.940 Robert Tseng: Yeah, billing by MTUs makes sense for them. They probably have more traffic than that, but, like, MTUs,

107 00:10:04.380 00:10:10.960 Robert Tseng: Oh yeah, you either get billed by users, or you get billed by Yvette account, or, like, a SaaS product.

108 00:10:11.320 00:10:16.719 Robert Tseng: user count is gonna be much lower. But, like, eCom, you know, like.

109 00:10:16.970 00:10:31.970 Robert Tseng: you’re… you know, most of the users that you’re tracking are not actually your customers, so it doesn’t make sense to bill off of MTUs, which was what Eden was doing before. So even just switching it from user to event billing kind of cut their bill by half, and then we optimize it a bit more from there.

110 00:10:32.400 00:10:34.059 Greg Stoutenburg: Got it, got it, okay. Yeah.

111 00:10:34.320 00:10:38.500 Greg Stoutenburg: Yeah, good to know as I look at the Mixed panel project next for Eden, then.

112 00:10:38.670 00:10:39.200 Robert Tseng: Yeah.

113 00:10:39.500 00:10:40.090 Greg Stoutenburg: Yep.

114 00:10:40.470 00:10:41.370 Greg Stoutenburg: Okay

115 00:10:41.950 00:10:51.499 Greg Stoutenburg: Alright, last question then. For these connections, I mean, right, this is just get an engineer and install the snippet, right? Pick your events, same way you did instrument anything else. Okay.

116 00:10:51.540 00:10:52.210 Robert Tseng: Yep.

117 00:10:52.250 00:10:53.010 Greg Stoutenburg: Cool.

118 00:10:53.580 00:10:55.410 Greg Stoutenburg: Alright, I think those are all my questions.

119 00:10:55.780 00:11:01.189 Robert Tseng: Yeah, I mean, if they had Unify, what you could do is you could go in there, you could add custom properties, like.

120 00:11:01.330 00:11:10.889 Robert Tseng: directly in here, which is nice, because you don’t have to instrument those as well. But you could, you know, you could do form captures and, like, do other, like, yeah, like.

121 00:11:10.910 00:11:25.500 Robert Tseng: if… if it’s, like, you know, it’s just… it is easier, because you don’t have to actually enrich events, like, you can enrich events just off of the UI here, not have to go and use code to do it, so, that’s why this is…

122 00:11:25.720 00:11:27.769 Robert Tseng: Helpful to start, but .

123 00:11:27.770 00:11:29.439 Greg Stoutenburg: I guess when I’m using it, so…

124 00:11:30.310 00:11:34.799 Greg Stoutenburg: Okay. Alright, yeah, that’s really helpful. Yeah, thanks a ton, Robert.

125 00:11:35.240 00:11:36.179 Robert Tseng: Yeah, of course.

126 00:11:36.640 00:11:38.109 Greg Stoutenburg: Alright, talk to you later.

127 00:11:38.110 00:11:38.740 Robert Tseng: Okay, man.

128 00:11:39.000 00:11:39.940 Greg Stoutenburg: Bye.