Meeting Title: [Eden] Client Data Pipeline Sync Date: 2025-07-09 Meeting participants: Robert Tseng, Awaish Kumar, Fireflies.ai Notetaker Tigran, Amber Lin, Annie Yu, Demilade Agboola, Josh
WEBVTT
1 00:03:16.720 ⇒ 00:03:18.820 Robert Tseng: Hey? Amber, anyways.
2 00:03:19.670 ⇒ 00:03:20.820 Amber Lin: Hell.
3 00:03:25.070 ⇒ 00:03:27.390 Robert Tseng: Just gonna wait another minute for people to trickle in.
4 00:03:46.540 ⇒ 00:03:48.619 Robert Tseng: Okay, I believe that
5 00:03:49.180 ⇒ 00:03:56.230 Robert Tseng: Cinnati would join later. He usually has a conflict with urban stems or something on on this day. So I’ll I’ll just get started.
6 00:03:56.380 ⇒ 00:03:57.686 Robert Tseng: And then
7 00:03:58.590 ⇒ 00:04:03.779 Robert Tseng: yeah, I think just to just to start. So we have, you know, and just to kind of
8 00:04:06.040 ⇒ 00:04:10.029 Robert Tseng: start high, higher level. And then we can get to the nitty gritty. So
9 00:04:10.150 ⇒ 00:04:20.159 Robert Tseng: yeah, I mean, amber is kind of slowly being transitioned onto this client so hopefully. The project management will look cleaner. And have you guys
10 00:04:20.519 ⇒ 00:04:30.099 Robert Tseng: probably just like working with me because I make you do a lot of stuff that you don’t necessarily expect to do but yeah, I think
11 00:04:30.480 ⇒ 00:04:32.720 Robert Tseng: where we’re doing kind of a
12 00:04:33.438 ⇒ 00:04:45.961 Robert Tseng: you know we’ll do a proper like grooming session and kind of I’ll be onboarding her. And Henry, who is a guy who started with us today. Like like later later today. So
13 00:04:46.450 ⇒ 00:04:51.230 Robert Tseng: Henry, it will take over
14 00:04:51.890 ⇒ 00:05:03.229 Robert Tseng: just like certain projects. I think I’ll have him assist on the Cdp work specifically, just to start. I think he’s gonna just keep running with what we started as in, you know. Wish and I
15 00:05:04.230 ⇒ 00:05:05.110 Robert Tseng: but
16 00:05:05.600 ⇒ 00:05:15.779 Robert Tseng: you know hopefully, well, you’re just gonna minimize interruption as much as we can. I’m just giving you a heads up that like I’m trialing this arrangement to see how how this will work.
17 00:05:16.405 ⇒ 00:05:24.864 Robert Tseng: So appreciate you kind of just being aware that there are gonna be people that maybe we don’t typically
18 00:05:25.620 ⇒ 00:05:32.700 Robert Tseng: work with on this client. But yeah, just we’re always trying to find ways to to level up, make things better. So
19 00:05:33.169 ⇒ 00:05:36.399 Robert Tseng: yeah, just wanted to kind of start with that
20 00:05:38.422 ⇒ 00:05:44.850 Robert Tseng: and then. Yeah, I guess as far as like, kind of work that’s been doing that has been done this past week.
21 00:05:47.010 ⇒ 00:05:49.170 Robert Tseng: Yeah, I think I’ve
22 00:05:50.100 ⇒ 00:06:07.470 Robert Tseng: kind of where I would want to spend this time. One is just kind of closing out on. I know there was a lot of back and forth on the pharmacy stuff. So because that’s top of mind for the team right now, any ad hoc escalations that have come out from there, and also like our
23 00:06:07.780 ⇒ 00:06:14.089 Robert Tseng: the the whole time to turnaround time investigation. I want to close that conversation out
24 00:06:14.565 ⇒ 00:06:26.269 Robert Tseng: and then I’ll probably want to spend some time talking about the cogs related work, because that’s just been a lot of his plate for some time. So I want to see how we can get that over the finish line.
25 00:06:26.736 ⇒ 00:06:55.149 Robert Tseng: and then, yeah, Annie, there’s like a couple small ad hoc things on your plate. In addition to like longer term projects. So there isn’t really too much like for you to deliver this week. I am slowing the pace of the team down so at least this week, so away. And I can. You know, we’re we’re definitely absorbing more of the hours. On the Cdp work. So that’s that’s that’s just a line of song where we’re at. So plus.
26 00:06:55.800 ⇒ 00:06:58.329 Robert Tseng: yeah, I guess any questions before I jump into it.
27 00:07:03.240 ⇒ 00:07:12.750 Robert Tseng: Okay, if not, then let’s talk about the pharmacy stuff. So I’m just gonna pull up the pharmacy channel. And then we’re just gonna kind of point out some things here. So
28 00:07:13.215 ⇒ 00:07:32.050 Robert Tseng: yeah, I think, based on, you know. Thank you, Annie and Dev, a lot of you guys kind of investigated this yesterday. This was my response to them. I think basically, if you haven’t, I guess, before I summarize it? Have both of you read this message like, do you agree with this? I guess I didn’t run it by you. But this was my understanding of what you’re the point that you’re trying to make to me.
29 00:07:36.831 ⇒ 00:07:39.840 Demilade Agboola: Yes, largely the same point.
30 00:07:40.030 ⇒ 00:07:42.569 Demilade Agboola: Yeah. I think you covered most of it.
31 00:07:43.310 ⇒ 00:07:46.450 Robert Tseng: Okay, cool. Yeah. So I think it was basically showing them.
32 00:07:46.710 ⇒ 00:07:52.020 Robert Tseng: I mean, yes, they were. They’re they were concerned about the whole 50, 58%, I think.
33 00:07:52.899 ⇒ 00:08:02.450 Robert Tseng: but then we’re also telling them, like, you know, this is a moving metric. It may not settle over 1010, 1014 days. So you know, Danny is kind of putting some.
34 00:08:02.580 ⇒ 00:08:20.230 Robert Tseng: Try to press them on like, what’s what’s really like. What can we do here? I don’t know, if you agree with my solution, which I’m basically saying, like, maybe we should just consider a wider margin of error. And this is regards to like the Timestamps themselves, will not be consistent across fast, and what we report and ship out
35 00:08:20.270 ⇒ 00:08:31.051 Robert Tseng: we don’t know what 3rd party tool they’re using. If it’s not ship. Oh, you know I there’s only a handful of other companies that do this. At one of my previous companies.
36 00:08:31.380 ⇒ 00:08:53.300 Robert Tseng: there, I forgot. I think it’s called ship station. It’s basically like a 3rd party like tool that just scrapes all of these like parcel companies like you usps Ups Fedex and they go. And they try to like, just basically create like a database of the of the packages, and like the and the in the Timestamps. And they they basically sell that as a service. So
37 00:08:53.579 ⇒ 00:09:05.599 Robert Tseng: that is the cheapest one out there. I wouldn’t be surprised if that’s what vast uses. But once again, it’s just an assumption we don’t fully understand what they use ship over is a very similar business model. So I think
38 00:09:06.430 ⇒ 00:09:30.379 Robert Tseng: that from what I’ve seen before, there is going to be some lag time there, but it won’t be more than you know, like 12 h. So I think, keeping that in mind that there’s a 12 h, probably margin of error, for, like what we can do in terms of getting the Timestamps to align. I think that should cover everything from after the order has left the pharmacy to the customer store.
39 00:09:30.862 ⇒ 00:09:49.360 Robert Tseng: But the question from my perspective is still well, ship doesn’t tell us when the order is actually sent to the pharmacy. And what’s happened? The timestamps that are happening there. We don’t really have another data source to audit against. We’re just taking what Bass tells us right?
40 00:09:49.660 ⇒ 00:09:53.760 Josh : Can’t you guys can’t you guys take when the order is completed
41 00:09:54.580 ⇒ 00:09:57.600 Josh : inside of like vast? So just.
42 00:09:57.600 ⇒ 00:09:58.340 Robert Tseng: Use right now.
43 00:09:58.560 ⇒ 00:09:59.280 Josh : Yeah.
44 00:10:00.549 ⇒ 00:10:07.809 Josh : we’ll just start the clock like, have a clock starting there. So it’s like time it takes from an order from a customer placing an order
45 00:10:08.560 ⇒ 00:10:17.949 Josh : to them, receiving, and if it goes over 4 or 5 days, then that should be a trigger, for necessarily the breach, and then the team can figure out where the fuck that’s happened.
46 00:10:22.930 ⇒ 00:10:23.590 Robert Tseng: Yeah.
47 00:10:24.096 ⇒ 00:10:25.110 Robert Tseng: Go ahead.
48 00:10:25.470 ⇒ 00:10:30.999 Demilade Agboola: I was gonna say, we could do that. I I think the reason why we didn’t is because the sla is
49 00:10:31.300 ⇒ 00:10:35.869 Demilade Agboola: more of the more of time between ordering and shipping.
50 00:10:36.790 ⇒ 00:10:42.350 Demilade Agboola: so we didn’t want to like add a new metric that, you know, is different.
51 00:10:43.220 ⇒ 00:10:54.500 Josh : I mean, you just call it out. You can keep both. I mean cause. Ultimately we should be able to track it at each step. But if you guys are saying that you can’t track it at each step, we need to have some sort of intermedium, some media area between those things.
52 00:10:55.390 ⇒ 00:10:56.730 Josh : You see what I’m saying
53 00:10:56.870 ⇒ 00:11:02.140 Josh : like, right now, you guys, can’t. You guys don’t have insight to where it’s locked up with the doctors right.
54 00:11:03.040 ⇒ 00:11:08.340 Robert Tseng: Correct. We can’t audit that. We just take whatever best to give us for that.
55 00:11:08.570 ⇒ 00:11:14.699 Josh : But you can round about round in a roundabout fashion. Kind of figure those numbers out.
56 00:11:14.850 ⇒ 00:11:16.009 Robert Tseng: Yeah, no, I agree.
57 00:11:16.680 ⇒ 00:11:18.950 Demilade Agboola: Yeah, we can have a ballpark estimates. Yes.
58 00:11:19.310 ⇒ 00:11:31.449 Robert Tseng: And we talked about this yesterday with having the second chart of not just being like the pharmacy kind of related like that that we were trying to expand it when the order was placed to when it leaves the pharmacy right?
59 00:11:40.470 ⇒ 00:11:41.879 Robert Tseng: Hello! Are you muted?
60 00:11:45.485 ⇒ 00:11:56.160 Annie Yu: No, in this ticket we were just fixing the sla, and then at least, according to ticket and the transcript up between you and Sarah, I added.
61 00:11:57.970 ⇒ 00:12:02.089 Annie Yu: the orders that were out of Sla.
62 00:12:04.540 ⇒ 00:12:19.850 Robert Tseng: Yeah. So we that was a whole conversation about increasing the denominator. We changed the like we we changed like the actual timestamp we from when the order I we made an adjustment there. Right then we just move the whole like
63 00:12:20.100 ⇒ 00:12:24.499 Robert Tseng: we. We moved the start of the clock earlier. Isn’t that? Isn’t that what we did.
64 00:12:24.830 ⇒ 00:12:33.700 Annie Yu: No, the denominator stays the same, which is all orders sent to pharmacy, not from when they were placed.
65 00:12:34.350 ⇒ 00:12:42.179 Robert Tseng: Okay? So we we still kept set the pharmacy. No, I think that. Yeah, we we should create a second view of when it starts from when order is completed.
66 00:12:42.290 ⇒ 00:12:49.049 Robert Tseng: because if we have both, we have both parts of that story, then we know what the difference is between when the order
67 00:12:49.230 ⇒ 00:13:00.310 Robert Tseng: is, we know we always know, like how many orders that were completed haven’t made it to pharmacy yet, and that’s that’s like the culprit bucket. That’s also one where, like
68 00:13:00.530 ⇒ 00:13:08.429 Robert Tseng: we, that’s that’s what needs to be. We have. That’s what we need to go. We can put it back on Rebecca’s team and tell them to investigate. Why haven’t these orders got made to the pharmacy?
69 00:13:08.970 ⇒ 00:13:11.410 Robert Tseng: That’s why we need to have those, both both of those views.
70 00:13:11.580 ⇒ 00:13:19.959 Annie Yu: That. Can you? Help me understand again? So for the second chart, for the side by side view, we want to have
71 00:13:20.190 ⇒ 00:13:23.980 Annie Yu: what? As denominator. And what’s numerator.
72 00:13:26.140 ⇒ 00:13:27.026 Robert Tseng: Let me.
73 00:13:27.920 ⇒ 00:13:31.327 Robert Tseng: I know that you put screenshots of the ticket, but I’d rather just
74 00:13:31.760 ⇒ 00:13:32.649 Annie Yu: Yeah, yeah.
75 00:13:47.388 ⇒ 00:13:49.269 Robert Tseng: What’s it called the name of it?
76 00:13:50.490 ⇒ 00:13:52.290 Annie Yu: The second one. Yeah.
77 00:13:52.290 ⇒ 00:13:52.880 Robert Tseng: Yep, yep.
78 00:13:58.720 ⇒ 00:14:22.684 Robert Tseng: okay, yeah. So this one I was like, okay, this is within the farm, like, once the order gets to the pharmacy, then there’s a turnaround time we’re tracking here. I’m saying that we should have like a second chart above this. That’s like from when the order order completed to like shipped out of pharmacy. Turnaround time we can. We can play with the wording there. But it, just it’ll move the the start of the clock earlier.
79 00:14:23.000 ⇒ 00:14:42.720 Robert Tseng: And then, you know, if if I were to be able to look at both of those. So let’s say, you know, on average, we’re seeing. It’s like like a 3 and a half. But maybe, like the average is like 4 and a half or something. Then we would be able to see the difference between like we, we need to know, like how many orders have been completed, or that you know, that haven’t been sent to pharmacy.
80 00:14:43.160 ⇒ 00:14:51.639 Robert Tseng: Yeah, there’s a delta. There’s a delta, and like being able to make sure that that Delta is captured and like the team can like get like access to that data.
81 00:14:51.890 ⇒ 00:14:57.050 Josh : Is the most important part of this this sheet. So it’s like, Hey, all of the orders are here
82 00:14:57.190 ⇒ 00:15:06.910 Josh : and then based on all the orders, then we also know how many orders have been sent to the pharmacy. There’s like one category of orders that have been placed, but have not made it to the pharmacy.
83 00:15:06.980 ⇒ 00:15:29.270 Josh : From there we’ll be able to tell, hey? Some of these were just denied. The doctors denied these orders and other ones. They just have not acted. And so like the this the Mx team knows like, Hey, I need to take action on these orders, and then the same thing then goes for like orders that have been sent to pharmacy that are just sitting for too long, because those are like, maybe an order status, error, or something.
84 00:15:29.370 ⇒ 00:15:35.720 Josh : See what I’m saying. Like, there’s gonna be like these delta points that happen where it’s like a bigger number to a slightly smaller number, to a slightly smaller number.
85 00:15:36.480 ⇒ 00:15:41.609 Josh : and like them, being able to get that data is the most important thing, because then they can actually take action on it.
86 00:15:45.900 ⇒ 00:15:51.160 Robert Tseng: Yeah, I mean, I think the ask is pretty clear in my in my mind. So I think the the
87 00:15:51.690 ⇒ 00:16:03.860 Robert Tseng: yes, we need to expand some of these definitions out of Sla is not just orders not shipped within 3 days of being set to pharmacy. So I don’t know. And and if you’re getting confused on like, where like how to expand this definition.
88 00:16:04.270 ⇒ 00:16:10.470 Robert Tseng: we’re we’re creating like a canonical sequence of like events from order all the way to to end of shift. Right
89 00:16:10.660 ⇒ 00:16:27.619 Robert Tseng: previously. We’re working with a limited set of events. Now we need to expand our definition with it, because there’s a there’s a bit of a gap between. There was that we, the assumption that all orders that were completed make it to pharmacy in a reasonable amount of time like that assumption is being challenged with what we’re seeing in in the operational
90 00:16:28.220 ⇒ 00:16:41.620 Robert Tseng: pick up. So the past past couple of weeks. Right? So that’s why we need to add that additional stage. We don’t necessarily need new tracking to do that. We can create proxy events, using what we already have. And so that’s why we’re gonna able to identify that delta that Josh is talking about
91 00:16:42.065 ⇒ 00:16:51.989 Robert Tseng: and if we need to expand these categories that we can talk, we can, we can talk about that. I think it would even be helpful to diagram this out. So you can, you can know, like all the different stages. But to me.
92 00:16:52.240 ⇒ 00:17:17.110 Robert Tseng: This out of sla of like the this is what the pharmacy is held accountable for. Once it arrives at their at their door. What hasn’t been shipped out within X number of days. But then there’s also something that happens that we need to go out and figure out which is, it’s also out of sla, or like an impact out of Sla, that before the order even arrives at the pharmacy. So I think these categories maybe need to be expanded. And I mean.
93 00:17:17.190 ⇒ 00:17:26.860 Robert Tseng: I I think, before you just go and start building things like, just run run it by me. Make sure that you’re you’re the mental model we have is like pretty clear on how we’re going to expand like our set.
94 00:17:27.859 ⇒ 00:17:29.519 Annie Yu: Yeah, yeah, okay, I will.
95 00:17:29.520 ⇒ 00:17:30.839 Robert Tseng: Okay, cool.
96 00:17:31.364 ⇒ 00:17:40.159 Robert Tseng: Yeah, I don’t think this requires any additional data modeling. Like, I think this is purely just a data visualization exercise. So like, I think we shouldn’t take very long on this.
97 00:17:42.790 ⇒ 00:17:46.960 Robert Tseng: Okay, cool. But I can
98 00:17:47.510 ⇒ 00:17:52.050 Robert Tseng: update the ticket. But I’ll probably have to do after this call. So I’m not gonna do it. Live here.
99 00:17:54.390 ⇒ 00:17:56.249 Robert Tseng: Cool any other questions on this?
100 00:18:00.130 ⇒ 00:18:03.241 Robert Tseng: No, okay. Then we’ll go back.
101 00:18:04.600 ⇒ 00:18:16.080 Robert Tseng: yeah. I saw I kind of closed out the tickets from the whole treatment. Id thing! I know you added to the model. There’s not necessarily an additional ask for it now.
102 00:18:16.190 ⇒ 00:18:25.400 Robert Tseng: I think there will be, and I don’t think we need to do it right like today. But I kind of I I mentioned it in the documentation of
103 00:18:25.630 ⇒ 00:18:32.309 Robert Tseng: what I was trying to do here. So okay, I think I’m just gonna reiterate it. If you can go in and look at this model that I had
104 00:18:32.420 ⇒ 00:18:34.555 Robert Tseng: kind of outlined doubts.
105 00:18:35.330 ⇒ 00:18:40.679 Robert Tseng: this idea of a treatment journey. Summary model. So this is going to be a separate
106 00:18:40.880 ⇒ 00:18:48.010 Robert Tseng: model that starts from like when they intake. This is purely from the patient, like it’s a canonical timeline, a sequence of
107 00:18:48.050 ⇒ 00:19:00.319 Robert Tseng: of patient actions from the start of their intake to when they, whatever interaction they have with the doctor. I mean this, it’s it’s expanding beyond just order stages at this point, right? Because this is what’s going to.
108 00:19:00.330 ⇒ 00:19:23.970 Robert Tseng: you know whether it’s Bobby or somebody else like, later on we’re going to be able to hand them. They’re going to be able to trigger campaigns off of like use, patient actions, which is a different set of events. That’s not just the orders that we’re talking about, Annie. So I kind of already, like mapped out some of those components, I think, want you to fill in the blanks on like our. If these are sufficient or like what we do and don’t have.
109 00:19:23.990 ⇒ 00:19:30.099 Robert Tseng: But I think that’s what you know, rather than waiting on Bobby to give the requirements like, I think this model is what
110 00:19:30.410 ⇒ 00:19:42.940 Robert Tseng: we’re gonna be able to enable once we’re done with the Cdp work for this next week. But like, yeah, this is this is what I think your your, what you what you did with pulling in the treatment Id is going to impact the most.
111 00:19:45.170 ⇒ 00:19:49.099 Robert Tseng: So I can create a spike for this. But I guess. Does that make sense.
112 00:19:50.517 ⇒ 00:19:54.510 Demilade Agboola: Yes, it does make sense. I
113 00:19:56.190 ⇒ 00:20:02.450 Demilade Agboola: I just wanted to like Flag that cause, you know, the task was there, and ultimately I don’t know if
114 00:20:02.980 ⇒ 00:20:06.369 Demilade Agboola: anybody will be handling the activation emails right now.
115 00:20:07.125 ⇒ 00:20:10.250 Demilade Agboola: But then, I just don’t want to just like waste
116 00:20:10.890 ⇒ 00:20:13.760 Demilade Agboola: and quote unquote doing something that will not be used.
117 00:20:14.360 ⇒ 00:20:29.009 Robert Tseng: Yeah, yeah, no. I think. I think what you did was fine. I’m just saying like, this is probably a directional gonna be headed in. Who like. You know how he’s gonna handle that transition. But he hasn’t responded yet. So
118 00:20:30.770 ⇒ 00:20:32.240 Demilade Agboola: Okay. Sounds good.
119 00:20:32.240 ⇒ 00:20:49.639 Robert Tseng: Okay, yeah, this is just wanting you to like, you know, check my concept. See? This makes sense. And then, like, what? What can we actually support with this. But I think once you know, once we build out the requirements here, then I’ll add it to cycle. But I think this is kind of what we’re what I think is the next best option.
120 00:20:51.070 ⇒ 00:20:56.970 Demilade Agboola: Okay, yeah, I’ll look through. I’ll look through and add some more information about it. Like, you know, maybe there’s stuff we can add.
121 00:20:57.220 ⇒ 00:21:07.682 Robert Tseng: Yeah, leave leave comments all over the same. Yeah, I think I’ve this is, this is kind of just my my working doc, rather than staying with tickets for all the CD stuff.
122 00:21:07.990 ⇒ 00:21:08.650 Demilade Agboola: Malta.
123 00:21:09.080 ⇒ 00:21:22.599 Robert Tseng: Okay, cool. And then last thing I wanna touch on before I give a wish. And I probably have a bunch of updates as the as the cogs work here. So we were trying to do the reconciliation and the duplication of cogs.
124 00:21:23.110 ⇒ 00:21:29.779 Robert Tseng: and also investigate like what additional cost wheels. So these are all things we kind of touched on last week.
125 00:21:30.090 ⇒ 00:21:31.730 Robert Tseng: How how far can we get here?
126 00:21:33.113 ⇒ 00:21:42.070 Demilade Agboola: So for cogs, the duplication isn’t really a hard part in the sense that it’s literally just adding something to the model to dedupe
127 00:21:42.741 ⇒ 00:22:03.469 Demilade Agboola: the problem is that there are multiple values of the cogs, and that is because of, you know, the historical like cogs, changes over time. It’s not a static value. So because that is handling, that is the tricky overall. Ask which is kind of another ticket there as well.
128 00:22:04.041 ⇒ 00:22:18.840 Demilade Agboola: And which is the retroactivity handling. So for that I am thinking of 2 different solutions, and I will try and see which one works best with the Eden team like Rebecca, and how that works best for her there are 2 ways. One is
129 00:22:19.330 ⇒ 00:22:23.140 Demilade Agboola: we have a cogs history sheet, which is just basically
130 00:22:25.650 ⇒ 00:22:35.309 Demilade Agboola: the variant id the product name and valid from cogs, and then valid from, and valid to. So anytime there’s a new cogs value.
131 00:22:35.630 ⇒ 00:22:37.900 Demilade Agboola: You just add in your rolling batch sheet
132 00:22:38.310 ⇒ 00:22:45.589 Demilade Agboola: and put the new cogs. Value the new variant. I like the variant id the product, name the new value of cogs.
133 00:22:45.750 ⇒ 00:22:54.390 Demilade Agboola: and the new valid from dates. So it’s valid from whatever date so like in in the case of when she Rebecca, came and said.
134 00:22:54.530 ⇒ 00:22:57.720 Demilade Agboola: It’s now valid from the 1st of June.
135 00:22:57.850 ⇒ 00:23:08.439 Demilade Agboola: That cogs will now be the new value. Old values would now have an end date, and that allows us to like, you know as well enriching our orders. Data, we can say the cogs.
136 00:23:08.990 ⇒ 00:23:11.689 Demilade Agboola: but it I know there’s this is associated with that.
137 00:23:12.740 ⇒ 00:23:22.909 Demilade Agboola: I don’t know how scalable that is especially because it still relies on Rebecca and her team like updating that over time.
138 00:23:23.590 ⇒ 00:23:25.950 Demilade Agboola: That is my drawback with that
139 00:23:26.090 ⇒ 00:23:30.540 Demilade Agboola: alternatively. If we can get the the
140 00:23:31.060 ⇒ 00:23:34.350 Demilade Agboola: pharmacy sheets when she when she gets it.
141 00:23:34.550 ⇒ 00:23:37.410 Demilade Agboola: We can try to then model, based off that
142 00:23:38.110 ⇒ 00:23:41.540 Demilade Agboola: which might be a more sustainable way.
143 00:23:42.428 ⇒ 00:23:48.809 Demilade Agboola: But that also puts that also puts like a huge ask.
144 00:23:49.110 ⇒ 00:23:59.939 Demilade Agboola: because I also, I would. Okay, I would capture it as Rebecca for that. But, like ultimately, that would require us to continuously like, manage the different sheets that come in
145 00:24:00.050 ⇒ 00:24:02.559 Demilade Agboola: and keep extracting the information we need.
146 00:24:02.860 ⇒ 00:24:08.599 Demilade Agboola: plus also like enriching it, because if it doesn’t come with the dates it’s valid from. We’ll need to manually enrich it
147 00:24:08.720 ⇒ 00:24:10.870 Demilade Agboola: on every single like, impute.
148 00:24:12.920 ⇒ 00:24:30.939 Robert Tseng: Yeah, okay, I mean, I hear you and yeah, so just to kind of quickly respond to that. I mean, this is somewhat related to what I kind of scoped out for a wish on the trait, on the we’re talking about customer data model and the traits there. So I kind of came up with a model there, that basically
149 00:24:33.080 ⇒ 00:24:34.540 Robert Tseng: like is
150 00:24:34.933 ⇒ 00:24:57.579 Robert Tseng: just like a, it’s. It’s a table full of each trait. And then it kind of just has a bunch of different columns that let us know the freshness and the volume, or whatever. So this works for this approach, because we expect the traits to be changing. They could happen they could change daily, and they could change it more than one time in a day. And so we do kind of need something that is consistently pulling updates from a stream of data.
151 00:24:57.660 ⇒ 00:25:08.790 Robert Tseng: But for cards doesn’t change daily, it doesn’t definitely doesn’t change twice at the same day. I think it changes very rarely, I guess you know, probably, like once a quarter is kind of what I would expect, or something less frequent.
152 00:25:09.213 ⇒ 00:25:13.020 Josh : Right now. It’s a little bit more more dynamic.
153 00:25:13.590 ⇒ 00:25:17.909 Josh : that is like, maybe, like once, like you have up to like once a week even
154 00:25:19.010 ⇒ 00:25:23.319 Josh : there’s a lot of supply chain considerations right now.
155 00:25:23.760 ⇒ 00:25:24.510 Robert Tseng: I see.
156 00:25:25.050 ⇒ 00:25:26.160 Robert Tseng: Okay? Well, then, if.
157 00:25:26.160 ⇒ 00:25:34.626 Josh : Less than ideal, less than less than ideal like, obviously, like, we wish it wasn’t like that. But it’s just kinda it is right now. It sucks.
158 00:25:35.860 ⇒ 00:25:36.580 Robert Tseng: I see.
159 00:25:37.340 ⇒ 00:25:39.219 Robert Tseng: Okay? Well, then, if it is
160 00:25:39.450 ⇒ 00:25:42.449 Robert Tseng: changing weekly, then I would prefer not
161 00:25:42.913 ⇒ 00:25:56.429 Robert Tseng: to rely on Rebecca’s team to to just manually update that because that’s just we’re gonna just miss things. So yeah, if you can kind of put both options in front of Rebecca and let her know that we prefer to get. You know this directly.
162 00:25:56.430 ⇒ 00:26:03.279 Josh : You can just go right to Christiana this week about a lot of that stuff. Just go right to Christiana.
163 00:26:03.540 ⇒ 00:26:08.970 Josh : You can copy Rebecca on the stuff. But Christian is gonna be the decision maker, for it.
164 00:26:09.250 ⇒ 00:26:10.350 Robert Tseng: Okay. Cool.
165 00:26:11.100 ⇒ 00:26:14.795 Josh : Hey? I gotta. I gotta jump. I gotta jump here in a minute. But I got one thing so
166 00:26:15.480 ⇒ 00:26:21.465 Josh : I’m just looking through my my, my for Josh report, and
167 00:26:22.400 ⇒ 00:26:25.959 Josh : was just hoping that we could get these new products added in
168 00:26:26.280 ⇒ 00:26:31.520 Josh : Friday, like as of yesterday, like I didn’t see any of the hormone stuff in in there.
169 00:26:36.500 ⇒ 00:26:39.209 Demilade Agboola: I believe you’re editing right now.
170 00:26:41.237 ⇒ 00:26:44.509 Demilade Agboola: Believe they are added in right now. The Hr team.
171 00:26:44.838 ⇒ 00:26:46.480 Robert Tseng: Stuff, isn’t it? In there.
172 00:26:46.480 ⇒ 00:26:47.110 Josh : Yeah.
173 00:26:47.580 ⇒ 00:26:59.260 Josh : I see. I see the kit like. So it’s like, here’s the thing like the naming conventions are kind of confusing me. So like in one report. I’ll say Hrt. Kit, and then the other one, it’ll say, like the actual product names.
174 00:26:59.400 ⇒ 00:27:04.090 Josh : And so I’m just trying to get some consistency, because, like other places, will say, like Estradiol cream.
175 00:27:04.540 ⇒ 00:27:09.389 Josh : And so I don’t know what’s driving some of that I just I’m calling it.
176 00:27:09.770 ⇒ 00:27:12.619 Josh : It’s just a little confusing for the end user.
177 00:27:12.620 ⇒ 00:27:13.450 Demilade Agboola: Yes.
178 00:27:13.650 ⇒ 00:27:23.330 Demilade Agboola: yeah, that’s actually kind of what I asked you yesterday to to provide some clarity on. So the orders the orders contain both the products, ids
179 00:27:23.570 ⇒ 00:27:25.700 Demilade Agboola: and the bundle ids.
180 00:27:25.870 ⇒ 00:27:28.289 Demilade Agboola: So depending on
181 00:27:29.060 ⇒ 00:27:36.680 Demilade Agboola: what we prioritize in our case. Fine statement, it will say, Hey, when the product ids it goes to this. Call it, you know.
182 00:27:36.970 ⇒ 00:27:53.279 Demilade Agboola: extra doll cream when it the bundle. I but the bundle id will not kick in then, because it would already have assigned the order to this. So right now the order, the flow is the bundle id comes first.st It will check if the bundle id matches hrt kits one
183 00:27:53.795 ⇒ 00:28:01.190 Demilade Agboola: and then associate that order to Hrt. Tickets 1, 2, 3, but then, obviously, the product Id names will not kick in as a result.
184 00:28:02.820 ⇒ 00:28:05.920 Demilade Agboola: So it kind of depends on how we want to prioritize them.
185 00:28:06.760 ⇒ 00:28:07.930 Josh : Interesting.
186 00:28:08.250 ⇒ 00:28:08.960 Demilade Agboola: Yeah.
187 00:28:11.590 ⇒ 00:28:19.599 Josh : cause. It’s weird, right cause like on the new product row as dashboard. It’s showing me. Hrt. Kit 3 had one sale yesterday.
188 00:28:21.530 ⇒ 00:28:22.490 Josh : right.
189 00:28:22.900 ⇒ 00:28:23.530 Demilade Agboola: Yeah.
190 00:28:24.090 ⇒ 00:28:25.283 Josh : And then
191 00:28:26.541 ⇒ 00:28:31.058 Josh : I forget which the other one is, and the other one. It’s saying, like Estradiol,
192 00:28:31.570 ⇒ 00:28:36.590 Josh : not some new product sales, and then then the new
193 00:28:36.730 ⇒ 00:28:41.930 Josh : row. Ass I I forget which one. I’m sorry I’m in my car, and I’m actually I’m about to walk in and have an appointment.
194 00:28:42.140 ⇒ 00:28:46.730 Josh : But we can talk about it tomorrow. I just wanted to bring it up
195 00:28:46.910 ⇒ 00:28:49.350 Josh : that I just saw something. So
196 00:28:49.520 ⇒ 00:28:52.340 Josh : we have. It sounds like, you guys have a lot of other stuff going on.
197 00:28:53.270 ⇒ 00:29:12.979 Demilade Agboola: Yeah, but I think we can talk about it tomorrow. But I know that for right now, unless it’s only has the product. Id, and not the bundle id. It will do extra, and then, if it has a bundle id, it would prioritize the bundle id, regardless of if it has a product, Id or not. So it’s call it the Hr tickets, 1, 2, 3.
198 00:29:13.160 ⇒ 00:29:22.659 Demilade Agboola: But potentially, this is also like a data quality issue. If you should ever be a product Id or a bundle id. Then we can look at that and try and get the better data.
199 00:29:26.170 ⇒ 00:29:29.459 Josh : Understood. Okay, cool. Thanks. Guys.
200 00:29:29.780 ⇒ 00:29:30.350 Demilade Agboola: Okay.
201 00:29:30.920 ⇒ 00:29:31.810 Robert Tseng: Thanks.
202 00:29:31.810 ⇒ 00:29:32.340 Josh : Yep.
203 00:29:33.314 ⇒ 00:29:46.400 Robert Tseng: Yeah, I know we’re approaching time. I don’t think that you all need to stay over, I think. Wish, can you? Can you stay on. I can stay on to talk through the updates. But I don’t think everyone else needs to be there. Necessarily.
204 00:29:50.210 ⇒ 00:29:54.719 Robert Tseng: Yeah, I mean, I did have a video summary with a wish.
205 00:29:54.850 ⇒ 00:30:06.850 Robert Tseng: So you all can go and take a look if you’re interested in kind of what our discussion was yesterday. I think you could just go log into the rainforge demo platform and you can pull up.
206 00:30:08.270 ⇒ 00:30:10.090 Robert Tseng: This still might fix this.
207 00:30:11.520 ⇒ 00:30:17.540 Robert Tseng: Okay, I guess you’ll have to go off of the regular dashboard and find.
208 00:30:27.630 ⇒ 00:30:33.390 Robert Tseng: did we not? Oh, okay, yeah. So this one, maybe I’ll just
209 00:30:35.510 ⇒ 00:30:40.940 Robert Tseng: yeah. So you guys want to go and figure out what we’re talking about. You can
210 00:30:41.750 ⇒ 00:30:44.490 Robert Tseng: review this review, this video, look at our transcript.
211 00:30:45.740 ⇒ 00:30:47.460 Robert Tseng: Yeah. Okay.
212 00:30:48.130 ⇒ 00:30:53.540 Robert Tseng: otherwise feel free to drop off unless there’s any other questions, and then a wish, I’ll I’ll stay on for another
213 00:30:53.800 ⇒ 00:30:57.280 Robert Tseng: 30. I can stay on just until we we talk through your updates.
214 00:30:58.763 ⇒ 00:30:59.626 Awaish Kumar: Okay.
215 00:31:01.460 ⇒ 00:31:02.470 Annie Yu: Alright! Thanks.
216 00:31:02.470 ⇒ 00:31:04.539 Demilade Agboola: Alright, thanks, guys! Bye.
217 00:31:07.490 ⇒ 00:31:12.310 Awaish Kumar: Basically, actually, I created the audit table
218 00:31:13.237 ⇒ 00:31:16.950 Awaish Kumar: and the user profile table as well. I have the queries
219 00:31:17.729 ⇒ 00:31:21.920 Awaish Kumar: but right now they are not in the warehouse. Because.
220 00:31:23.168 ⇒ 00:31:30.800 Awaish Kumar: I’m getting issue with running that because this union carry is is basically
221 00:31:30.960 ⇒ 00:31:33.749 Awaish Kumar: it’s a very big one. And
222 00:31:36.150 ⇒ 00:31:42.090 Awaish Kumar: yeah, yeah. So basically, this is the like
223 00:31:42.470 ⇒ 00:31:45.229 Awaish Kumar: processing more than 10 Gb’s of data
224 00:31:46.273 ⇒ 00:31:50.109 Awaish Kumar: and it is not allowed in bigquery. And so I’m
225 00:31:50.280 ⇒ 00:31:53.709 Awaish Kumar: I have like find ways to split it up.
226 00:31:54.130 ⇒ 00:32:00.710 Awaish Kumar: But now, like with the Dvt how to automate that process, I’m working on that one. So once
227 00:32:01.389 ⇒ 00:32:13.149 Awaish Kumar: maybe I I can create up some python DVD Python models or something like that to get it working. So after that, like we can easily process in batches
228 00:32:13.540 ⇒ 00:32:17.840 Awaish Kumar: like the like, the batch of 50 columns.
229 00:32:18.410 ⇒ 00:32:23.907 Awaish Kumar: and then merge it together to create a final table. So I’m I’m just
230 00:32:24.550 ⇒ 00:32:26.820 Awaish Kumar: working on that pipeline right now.
231 00:32:27.740 ⇒ 00:32:30.350 Awaish Kumar: So it will be like done in.
232 00:32:32.050 ⇒ 00:32:37.545 Awaish Kumar: maybe in an hour or something. And then I also have a user profile profiles
233 00:32:39.170 ⇒ 00:32:46.410 Awaish Kumar: kerry. And so what I’ve been when I work on that like once, we have this user
234 00:32:47.120 ⇒ 00:32:53.920 Awaish Kumar: like this trades table, we can like identify meaningful column by providing some filters like
235 00:32:55.810 ⇒ 00:32:59.970 Awaish Kumar: the columns with non, 90% normal values and
236 00:33:00.400 ⇒ 00:33:00.880 Robert Tseng: Yeah.
237 00:33:00.880 ⇒ 00:33:06.950 Awaish Kumar: More than more than 10,000 unique users and things like that. And we get some
238 00:33:07.763 ⇒ 00:33:14.090 Awaish Kumar: a subset of columns which will be part of the user table. And then, yeah, these both will be
239 00:33:14.912 ⇒ 00:33:17.387 Awaish Kumar: there, like sometime today. And then
240 00:33:18.600 ⇒ 00:33:22.906 Awaish Kumar: after that, the 3rd part will be to get it.
241 00:33:25.830 ⇒ 00:33:38.010 Awaish Kumar: like, enriched customer model, which basically will join this user table and our existing table. And and the features you mentioned, like Ltv, total orders and things like that.
242 00:33:39.540 ⇒ 00:33:52.769 Robert Tseng: Great. Yeah, I know that you were kind of building on top of this model. I don’t know if the massive unions is the best way to do it. That was just like every I was just trying to hack it together and get something to work. So if you can run something better for efficient than that.
243 00:33:52.890 ⇒ 00:33:55.248 Robert Tseng: I’m sure there’s better ways.
244 00:33:56.720 ⇒ 00:33:57.720 Awaish Kumar: Yeah.
245 00:33:57.860 ⇒ 00:34:03.530 Awaish Kumar: yeah, I’m just trying, like, I’m working to automate this. But this is the only way to create this kind of table.
246 00:34:03.720 ⇒ 00:34:14.009 Awaish Kumar: But yeah, like, we. I’m just trying every way to automate all this. So we don’t have to provide the column names and anything. So it so it will just read from schema.
247 00:34:15.443 ⇒ 00:34:21.799 Awaish Kumar: Create batches, run the job and create a final table, so it will be.
248 00:34:22.179 ⇒ 00:34:34.039 Robert Tseng: Okay. And I mean, I did call out a few issues. Even my, even my query doesn’t fully work like I I mean, it still broke out a few different parts. Obviously this wasn’t the actual field name I was trying to use some I don’t know. I was trying to.
249 00:34:34.499 ⇒ 00:34:56.628 Robert Tseng: because they they have like some of the traits are just named really weirdly and like it, doesn’t, you know, traditional sequel logic of, you know, the single quotes doesn’t work to catch it. Things like address one or I think one of them was actually called like hex backslash, or something like, and I tried to like fix it by, you know, using using trying to catch, use a
250 00:34:57.219 ⇒ 00:35:08.769 Robert Tseng: these these type of operators. But it didn’t actually do anything. So yeah. So I’m I’m sure you’re seeing the same syntax errors that I did. But yeah, I think that’s why I didn’t end up
251 00:35:09.049 ⇒ 00:35:11.599 Robert Tseng: writing a script because I didn’t really feel like I could
252 00:35:12.219 ⇒ 00:35:17.639 Robert Tseng: sit there and think through a dynamic way to to get everything right. So I just
253 00:35:17.789 ⇒ 00:35:28.329 Robert Tseng: what it did. Kind of what for this approach? But I mean, it sounds like you. Yeah, yeah, I think I think what you’re doing is probably that like, that’s the right approach. I just
254 00:35:28.619 ⇒ 00:35:34.559 Robert Tseng: yeah. Anyway, I just I I think I understand what you’re what you’re what you’re getting at. Okay. So.
255 00:35:34.560 ⇒ 00:35:35.080 Awaish Kumar: So.
256 00:35:35.080 ⇒ 00:35:35.570 Robert Tseng: That.
257 00:35:35.570 ⇒ 00:35:36.460 Awaish Kumar: What?
258 00:35:37.370 ⇒ 00:35:41.320 Awaish Kumar: Yeah. One brief. 2 table are ready today. I will just let you know in the slack.
259 00:35:41.470 ⇒ 00:35:45.909 Awaish Kumar: and after that I will start on the the initial customer model.
260 00:35:46.480 ⇒ 00:36:05.120 Robert Tseng: Great. Okay, yeah, I think I will. I didn’t spend much time in customer I/O, yesterday. I’m gonna go back in there. And yeah, I’m gonna probably clean up some of this. I know that these are just like random sub tasks and things noise. There’s a lot of noise in here. Just to recap
261 00:36:05.774 ⇒ 00:36:33.139 Robert Tseng: we’re trying to build out this model so that it becomes the single customer data model that we used to push into all tools rather than having a segment, have its own model. You know bigquery. We have them customers. It doesn’t use segment. And then Customer I/O has its own set as well. Rob never got back to me on like how he built the customer data model in in customer I/O. But yeah, I’m gonna keep trying to go in and and look at that. See? What could turn off?
262 00:36:33.593 ⇒ 00:36:45.049 Robert Tseng: I think specifically, I wanted to see if there’s anything in these 41 attributes that we can’t pull out of segment or out of our out of bigquery right now, and just try to understand, like, you know, what?
263 00:36:45.320 ⇒ 00:37:03.489 Robert Tseng: Yeah, like, you know, maybe there’s like a couple of things something like empower membership status. This was like a previous before they use circle. I think they were using some other provider. That was like a membership community thing. So this is like one field that I know never went into our data warehouse. It seems like
264 00:37:03.520 ⇒ 00:37:19.140 Robert Tseng: even I would guess that Rob just hooked up in power into Customer I/O. And then like that was just like a label there, this doesn’t exist. We don’t. This is not even active anymore. So it can be removed. But yeah, there’s just like, certain it’s just kind of.
265 00:37:19.820 ⇒ 00:37:26.179 Robert Tseng: I just have to keep auditing all of these fields pretty much, and figure out what we can what we should actually pull into
266 00:37:26.630 ⇒ 00:37:42.239 Robert Tseng: Customer I/O. If there’s anything in there that we currently don’t have that we need to somehow like reroute and bring it into the warehouse first.st so yeah, hopefully, I’ll get here on that today. And as you’re kind of catching up there, yeah.
267 00:37:43.190 ⇒ 00:37:43.850 Awaish Kumar: Okay.
268 00:37:45.450 ⇒ 00:37:46.300 Robert Tseng: Yeah,
269 00:37:47.550 ⇒ 00:37:55.069 Robert Tseng: okay, so that’s it. I think I’m just gonna spend some time customer. I/O, and then I’m gonna onboard. Henry, I think Henry has a better.
270 00:37:55.950 ⇒ 00:38:11.529 Robert Tseng: I mean, I’m curious how he thinks I’ve been approaching this. But yeah, I think we’ll we’ll talk. We’ll talk more about this, let me know if you run into anything. This is still biggest priority for me. So I wanna unblock you whenever you can. Yeah.
271 00:38:12.964 ⇒ 00:38:14.610 Awaish Kumar: Okay, sure, I’ll let you know.
272 00:38:14.840 ⇒ 00:38:16.559 Robert Tseng: Okay. Alright, thanks. Aish.
273 00:38:16.870 ⇒ 00:38:17.520 Awaish Kumar: Right.
274 00:38:17.810 ⇒ 00:38:18.360 Robert Tseng: Bye.