Meeting Title: Magic Spoon — Brainforge sync Date: 2026-02-20 Meeting participants: Demilade Agboola, Michael Thorson, Justin Tabarini


WEBVTT

1 00:01:12.330 00:01:14.020 Michael Thorson: Hey, what’s up?

2 00:01:18.840 00:01:20.579 Demilade Agboola: Hi, Mike. Hi, Michael, how you doing?

3 00:01:20.900 00:01:22.530 Michael Thorson: Good! How you doing?

4 00:01:22.660 00:01:30.740 Demilade Agboola: Pretty good, it’s been a long week, but it’s… we’re, like, at the end, so that’s pretty good.

5 00:01:31.100 00:01:33.409 Michael Thorson: Dude, you’re telling me, it’s been crazy.

6 00:01:34.080 00:01:37.079 Michael Thorson: Where are you, where are you based out of again? I’m totally spacing.

7 00:01:37.700 00:01:43.820 Demilade Agboola: Malta, so it’s, like, an island off the coast of Italy. It’s its own country, but it’s off the coast of Italy.

8 00:01:43.940 00:01:47.610 Michael Thorson: Oh, yeah, yeah, yeah. How long have you been there? Is that where you’re from, or…

9 00:01:47.610 00:01:53.050 Demilade Agboola: Oh, no, no, no, so I’m from Nigeria, and I came here, like, 3 years ago, so I’ve been here for 3 years.

10 00:01:53.410 00:01:55.160 Demilade Agboola: Nice, nice.

11 00:01:55.160 00:01:58.699 Michael Thorson: Was it work that took you to Malta, or was it choice, or…

12 00:01:58.700 00:02:04.150 Demilade Agboola: I decided to do, like, a digital nomad, like, visa thingy, so that’s what brought me here.

13 00:02:04.500 00:02:06.330 Demilade Agboola: But yeah.

14 00:02:06.930 00:02:15.230 Demilade Agboola: I… I do spend… I do visit, like, the US quite a bit. I’m usually, like, when I do visit, I mean, like, Minnesota, most of the time.

15 00:02:16.580 00:02:19.530 Demilade Agboola: Thankfully, I’m not around right now, it’s really cold.

16 00:02:19.530 00:02:26.760 Michael Thorson: Seriously, it’s, like, so bleak in New York, and I’m sure Minnesota is just, like, even gnarlier, but…

17 00:02:26.760 00:02:27.999 Demilade Agboola: Yeah, fair, bye.

18 00:02:28.420 00:02:31.650 Demilade Agboola: Actually, I’ve been to, like, New York a couple times, it’s pretty cool.

19 00:02:31.910 00:02:37.139 Demilade Agboola: pretty busy. I don’t think I could live there, but it was… it was a nice… it was a nice trip.

20 00:02:37.460 00:02:40.979 Michael Thorson: Yeah, man, Malta sounds nice right about now, I’ll tell ya.

21 00:02:41.760 00:02:46.539 Demilade Agboola: Yeah, it’s, like, in the 40s, 50s, temperature-wise. It’s not… it’s not the worst.

22 00:02:47.300 00:02:52.600 Michael Thorson: Bingo. That’s awesome. Do you have, like, what’s, what’s… what brings you to Minnesota? Like, work stuff, or…

23 00:02:52.600 00:02:54.980 Demilade Agboola: Oh, no, no, so my girlfriend lives there, so,

24 00:02:55.530 00:02:59.329 Demilade Agboola: Constantly, shuttling up and down because of that.

25 00:02:59.760 00:03:03.019 Michael Thorson: Nice. Wow. World traveler. Nomad, I guess.

26 00:03:03.370 00:03:05.530 Demilade Agboola: Exactly, living up to the title.

27 00:03:06.120 00:03:07.830 Michael Thorson: Yeah, seriously, that’s awesome.

28 00:03:07.830 00:03:09.180 Demilade Agboola: Yeah, it’s pretty cool.

29 00:03:10.100 00:03:16.149 Demilade Agboola: Okay, I think this is the crew. Not sure, let me quickly confirm, based off the invite.

30 00:03:17.380 00:03:18.519 Michael Thorson: Yeah, Mary’s out today.

31 00:03:19.310 00:03:20.190 Demilade Agboola: Yeah.

32 00:03:20.580 00:03:24.710 Demilade Agboola: Yeah, no time also declined, but… yes, I think this is the crew.

33 00:03:25.180 00:03:28.820 Michael Thorson: Cool. Alright, so let’s just kind of walk through what’s being done.

34 00:03:28.820 00:03:29.579 Demilade Agboola: this week.

35 00:03:35.040 00:03:36.890 Demilade Agboola: Can you see my screen?

36 00:03:38.550 00:03:39.460 Michael Thorson: Bingo.

37 00:03:39.460 00:03:41.050 Demilade Agboola: Alright, all good, let’s go.

38 00:03:41.840 00:03:46.689 Demilade Agboola: So yeah, typical Project North Stars,

39 00:03:47.450 00:03:52.919 Demilade Agboola: So we’re just gonna, like, talk through, like, the Spins API update from this week.

40 00:03:53.330 00:03:57.549 Demilade Agboola: The MMA Mart update this week, render transition and takeover

41 00:03:57.760 00:04:03.419 Demilade Agboola: As well as, like, just the next engagement and renewal, and then next step on blockers.

42 00:04:04.210 00:04:12.789 Demilade Agboola: So yeah, this week, like, spins have responded that, they will not be able to… will not be able to aggregate these values based off weekly data.

43 00:04:13.220 00:04:18.250 Demilade Agboola: So what we’re trying to do now is just get that based off of multiple timeframes.

44 00:04:18.790 00:04:25.890 Demilade Agboola: And so now, what we’ve also been doing is, like, Extracting for multiple…

45 00:04:26.190 00:04:28.929 Demilade Agboola: Brands, and that has been done.

46 00:04:29.260 00:04:36.529 Demilade Agboola: And so the next step is also to extract data for multiple timeframes. So that’s kind of the next step with the Spint API.

47 00:04:37.560 00:04:50.910 Demilade Agboola: For the MMM art, forgot to send the pull request in, but I’ve actually done it. I know you made the request for, like, the base table, so that’s been done. And also it has been rerun with that data, against…

48 00:04:51.020 00:04:56.449 Demilade Agboola: So, for census region and… Magic Spoon against,

49 00:04:57.200 00:05:00.329 Demilade Agboola: The weekly table, so that’s… that’s in there as well.

50 00:05:00.510 00:05:17.190 Demilade Agboola: I think that’s basically it. So, like, we’ve done the base table, we’ve been able to integrate everything else, and so all we just need is the ACV and TDP. I’m… before the call, because I remembered when I was putting the slides together, I just remembered I hadn’t put the PR, so I have, sent the PR in.

51 00:05:17.810 00:05:22.150 Demilade Agboola: So it can be reviewed, and anyone who can merge it can merge it.

52 00:05:25.860 00:05:32.390 Demilade Agboola: Yeah, so basically also, like, vendor transition and takeover. So right now, like, that’s currently the phase we’re in. We’re trying to…

53 00:05:32.690 00:05:41.390 Demilade Agboola: You know, have meetings with them, ensure that we are, currently defining what, like, the entire structure

54 00:05:41.680 00:05:44.450 Demilade Agboola: What was… what is currently being handled?

55 00:05:44.730 00:05:48.600 Demilade Agboola: Any documentation that we may need in terms of

56 00:05:49.140 00:05:56.829 Demilade Agboola: Runbooks, and how things break, or just, like, you know, things that they might, you know, know, or any,

57 00:05:57.300 00:06:00.139 Demilade Agboola: Knowledge that they could be able to transfer over to us.

58 00:06:00.330 00:06:01.820 Demilade Agboola: Prior to them leaving.

59 00:06:02.050 00:06:07.330 Demilade Agboola: But yeah, so that’s currently what’s going on, and we’re just ensuring that we’re on top of things in that sense.

60 00:06:09.460 00:06:16.400 Demilade Agboola: And then, yeah, so we’re also talking about, like, the renewal and SOW. I know that stuff we’re talking with Mary.

61 00:06:16.580 00:06:21.970 Demilade Agboola: And I know… Tom will send over some, document to…

62 00:06:22.870 00:06:28.099 Demilade Agboola: Josh, because I know, Joshua, because I know, Mary’s out of office today.

63 00:06:28.360 00:06:32.330 Demilade Agboola: So I believe OTAM is in the process of doing that. I know

64 00:06:33.130 00:06:42.099 Demilade Agboola: It’s a bit slug… it’s a bit slow, because I know Tommy’s talking to Robert. Robert’s the other CEO. They’re trying to finalize some things, and we’ll… once that’s done, they’ll send it over.

65 00:06:42.390 00:06:50.310 Demilade Agboola: So once that’s sent over, and every other thing is finalized, then we can move into the next phase of the engagement.

66 00:06:52.730 00:06:57.269 Demilade Agboola: Okay, so I think in summary, just like a high level of this week,

67 00:06:57.970 00:07:02.290 Demilade Agboola: For the spins, basically, we’re just trying to extract the data across multiple time frames.

68 00:07:04.810 00:07:13.820 Demilade Agboola: MMM Matters done, at least for now, unless there’s anything, like, Justin or Michael have to, like, in terms of feedback, which you want maybe

69 00:07:14.040 00:07:18.079 Demilade Agboola: things changed or anything done, but other than that, everything has been done.

70 00:07:23.000 00:07:41.370 Michael Thorson: What’s… yeah, I had a question, maybe this is for Ashwini as well, like, what’s actually in warehouse at current state? I haven’t been able… he did an extract, I think, yesterday, and we had some, like, limitations, and I just wanted to make sure I understand, like, what exactly was landed in warehouse for the one-week time frame.

71 00:07:41.470 00:07:46.680 Michael Thorson: Do you know, or have you picked up that, or is that a question for Ashwini?

72 00:07:46.680 00:07:54.399 Demilade Agboola: I do not know, but I could always… I could always ask him, and ask him to, like, message you on the group, and just… just put it there, so what was extracted this week.

73 00:07:54.710 00:08:14.959 Michael Thorson: Yeah, I’ll ask him some, like, core questions, and make sure I get a summary of, like, what exactly is in Warehouse, and I’ll also take a look at that in parallel. So, I’ll just, like… I want to make sure everyone’s aware of exactly, like, current state that Ashwini landed, and just, like, a line there, before we start to pull the higher-level timeframes of, like, 4-week, 12-week, etc.

74 00:08:14.960 00:08:20.700 Michael Thorson: Okay. Because there’s still, like, a ton of value of having, like, a full backfill of our week

75 00:08:20.700 00:08:21.650 Michael Thorson: to weak.

76 00:08:21.910 00:08:26.970 Michael Thorson: spins data for both Magic Spoon and, like, all other, relevant brands.

77 00:08:27.190 00:08:35.789 Demilade Agboola: Okay, that’s fair, that’s fair. So once we do that, then you want us to do the multiple time frames of 4, 12, 24, 52, okay.

78 00:08:35.909 00:08:37.469 Demilade Agboola: Yeah. Alright, sounds good.

79 00:08:38.620 00:08:51.989 Michael Thorson: Cool. So that’s… yeah, that’s my big question on the data engineering, like, side of things, is just, like, what’s our current state? And, like, before we step into the multiple time frames and, like, complicate things even further, just, like, make sure we have the right

80 00:08:52.140 00:08:54.700 Michael Thorson: Backfill in place, and then…

81 00:08:55.030 00:09:15.219 Michael Thorson: Number 2 is more on the modeling side, now that we have this kind of, like, large data lake, and we’re, like, expanding to different brands. We talked internally about, like, next steps in terms of modeling, because we’ve delivered, like, an MMM model, but there’s also different use cases that we’re looking to extract from that core

82 00:09:15.220 00:09:18.549 Michael Thorson: Like, like, lake of information.

83 00:09:18.550 00:09:28.139 Michael Thorson: are we, like, is that still, like, that’s kind of the deliverable we were driving towards with the spins delivery in warehouse? It’s, like, actually, like, workable, usable models?

84 00:09:28.270 00:09:37.049 Michael Thorson: Yeah, what’s your… what do you think your timeline or, like, roadmap is for tackling or, like, standing up those initial models, Demi?

85 00:09:40.240 00:09:47.110 Demilade Agboola: So, it depends. I know, in that case, we’ll need to have, like, conversations around what that looks like, what the initial models look like.

86 00:09:47.830 00:10:06.799 Demilade Agboola: In terms of, like, what level of granularity are we looking at? Are we going to be looking at different, like, brands? Do we want to stand them up in terms of brands, in terms of regions? Like, how do we… or geography level, how do we want to go about that? And once we have a clearer idea of what that looks like, we can then have a clearer idea of how things come together to make

87 00:10:06.850 00:10:08.979 Demilade Agboola: Whatever insights you want to see.

88 00:10:09.140 00:10:21.620 Demilade Agboola: Nice. And so once we have that, like, flow, we can then decide how we want to, one, break it down, and two, create, like, a timeline for that. So it’s kind of hard to determine

89 00:10:21.770 00:10:25.590 Demilade Agboola: timeline right now, when I don’t really know the full scope of what that entails.

90 00:10:26.220 00:10:28.319 Michael Thorson: Cool. Yeah, we, I…

91 00:10:28.320 00:10:51.889 Michael Thorson: I’m prepping kind of, like, a… treating, like, a spins data mart roadmap, because there’s obviously a lot of ways we can slice this, and, like, we don’t want to report, like, multiple reporting levels in the same report, for example. We want to just have, like, one level of granularity, and we’ll roll that up in our BI tool. So, we have kind of, like, a priority hit list of, like, hey, these are the most important, like, first two, and, like, the set of filters.

92 00:10:51.960 00:11:02.030 Michael Thorson: I’d love to just, like, since we’re… we have, like, technical people on the call, just, like, maybe take a look at them together before today’s call ends, and just be like, hey, is this the right way to communicate?

93 00:11:02.130 00:11:04.889 Michael Thorson: how we want to go about these data marts.

94 00:11:05.130 00:11:06.010 Demilade Agboola: Sure. Gabby.

95 00:11:06.120 00:11:15.620 Michael Thorson: Cool, I’m like, yeah, we can hop into that after that, just want to leave, like, any other questions on this week, JT, or Demi, or, like, anything you need from us, high level?

96 00:11:16.390 00:11:27.420 Demilade Agboola: Right now, no. I think the only, like, major, like, blocker is just the contract stuff, but, like, personally, that’s, like, above me right now, so just waiting for, like, once that’s confirmed.

97 00:11:27.530 00:11:28.580 Demilade Agboola: We’re all good.

98 00:11:29.130 00:11:29.820 Michael Thorson: Cool.

99 00:11:30.390 00:11:34.550 Michael Thorson: Yeah, same note. We’re all, we’re all aligned there. No fans.

100 00:11:35.130 00:11:35.900 Michael Thorson: Cool. Okay.

101 00:11:36.100 00:11:42.059 Michael Thorson: Yeah, I’ll hop in, though. Let me, share my screen, where’d I put that?

102 00:11:42.920 00:11:49.770 Michael Thorson: Yeah, so, like, I kind of split this out, like, after our internal conversations, we had this, like, massive working doc, you know?

103 00:11:50.030 00:11:57.810 Michael Thorson: It’s like, I’m trying to keep it as updated as possible, but I split it out into, like, an ETL roadmap, where this is where I’ve been communicating with Ashwini of, like.

104 00:11:58.000 00:12:05.300 Michael Thorson: what we need to do, what polls do we want first? So, I think this is maybe what he did this week. He’s, like, through number 3.

105 00:12:05.630 00:12:12.009 Michael Thorson: Pull the backfill for, like, all brands and, like, all geographies.

106 00:12:13.780 00:12:27.420 Michael Thorson: so doing kind of, like, a brand-level extract, and then there’s also, like, a UPC-level extract. So this is, like, for only a couple brands, we want to look at UPC. For context, I mean, like, these are our competitors.

107 00:12:27.420 00:12:29.599 Demilade Agboola: So it’s like, we don’t necessarily want…

108 00:12:29.600 00:12:31.279 Michael Thorson: A bunch of random…

109 00:12:31.400 00:12:37.120 Michael Thorson: like, UPC noise, because it’s, like, that’s… that’s, like, a singular product, so the data gets, like, massive.

110 00:12:37.290 00:12:48.759 Michael Thorson: So, like, yeah, for the UPC reporting level, we’re only looking at this, like, very, like, smaller subset. So that’s just, like, important to kind of keep in mind. And then, like, at the brand level, so it’s, like.

111 00:12:48.860 00:12:59.349 Michael Thorson: basically it’ll be, like, a brand, so it’ll be, like, you know, Catalina Crunch, they have multiple subcategories, so they could be selling any, like, cereal, or, like, granola, or, you know, etc, etc.

112 00:12:59.580 00:13:08.440 Michael Thorson: But it’ll be rolled up to, like, the product category, which is, like, where we do most of our analysis. So it’s like, how is Magic Spoon Cereal doing versus…

113 00:13:08.850 00:13:24.630 Michael Thorson: Cascading Farms or Catalina Crunch cereal, you know? So, that’s kind of the primer of, like, how we view the data, usually. And then, obviously, like, the different slices come into, like, these other key, dimensions, which I’m sure you’re familiar with.

114 00:13:25.040 00:13:32.329 Michael Thorson: But, yeah, so that’s kind of the ETL roadmap, and I’ll keep kind of working with Ashwini there, and then this is, I don’t know,

115 00:13:32.490 00:13:35.439 Michael Thorson: This is kind of what we landed on for, like.

116 00:13:35.650 00:13:55.600 Michael Thorson: the most important, kind of, like, starter data marts that we’re interested in. And I’m trying to, like, name them. I don’t really have a naming convention that I’m tied to, so if you, like, want to update these, this could be, like, an input field. But these are, like… I wanted to provide, like, a set of very discrete, like, filter guides, so it’s like…

117 00:13:55.780 00:14:08.589 Michael Thorson: geography level is total US, so this is, like, not really a regional view. It’s, like, what’s the total roll-up across the whole country, like, nationally, for… I think we actually want to do, like, brand, yeah.

118 00:14:09.690 00:14:17.640 Michael Thorson: Yeah, we want to do this. There we go. So yeah, these would be, like, a set of filters, though, that we want to pull in to a final data mark.

119 00:14:18.110 00:14:24.560 Michael Thorson: So, yeah, I mean, at first glance, do you have any questions here? Or, like, from what you know about the data, are you, like…

120 00:14:25.090 00:14:30.880 Michael Thorson: is this… does this make sense, or is this, like, what… what are these fields? I haven’t really, like, gotten familiar with them yet.

121 00:14:32.860 00:14:48.719 Demilade Agboola: I will say, like, in terms of geography level, I do get that to a certain extent. I mean, I don’t necessarily know what every single geography level stands for. For instance, like, census Region or RMA, but I know, like, what all U.S. stands for. Reporting level, no, as well.

122 00:14:49.320 00:14:52.670 Demilade Agboola: But I think the idea here is…

123 00:14:53.460 00:14:56.839 Demilade Agboola: In terms of what the, like, final, like, mark name is…

124 00:14:57.270 00:15:00.330 Demilade Agboola: I get that. I think what I’m trying to…

125 00:15:00.890 00:15:08.910 Demilade Agboola: what I’m trying to understand now is, are we going to be doing this by, like, quantity? Are we doing this by, like, what metric are we looking at in terms of, like.

126 00:15:09.230 00:15:24.359 Demilade Agboola: the spin total US subcategory or brands. Do we want to have it in terms of, like, a flat table with all the data for those, like, this criteria? Or do we want to, like, roll it up to a certain level of granularity, and then we say, this is the revenue by week?

127 00:15:24.500 00:15:27.220 Demilade Agboola: For these categories, like, for this, like, filters.

128 00:15:28.490 00:15:33.170 Michael Thorson: Yeah, let me see if the right… yeah, I mean… the…

129 00:15:34.010 00:15:36.029 Michael Thorson: How do I… how do I put this?

130 00:15:37.700 00:15:45.490 Michael Thorson: this is kind of, like… I’m trying to think… so this is, like, yeah, here’s geography level. So this is, like, an example extract that we were working with, just to, like.

131 00:15:45.660 00:15:59.119 Michael Thorson: this is something we thought was, like, useful, so it’s like, we want to understand, like, in the US, like, what is the total dollar roll-up in, like, all regions, all stores, all, like, categories, and that’s why we, like, filtered on total geography level.

132 00:15:59.120 00:15:59.500 Demilade Agboola: Yeah.

133 00:15:59.500 00:16:11.420 Michael Thorson: like, total US geography level, and in this specific report, it’s kicking out just total US MULO, which is, like, top, top level, like, this is, like.

134 00:16:11.530 00:16:17.609 Michael Thorson: How much cereal was sold, for example, and granola in, like, All geographies together.

135 00:16:17.610 00:16:24.769 Demilade Agboola: There’s gonna be a couple more geographies that fall under this total US, and that’s okay, like, we want that. They should be…

136 00:16:24.770 00:16:31.179 Michael Thorson: From what we understand, like, they should all be, like, additive, like, they should be mutually exclusive, and we should be able to, like, aggregate.

137 00:16:31.180 00:16:32.329 Demilade Agboola: Comment up, yeah.

138 00:16:32.330 00:16:34.910 Michael Thorson: Yeah, exactly. And…

139 00:16:35.160 00:16:51.599 Michael Thorson: So I think this is, like, kind of our example of, like, a data mart. This is, like, for a leadership team to be like, how is Magic Spoon doing, like, across the whole country? And that’s, like, why we’re, like, filtering only on account type equals total U… or, like, sorry, geography levels total US.

140 00:16:53.080 00:16:54.110 Michael Thorson: Sweet.

141 00:16:54.670 00:16:55.810 Michael Thorson: So, yeah.

142 00:16:56.250 00:17:05.729 Demilade Agboola: And will we be doing this for just, like, Magic Spoon, or are we going to do this for across multiple brands, so that we can start to have, like, comparative, like, okay, so across…

143 00:17:06.319 00:17:14.580 Demilade Agboola: Magic Spoon, this is what we’re doing. Widabix is doing this, you know, like, we can start to have, like, those sort of comparison.

144 00:17:16.190 00:17:20.940 Michael Thorson: It’s a good call-out. There’s probably a use case for a data mart that’s, like.

145 00:17:21.170 00:17:37.189 Michael Thorson: Either, I’d say. Jt, do you have an opinion here? Like, I was thinking we could start with Magic Spoon, just start small, make sure we really understand, like, what we’re looking at, and then, like, we just go in later, and we can just, like, update the brand filter from Magic Spoon to brand equals any.

146 00:17:37.740 00:17:48.719 Michael Thorson: That’s kind of my thought. Like, kind of have an iterative design approach, where it’s like, let’s start small, keep it focused, and then, like, when we’re ready, we can move this from dbt, this filter from dbt, into our analytics layer.

147 00:17:50.910 00:17:52.729 Justin Tabarini: Yeah, I think, I think the…

148 00:17:52.870 00:17:56.709 Justin Tabarini: I agree with what Michael’s saying. Basically, let’s design it for all brands.

149 00:17:56.840 00:17:57.490 Michael Thorson: Right.

150 00:17:57.490 00:18:00.170 Justin Tabarini: That will QA against Magic Spoon.

151 00:18:00.470 00:18:11.019 Justin Tabarini: Okay. So all validation will be done against Magic Spoon, but we will build it in a way where it can deal with multiple brands. That make sense?

152 00:18:11.020 00:18:16.300 Demilade Agboola: Yeah, that makes sense. So, like, the goal is obviously all the brands, but then we’ll start off with Magic Spoon. Yeah.

153 00:18:18.900 00:18:19.730 Demilade Agboola: Okay.

154 00:18:20.290 00:18:28.609 Demilade Agboola: And do we have, like, a list of metrics we care for, across, like, Magic Spoon, or, like, we will be using to, you know, just look at?

155 00:18:28.710 00:18:44.580 Demilade Agboola: Or are we just going to be using… because Spain has a lot of data, that’s kind of why I’m asking. So there’s a lot of data in Spain. I want to know, like, what’s, like, what matters most to the business, what… how do we know that, like, this is what stakeholders need to see to make their decisions?

156 00:18:44.650 00:18:50.470 Demilade Agboola: Do we have, like, some sort of docs on that, or do we have an idea of what they look at on a day-to-day or a week-to-week basis?

157 00:18:50.910 00:19:07.009 Michael Thorson: Ashwini, me, and JT already did that on the upstream side, so whatever we’re pulling from spins, we’ve deemed as useful for the business. So just go ahead and, like, let’s pull in all available measures for now, and we’ll hide those when we, see fit.

158 00:19:07.280 00:19:08.890 Demilade Agboola: Okay, sounds good, sounds good.

159 00:19:09.750 00:19:11.340 Demilade Agboola: Okay,

160 00:19:14.580 00:19:16.969 Demilade Agboola: That makes a lot of sense.

161 00:19:19.030 00:19:23.809 Demilade Agboola: Also, in terms of modeling, again, we’ll get… once we get to that bridge, it becomes clearer.

162 00:19:24.480 00:19:29.359 Demilade Agboola: And this might be a bit, like, forward, but I’m just wondering, how much of it do we want to pass to Omni?

163 00:19:29.580 00:19:35.649 Demilade Agboola: to do, like, the modeling, or… and how much, like, pre-aggregated do we want to do with dbt?

164 00:19:36.710 00:19:40.580 Michael Thorson: Good question. We…

165 00:19:41.610 00:19:53.920 Michael Thorson: I think the biggest thing, the biggest open is, like, I think a lot of the modeling will just be pass the data through to Omni, and, like, we do a lot of the, kind of, table builds and, like, aggregation levels in

166 00:19:53.920 00:20:01.650 Michael Thorson: D in Omni, excuse me. The one thing that I think is an open question is how we want to handle period-over-period calculations.

167 00:20:01.770 00:20:09.840 Michael Thorson: I think a lot of this retail, it’s really important to, like, compare what sales this week versus the same sales… like, same week last year.

168 00:20:09.960 00:20:26.689 Michael Thorson: is a super common one. So that’s a… that’s an opportunity where I’m like, we might want to start to build in, like, a previous period, into our dataset, like, into the, like, final data mart. That’s kind of what I was thinking in terms of, like, CTEs, is, like.

169 00:20:27.520 00:20:31.650 Michael Thorson: one… I think that’s, like, probably the most common is, like, one year ago.

170 00:20:32.360 00:20:33.219 Demilade Agboola: Okay, sounds good.

171 00:20:33.220 00:20:36.500 Michael Thorson: So keep that in mind. I personally…

172 00:20:36.670 00:20:44.779 Michael Thorson: like, we… that period-over-period stuff we can also do in Omni, like, I think it’s fairly straightforward. So it’s like…

173 00:20:45.470 00:20:53.989 Michael Thorson: we can always build that into the data model DBT when we see a need for it, rather than, like, over-engineer it, now. But, JT, any thoughts there?

174 00:20:55.370 00:21:01.430 Justin Tabarini: Yeah, generally, I think it’s, like, data cleaning and dbt aggregation and Omni is the easiest way to do it.

175 00:21:01.930 00:21:04.780 Justin Tabarini: So, like, aligned with everything you just said, it’s just…

176 00:21:05.580 00:21:06.080 Michael Thorson: Yeah.

177 00:21:06.080 00:21:10.559 Justin Tabarini: let’s add some cleaning periods, maybe? Like, if we need to clean anything.

178 00:21:10.730 00:21:18.719 Justin Tabarini: Probably… Let’s build a mart at 1G level of… geography aggregation?

179 00:21:19.250 00:21:28.749 Justin Tabarini: Like, er… my bad, geography level, there we go, that’s the word. Like, we probably don’t want a mart to have two geography levels, or that can lead to, like, incorrect summing.

180 00:21:29.600 00:21:30.330 Demilade Agboola: Okay.

181 00:21:30.330 00:21:36.750 Justin Tabarini: So, like, we’ll have to… pre-filter, like, build a mart, that’s total US, build a mart, that’s RMA.

182 00:21:38.060 00:21:41.990 Justin Tabarini: But… the aggregation will happen in Omni.

183 00:21:42.320 00:21:45.039 Justin Tabarini: We’ll just need, like, a… we don’t want to give…

184 00:21:45.350 00:21:49.870 Justin Tabarini: put a dataset into Omni, where you can easily aggregate.

185 00:21:49.970 00:21:51.400 Michael Thorson: Incorrectly.

186 00:21:53.200 00:21:59.439 Michael Thorson: Yeah. And I think these filters, like, if you follow this logic, Demi, this should…

187 00:21:59.960 00:22:01.590 Justin Tabarini: Lead to form arts.

188 00:22:01.590 00:22:03.940 Michael Thorson: Yeah, exactly, and lead to, like, healthy, like…

189 00:22:04.580 00:22:08.810 Michael Thorson: Like, that can’t be, like, double counting certain regions or whatever.

190 00:22:08.810 00:22:09.640 Demilade Agboola: Yeah.

191 00:22:09.640 00:22:10.350 Michael Thorson: Yeah.

192 00:22:10.670 00:22:20.749 Demilade Agboola: No, no, I agree. I think even, like, which is part of why, even with the data that we currently have, we did the base layers, where we have, like, the region and brand Magic Spoon.

193 00:22:20.850 00:22:33.440 Demilade Agboola: and the region, and the region, census region, and the brand Magic Spoon, and we did another region of RME, our geography level as RME, and the brand Magic Spoon. So kind of already from…

194 00:22:33.620 00:22:45.600 Demilade Agboola: the base tables, we try to ensure that we’re splitting it out in such a way that, further down the line, we will run into the risk of aggregating it incorrectly.

195 00:22:45.890 00:22:46.870 Michael Thorson: Yeah, for sure.

196 00:22:47.490 00:22:54.500 Michael Thorson: Yeah, and I think these two, like, B and C, are, like, the key ones, because that’s where, like, the biggest mistakes can happen, I think.

197 00:22:54.640 00:23:01.410 Michael Thorson: There are a couple filters that, like, we can pass through to Omni. Like, TPL versus HWI, for example, is, like.

198 00:23:02.100 00:23:04.389 Michael Thorson: Magic Spoon will…

199 00:23:04.660 00:23:10.189 Michael Thorson: be, like, 100% sales will be in TPL, and 100% will be in HWI, which is interesting.

200 00:23:10.350 00:23:11.780 Michael Thorson: But we want…

201 00:23:11.780 00:23:12.689 Demilade Agboola: How does that happen?

202 00:23:13.040 00:23:18.300 Michael Thorson: We’ll force a filter in Omni. Basically, like, TPL and HWI are, like, what,

203 00:23:18.510 00:23:35.820 Michael Thorson: like, what, category is this brand in? So it’s, like, I think total product universe… total product, like, it’s, like, both health brands and, like, Cheerios, like, a more traditional, like, cereal company, for example. And then HWI is, like, health…

204 00:23:36.230 00:23:54.610 Michael Thorson: brands only. So, yeah, the list of brands in TPL will be larger than the list of brands in HWI. So, like, I think leaving this in the model is, like, important, because we want to have a toggle, like a forced filter in Omni, so, like, users can split back and forth.

205 00:23:55.000 00:23:56.370 Demilade Agboola: Gotcha, gotcha.

206 00:23:56.370 00:24:02.319 Michael Thorson: Yeah, so that’s why I, like, wanted to add, like, these are filters we want to actually leave in Omni,

207 00:24:02.960 00:24:06.959 Michael Thorson: Maybe, but, like, don’t filter out on the dbt side. Does that make sense?

208 00:24:06.960 00:24:09.790 Demilade Agboola: Yeah, yeah, that does make… that does make sense.

209 00:24:10.020 00:24:10.710 Michael Thorson: You’re good.

210 00:24:11.080 00:24:11.810 Michael Thorson: Cool.

211 00:24:14.040 00:24:14.680 Demilade Agboola: Okay.

212 00:24:15.060 00:24:21.840 Demilade Agboola: Right now, no other questions. I’m pretty sure once, like, once we start getting to it, like, questions will come up, as they tend to do.

213 00:24:22.000 00:24:24.550 Demilade Agboola: And then I’ll definitely be able to sync with you.

214 00:24:25.510 00:24:36.189 Michael Thorson: Oop, I’ll tag in this, you know where it is, and we can just kind of, like, let me know… feel free to punch in the deliverable dates that you think you can meet, just so we’re, like, aligned. I put in some, like.

215 00:24:36.750 00:24:45.439 Michael Thorson: You know, directional dates here, just in terms of, like, prioritization, but obviously these hinge on the other projects in the workstream.

216 00:24:46.120 00:24:47.650 Demilade Agboola: Okay, alright, sounds good.

217 00:24:49.700 00:24:52.780 Demilade Agboola: Can you just be sure I have access to it? Like, the share?

218 00:24:53.130 00:24:53.690 Michael Thorson: Yep.

219 00:24:54.350 00:24:59.040 Michael Thorson: Should be public… yeah, anyone can link, and I’ll just make sure it goes straight to your inbox, too.

220 00:25:01.780 00:25:02.480 Michael Thorson: Cool.

221 00:25:03.030 00:25:08.460 Demilade Agboola: Sounds good. I’ll be sure to take a look at this early next week, and I’ll get back to you.

222 00:25:09.520 00:25:15.720 Michael Thorson: Nice. Super excited to get some data in Omni and actually, like, start to… Mess around with some things.

223 00:25:15.720 00:25:16.190 Demilade Agboola: Yeah.

224 00:25:16.190 00:25:32.200 Michael Thorson: I’ll follow up with Ashwini on the ETL roadmap, and then see what his status is as well, because I think, like, once we have the larger timeframes in, that’ll shift the dynamics on this roadmap, and we’ll just keep adding in, like, different,

225 00:25:32.330 00:25:34.229 Michael Thorson: Marts here that we’re looking for.

226 00:25:35.080 00:25:36.440 Demilade Agboola: Okay, alright, sounds good.

227 00:25:37.120 00:25:43.420 Demilade Agboola: I will reach out to Ashwini as well, just to ask him to send, like, what was loaded, like, the current state of,

228 00:25:43.660 00:25:46.840 Demilade Agboola: data in Redshift right now,

229 00:25:47.180 00:25:49.329 Demilade Agboola: We can continue from next… from that next week.

230 00:25:49.940 00:25:50.590 Michael Thorson: Thanks.

231 00:25:50.590 00:25:51.640 Demilade Agboola: Alright, cool.

232 00:25:51.870 00:25:53.470 Michael Thorson: Making progress, love it.

233 00:25:53.470 00:25:54.690 Demilade Agboola: Alright, sounds good.

234 00:25:56.390 00:25:58.709 Demilade Agboola: Alright, see you guys next week. Take care.

235 00:25:58.710 00:25:59.609 Justin Tabarini: Next week?

236 00:25:59.610 00:26:00.530 Demilade Agboola: Alright, bye.