Meeting Title: Insomnia Knowledge Handoff Date: 2025-11-21 Meeting participants: Casie Aviles, Amber Lin
WEBVTT
1 00:04:11.790 ⇒ 00:04:12.970 Amber Lin: Hello!
2 00:04:13.670 ⇒ 00:04:14.980 Casie Aviles: Oh. Hey, Amber.
3 00:04:17.000 ⇒ 00:04:20.010 Amber Lin: Okay, how do we want to do this?
4 00:04:22.260 ⇒ 00:04:26.710 Casie Aviles: I think… Just, you know,
5 00:04:27.290 ⇒ 00:04:32.300 Casie Aviles: What are the kind of, like, the tasks that you want me to help with?
6 00:04:33.130 ⇒ 00:04:39.479 Casie Aviles: And then, like, I guess I just have some other questions as well on… how I would
7 00:04:39.600 ⇒ 00:04:49.039 Casie Aviles: you know, do the analysis, how you typically approach the analysis task, and… because since… Yeah, that’s kind of…
8 00:04:49.360 ⇒ 00:04:55.350 Casie Aviles: I don’t know, like, what’s the typical process? Like, where do I start? What do I need to…
9 00:04:56.370 ⇒ 00:05:00.370 Casie Aviles: Produce as an output, you know, just the high-level stuff.
10 00:05:00.830 ⇒ 00:05:07.100 Amber Lin: I see. I think we can start on something together right now.
11 00:05:07.260 ⇒ 00:05:16.000 Amber Lin: Right now, I think I’ve just sent them an analysis. I can walk you through how I approached that, and…
12 00:05:16.000 ⇒ 00:05:16.430 Casie Aviles: Yes.
13 00:05:16.430 ⇒ 00:05:18.809 Amber Lin: I can walk through the…
14 00:05:18.940 ⇒ 00:05:25.460 Amber Lin: type analysis that I’ve done so far, because I also don’t…
15 00:05:26.150 ⇒ 00:05:30.990 Amber Lin: exactly know what will be coming up, because right now, I just sent it in
16 00:05:32.260 ⇒ 00:05:35.250 Amber Lin: I just sent in this.
17 00:05:36.020 ⇒ 00:05:37.299 Amber Lin: Let me show you.
18 00:05:37.530 ⇒ 00:05:43.499 Amber Lin: So, we started out by looking at how does customers
19 00:05:44.040 ⇒ 00:05:59.049 Amber Lin: compare throughout the life cycle. It’s like, life cycle as in when they make their first purchase, and then they make their second purchase, because it is a cookie company, so it’s more simple, they just keep buying more cookies, so it’s more of who buys
20 00:05:59.860 ⇒ 00:06:09.639 Amber Lin: like, 3… who returns 3 times, 4 times, it’s like that. And then we look at, for each of these, say, stages.
21 00:06:09.790 ⇒ 00:06:13.950 Amber Lin: How do they behave differently? So this is…
22 00:06:15.530 ⇒ 00:06:24.610 Amber Lin: Like, how… what’s the conversion rate between the different stages? So, first to second order, second to third order, etc.
23 00:06:25.320 ⇒ 00:06:28.310 Amber Lin: I see. And then this is the time between…
24 00:06:29.310 ⇒ 00:06:30.870 Casie Aviles: the purchases…
25 00:06:31.080 ⇒ 00:06:36.439 Amber Lin: So… they make their first order. When do they usually make their second order?
26 00:06:37.050 ⇒ 00:06:41.810 Amber Lin: This is, like, good distribution of… Product, so it’s just…
27 00:06:41.980 ⇒ 00:06:46.500 Amber Lin: X percent of purchases are boxes, the other percent are this.
28 00:06:46.640 ⇒ 00:07:04.149 Amber Lin: So it’s mostly still, like, summary statistics. Like, these are all still descriptive. We’re not doing anything predictive here yet, so it’s all taking this data, and I asked Cursor to say, hey, help me
29 00:07:04.300 ⇒ 00:07:08.529 Amber Lin: find was the percentage of these
30 00:07:08.820 ⇒ 00:07:24.470 Amber Lin: products, so you just give it a prompt. And sometimes you need to point out, use this field, use that field, so you might need to explore the dataset first, which Cursor can also help with, but, I find that giving it the fields is helpful. So in this case.
31 00:07:24.560 ⇒ 00:07:40.960 Amber Lin: I gave it the product names, so a little list of product IDs, and then product types. And so then I asked her, so, hey, group all the purchases, assign them…
32 00:07:41.500 ⇒ 00:07:48.069 Amber Lin: based on… assign them types, based on a product ID, and then summarize what the percentage is.
33 00:07:49.110 ⇒ 00:07:49.500 Casie Aviles: Okay.
34 00:07:49.500 ⇒ 00:07:57.880 Amber Lin: And so that’s, like, that’s the type of prompt I gave it. And this one is a combination of this and, like, the different…
35 00:07:58.260 ⇒ 00:08:05.520 Amber Lin: order life cycles. So this is essentially what is the distribution between the… between the products?
36 00:08:05.900 ⇒ 00:08:13.399 Amber Lin: When they make their first order, on average, when they make the second order, so what changes in product type?
37 00:08:14.440 ⇒ 00:08:14.960 Amber Lin: So…
38 00:08:15.030 ⇒ 00:08:15.570 Casie Aviles: Hmm.
39 00:08:16.490 ⇒ 00:08:22.370 Amber Lin: This stalled pretty much the same prompt, it’s just also adding in… this is, like, over time.
40 00:08:22.650 ⇒ 00:08:30.169 Amber Lin: This is, over the different life cycles, so it’s like you’re slicing it in a slightly different way, and it gives you something different.
41 00:08:31.470 ⇒ 00:08:36.280 Casie Aviles: And then you gen… you create, like, recommendations based on what you find.
42 00:08:36.280 ⇒ 00:08:46.020 Amber Lin: Yeah, yeah, so you look at it, and you see, hmm, this is interesting, oh, this green one spiked here. I was like, oh, that’s… that’s unnorm, like, that’s…
43 00:08:46.120 ⇒ 00:08:58.710 Amber Lin: not normal, so you look at stuff that’s a little bit off, and you’re like, oh, if this is… like, if the data’s right, because sometimes the data’s wrong, then this means something, because that means, like.
44 00:08:58.710 ⇒ 00:09:14.570 Amber Lin: the boxes were okay, and then they spiked, so they probably promoted boxes pretty hard during that period, and then the customer behavior shifts. Similar here of, you see this is pretty flat, but then this is a big jump from 28 to 42.
45 00:09:14.570 ⇒ 00:09:20.380 Amber Lin: So, then I can conclude, well, first of all, only 28 Get their second.
46 00:09:21.370 ⇒ 00:09:27.900 Amber Lin: That’s not a lot of people. But when they do, you see that the curve here is pretty… like, the number’s pretty high, they’re pretty…
47 00:09:27.900 ⇒ 00:09:44.830 Amber Lin: Even so we can say, oh, once they do make their second cookie purchase, then they’re more likely to return to get the new ones. So, it’s just, once you see something, you can make sense of, oh, this might cause this, that might mean that. I think…
48 00:09:44.860 ⇒ 00:09:54.800 Amber Lin: Like, you work with AI a lot, like, those are… we’re just making educated guesses, and we can also always go back and verify them, but,
49 00:09:55.120 ⇒ 00:10:08.469 Amber Lin: Most of the times, the client wants to see… the client needs a guest, and then they can layer in what they’ve seen on the ground. They’re like, oh yeah, you’re right, like, we saw this before.
50 00:10:08.650 ⇒ 00:10:09.849 Casie Aviles: I see, yeah.
51 00:10:09.850 ⇒ 00:10:10.450 Amber Lin: Yeah.
52 00:10:12.070 ⇒ 00:10:19.490 Amber Lin: This is still very similar, this is, like, based on their first, second, third cookie of how many items they get each order.
53 00:10:19.680 ⇒ 00:10:25.929 Amber Lin: And it’s, and, like, the more stuff they get, they get less stuff, so…
54 00:10:26.050 ⇒ 00:10:42.030 Amber Lin: what does that mean? And it’s like, oh, when somebody does their first order, they might get a few things, they might try something out, but then in the future, when they’re… they’ve bought Insomnia, like, 5 or 6 times, they might just get.
55 00:10:42.630 ⇒ 00:10:49.789 Amber Lin: the cookie that they like. They just might get one. So that’s… that’s some insights on their customer behavior.
56 00:10:50.570 ⇒ 00:10:55.319 Amber Lin: And usually when it’s the first pass, I can only…
57 00:10:55.600 ⇒ 00:11:05.940 Amber Lin: I usually do these insights, and then I’m like, oh, that might be interesting, but it takes a… it takes another… it takes a different brain to make, like, this type of…
58 00:11:06.080 ⇒ 00:11:06.700 Amber Lin: Recommended.
59 00:11:06.700 ⇒ 00:11:07.570 Casie Aviles: of…
60 00:11:07.570 ⇒ 00:11:14.799 Amber Lin: this is what I see, but this is what you can do. But, what’s helpful is that if you just make these observations.
61 00:11:15.120 ⇒ 00:11:25.190 Amber Lin: And then you put it to ChatGBT, and then you say, hey, these are the insights, what are some actionable things that
62 00:11:25.750 ⇒ 00:11:36.130 Amber Lin: the client can do. Sometimes they’re bad, sometimes they’re okay, but if you get a list of options, you might find the best one, and
63 00:11:36.380 ⇒ 00:11:37.680 Amber Lin: Like, it’s…
64 00:11:37.700 ⇒ 00:11:40.120 Casie Aviles: It’s easier…
65 00:11:40.440 ⇒ 00:11:41.250 Amber Lin: Yeah.
66 00:11:41.450 ⇒ 00:11:47.580 Amber Lin: So I think that will make it easier, so you don’t have to worry about, oh, I can’t make recommendations, like…
67 00:11:47.740 ⇒ 00:12:03.149 Amber Lin: if you have the insights to back it up, like, it means something, but since I don’t… I’ve never done this before, so I don’t… don’t know what it means, but AI will know. Like, it has seen other people make conclusions before, so it will be more helpful.
68 00:12:03.690 ⇒ 00:12:09.890 Amber Lin: Like, say here, right, the… they buy less cookies when they’re returning.
69 00:12:10.080 ⇒ 00:12:25.459 Amber Lin: what does that mean for the client? Like, yes, they buy less cookies, but what can they do? Like, does that mean they’re doing something wrong that people are getting less, or is that something that the customers are just doing, and there’s nothing I can do about it? So, we can tell the… tell the client, say, hey.
70 00:12:26.050 ⇒ 00:12:40.329 Amber Lin: when they return, the average order value is smaller, because they just want to have one cookie. But you can work towards increasing that, so what can you do to make those
71 00:12:40.710 ⇒ 00:12:45.969 Amber Lin: returning order is a little bit bigger, so maybe you have,
72 00:12:46.850 ⇒ 00:13:02.019 Amber Lin: add-ons, so it’s like, buy more cookies for this, or upsell other cookies, so it’s like, there’s different ways to, help them make more money. It’s like, everything is, like, help them make more money, help them not lose money, so if you think about it that way, it’ll be…
73 00:13:02.020 ⇒ 00:13:06.810 Casie Aviles: Yeah, that’s pretty much the end goal, right? Yeah. Like, how they can make more revenue.
74 00:13:07.110 ⇒ 00:13:08.380 Casie Aviles: Yeah.
75 00:13:09.110 ⇒ 00:13:11.520 Amber Lin: And… Wink.
76 00:13:11.660 ⇒ 00:13:23.370 Amber Lin: once I looked at this, right, I saw this graph, I saw that the boxes went up, I looked at this graph, I saw that, okay, there seems to be more classic cookies.
77 00:13:23.370 ⇒ 00:13:41.800 Amber Lin: as you go, slightly less boxes, but not that much less. So I said, okay, let’s, separate this out a little bit further, see if… because there’s different types of cookies, there’s different types of boxes. Yeah. So looking at the types of boxes, and you can see, oh,
78 00:13:42.980 ⇒ 00:13:53.210 Amber Lin: like, I summarized the box names into 6 packs, 9 packs, 12 packs, they have quite a few. And so you can see, oh, they get…
79 00:13:53.900 ⇒ 00:13:58.099 Amber Lin: When they first order, most of them get 6 packs.
80 00:13:58.250 ⇒ 00:13:58.920 Casie Aviles: Hmm.
81 00:13:58.920 ⇒ 00:14:06.050 Amber Lin: And then later on, they… They’re more open to getting bigger ones, so 9 packs and 12 packs.
82 00:14:06.140 ⇒ 00:14:25.529 Amber Lin: So, that was my first observation, and then, and then you can think about, okay, why… why might that happen? So, first is you notice a trend, and then you try to find a reason for that trend. It’s like, oh, why do they get six-packs when they first…
83 00:14:25.690 ⇒ 00:14:39.039 Amber Lin: When they first order it. Yeah. And it’s because when you look at their website, they have free delivery. And 6 packs is, like, 23, free delivery is $20.
84 00:14:39.130 ⇒ 00:14:56.230 Amber Lin: So, that’s why they would choose this, but if they’re returning, and this doesn’t count people who get classic cookies. I think when they’re returning, some of them go get individual cookies, right? And those who don’t, those who still buy boxes.
85 00:14:56.510 ⇒ 00:15:04.890 Amber Lin: probably… like, lies the value of boxes, and they’re… they’re starting to get bigger boxes. I think that’s the…
86 00:15:05.040 ⇒ 00:15:07.669 Amber Lin: Like, that’s the narrative, like, I think.
87 00:15:07.670 ⇒ 00:15:08.090 Casie Aviles: boyfriend.
88 00:15:08.090 ⇒ 00:15:23.100 Amber Lin: rubber means, tell the story, is what it means. Oh, I come onto the website, I see there’s $20 free delivery. I was like, okay, I’ll get a six-pack. And then if… if I come back again, I say,
89 00:15:23.430 ⇒ 00:15:35.579 Amber Lin: I don’t want an individual cookie because I think the boxes are good, I might buy a bigger pack. So that’s the story there, and then we can make the recommendation and say, hey, hey, how do you make people
90 00:15:35.730 ⇒ 00:15:38.139 Amber Lin: Buy bigger packs earlier.
91 00:15:38.250 ⇒ 00:15:43.350 Amber Lin: Or, how do you… Because that’s more money for them.
92 00:15:43.640 ⇒ 00:15:46.500 Amber Lin: And that means, okay, you can show…
93 00:15:46.990 ⇒ 00:16:05.559 Amber Lin: on the website, you can say, like, try to really generate this. I didn’t… I didn’t know this was possible. You can show returning customers a different thing on the website, because you can change that, and so that might make them get the cookie. Let me see if I can show you.
94 00:16:06.660 ⇒ 00:16:07.390 Casie Aviles: Yeah, it looks like…
95 00:16:07.390 ⇒ 00:16:07.830 Amber Lin: Yeah.
96 00:16:07.830 ⇒ 00:16:09.750 Casie Aviles: It’s really about, like, tracing…
97 00:16:10.170 ⇒ 00:16:15.760 Casie Aviles: putting yourself in the shoes of the customer, I guess, and using the data that we have.
98 00:16:16.770 ⇒ 00:16:24.170 Amber Lin: So you go here, and then, like, they have this, like, free delivery. They have this banner.
99 00:16:25.390 ⇒ 00:16:26.170 Amber Lin: That…
100 00:16:26.410 ⇒ 00:16:33.180 Amber Lin: Makes, like, that looks interesting, looks tasty, and then some of them have the packs on the top, some of them…
101 00:16:33.290 ⇒ 00:16:35.020 Amber Lin: I like this.
102 00:16:35.570 ⇒ 00:16:39.500 Amber Lin: So, similar thing for the cookies, it was like.
103 00:16:39.750 ⇒ 00:16:46.750 Amber Lin: what cookies do people get? Because if we just say people get more classic cookies, it’s harder for them to…
104 00:16:46.750 ⇒ 00:17:05.549 Amber Lin: know what to recommend, because, in an… like, if I were to receive an email from Sonya Cookies, like, if they tell me, go get a classic cookie, it’s harder for me to want to do that, but instead, if they say a very specific cookie, I might like that flavor, I might go.
105 00:17:06.050 ⇒ 00:17:19.590 Amber Lin: And so here, we look at, for the first, second, third, fourth, like, for these orders, let me… and I mapped it out, and then you can see something’s interesting here, right? You can see this is the biggest chunk.
106 00:17:20.220 ⇒ 00:17:35.959 Amber Lin: And I was like, oh, what’s that? What’s the one that people get the most on their first order? And that’s chocolate chunk. I was like, oh, that’s very interesting, because that’s a… that’s the base… most basic flavor ever. And that means, okay, people, on their first order.
107 00:17:36.120 ⇒ 00:17:39.929 Amber Lin: like to get more basic things. And…
108 00:17:41.520 ⇒ 00:17:42.270 Casie Aviles: Yeah.
109 00:17:42.270 ⇒ 00:17:59.929 Amber Lin: Probably the safer option. Yeah, and then you can see something increases here. It’s like, what are these that grow? Because they are so small when they started, but seems like returning customers like these things, and you can look, here at the stuff that increased, like.
110 00:17:59.950 ⇒ 00:18:05.639 Amber Lin: So this is another trend, right? The trend is like, oh, these more unique flavors
111 00:18:06.010 ⇒ 00:18:25.189 Amber Lin: they look more interesting. Like, cookies and cream, this very long name, another very long name, as a caramel apple, these all look interesting. And that tells us, like, okay, when people come back again, they don’t want the basic flavor, because they don’t get the chocolate chunk anymore, but they get the more interesting stuff.
112 00:18:25.910 ⇒ 00:18:26.660 Casie Aviles: Yeah.
113 00:18:26.870 ⇒ 00:18:36.179 Amber Lin: So what does that tell us? If they… if they… when they return, they like more interesting stuff? Like, what does that mean?
114 00:18:36.380 ⇒ 00:18:40.100 Amber Lin: like, how can we use that to get more money? Like, what do you think?
115 00:18:42.950 ⇒ 00:18:49.110 Casie Aviles: Oh, I guess it’s something that we can…
116 00:18:50.820 ⇒ 00:18:56.919 Casie Aviles: what’s the right term? Market more, or sell more during when they’re returning?
117 00:18:57.160 ⇒ 00:18:58.570 Casie Aviles: Yeah, so…
118 00:18:58.670 ⇒ 00:19:01.669 Amber Lin: Totally. That’s… that’s exactly the thing, because…
119 00:19:02.080 ⇒ 00:19:06.249 Amber Lin: Remember this? They have problems getting people to come back.
120 00:19:06.730 ⇒ 00:19:07.170 Casie Aviles: Yeah.
121 00:19:07.170 ⇒ 00:19:17.009 Amber Lin: only get 20%, and they only get 40% out of that, and then 40% out of that. So each one of them, they’re losing a lot down the funnel.
122 00:19:17.290 ⇒ 00:19:18.599 Casie Aviles: So is that… I see.
123 00:19:18.730 ⇒ 00:19:28.719 Amber Lin: Opposite triangle, they’re losing a lot of people, and anything they can do to make that funnel a bit wider to keep more people is a lot more money for them.
124 00:19:28.720 ⇒ 00:19:30.829 Casie Aviles: So that means, like, if you can…
125 00:19:31.230 ⇒ 00:19:36.019 Amber Lin: get people to come back, you know what they like, and you can say, oh, I want these
126 00:19:36.310 ⇒ 00:19:43.250 Amber Lin: new flavors, and they will come back, then it means that you might earn a lot more money by doing that. So that is.
127 00:19:43.250 ⇒ 00:19:44.400 Casie Aviles: Yeah, I see, yeah, yeah.
128 00:19:44.400 ⇒ 00:19:45.670 Amber Lin: recommendation there.
129 00:19:46.760 ⇒ 00:19:58.170 Amber Lin: And that’s, like, that’s the… that’s, like, that’s the story. There’s, like, what are the products, what do people want when they come back? What do they people want when they first purchase?
130 00:19:58.850 ⇒ 00:20:03.310 Amber Lin: And we said, Daylight Basics,
131 00:20:03.630 ⇒ 00:20:07.789 Amber Lin: They like 6-packs boxes. When they return, they like…
132 00:20:07.990 ⇒ 00:20:14.560 Amber Lin: Most of the times, they just get one thing, they’re pretty okay with the basics, but if there’s something cool, because these don’t…
133 00:20:14.990 ⇒ 00:20:27.910 Amber Lin: I checked their website, and these are not always on. Like, sometimes they’re limited, so that means that if there’s something cool, then people will get it, and that, like, we have their recommendations over here.
134 00:20:28.960 ⇒ 00:20:34.820 Casie Aviles: Yeah, yeah, that makes sense. Like, even when I buy something, I would…
135 00:20:35.140 ⇒ 00:20:39.800 Casie Aviles: When I return, I would like something that’s interesting.
136 00:20:40.270 ⇒ 00:20:40.590 Amber Lin: Yeah.
137 00:20:40.590 ⇒ 00:20:43.119 Casie Aviles: I wouldn’t get the same thing over and over again.
138 00:20:43.470 ⇒ 00:20:44.100 Amber Lin: Yeah.
139 00:20:44.570 ⇒ 00:21:00.029 Amber Lin: That’s… and a lot of the things, like, when you find it, you feel like, oh, this is common sense, but I think for the… for the business, it’s still something more interesting, if you can show, like, a trend that’s… they haven’t seen before. They’re like, oh, this is cool.
140 00:21:00.320 ⇒ 00:21:04.149 Amber Lin: And something… this is what I did today, so…
141 00:21:04.430 ⇒ 00:21:11.930 Amber Lin: This is essentially a repeat of what I did before, but specifically, Focusing, again, on a…
142 00:21:12.020 ⇒ 00:21:24.509 Amber Lin: spike in a different trend, so I looked at it as I saw. In June, they started getting more boxes, so I just compared before and after June, what was the… what was the difference?
143 00:21:24.510 ⇒ 00:21:33.130 Amber Lin: And then you can see conversion rates dropped on the first, second, and second, and third, so the earlier conversion rates dropped.
144 00:21:33.480 ⇒ 00:21:42.700 Amber Lin: What does that mean for them? What can they do about it? And they also had good news, like, the time to purchases shrank significantly.
145 00:21:43.500 ⇒ 00:21:50.319 Amber Lin: Because that’s 100 days, and then this is… Like, this is the rest.
146 00:21:51.360 ⇒ 00:22:08.039 Amber Lin: So, these are also, like, interesting insights for them. It tells them what’s working, because they might have just promoted boxes, because it’s more… the boxes are a bigger purchase, but they might not know that, okay, customers who get boxes actually
147 00:22:08.340 ⇒ 00:22:18.619 Amber Lin: Purchase more frequently, and return sooner, and that means a lot to them, because the more you return, the faster you return, like, the more purchases you make, and
148 00:22:18.900 ⇒ 00:22:22.060 Amber Lin: The, like, the more value they make off of them.
149 00:22:22.310 ⇒ 00:22:26.720 Amber Lin: So, like, that’s another thing that I found today.
150 00:22:27.600 ⇒ 00:22:28.590 Amber Lin: Okay.
151 00:22:28.590 ⇒ 00:22:29.980 Casie Aviles: Yeah, that’s very helpful.
152 00:22:30.270 ⇒ 00:22:37.949 Amber Lin: Yeah, and I can show you my cursor. So there’s a few steps. I know you already have cursor set up, right?
153 00:22:38.290 ⇒ 00:22:41.890 Casie Aviles: Yeah, so you’re using, like, Python notebooks here, right?
154 00:22:42.180 ⇒ 00:22:58.410 Amber Lin: Yeah, I’m using Python notebooks, and it asks me to write SQL and, like, whatever, because, like, I can read SQL, I can’t really read Python, but it doesn’t matter, because I first, I used this, let’s see…
155 00:22:59.390 ⇒ 00:23:08.549 Amber Lin: I think you know this better than I am. I have data folder, I have the notebooks, I try to separate them, I’m putting it all in the Brainforge platform, I have not.
156 00:23:08.740 ⇒ 00:23:15.260 Amber Lin: submitted a PR for a while, so only this one is synced. But I have that…
157 00:23:15.750 ⇒ 00:23:27.429 Amber Lin: And, so I connected to Mother Duck using a CLI, and then I asked it to, okay, first, let’s start with the first one I did for Insomnia.
158 00:23:29.810 ⇒ 00:23:32.550 Casie Aviles: And you’re using Braze data here, right?
159 00:23:33.860 ⇒ 00:23:36.879 Amber Lin: Whatever’s in… yeah, whatever’s in here.
160 00:23:37.310 ⇒ 00:23:38.040 Casie Aviles: Okay.
161 00:23:38.460 ⇒ 00:23:39.030 Amber Lin: Yeah.
162 00:23:40.620 ⇒ 00:23:41.580 Amber Lin: So…
163 00:23:41.580 ⇒ 00:23:45.479 Casie Aviles: And we aren’t using FDA data yet, right?
164 00:23:45.790 ⇒ 00:23:46.850 Casie Aviles: Or…
165 00:23:46.850 ⇒ 00:23:50.119 Amber Lin: Is that in Braze? I don’t know if we have it.
166 00:23:50.490 ⇒ 00:23:51.600 Casie Aviles: Yeah, I don’t catch that.
167 00:23:51.900 ⇒ 00:23:55.849 Casie Aviles: I don’t think it’s in Braze. Those were, like, the Uber…
168 00:23:56.280 ⇒ 00:23:56.870 Amber Lin: Yeah, I guess so.
169 00:23:56.870 ⇒ 00:23:57.490 Casie Aviles: DoorDash.
170 00:23:57.490 ⇒ 00:24:12.260 Amber Lin: I’ve never seen it before, so that… I think that’s the next step, because, Matt, who handles FDA, has something that he wants us to help with. I think that’s a pretty easy cursor assignment you can… you can start with, because…
171 00:24:12.260 ⇒ 00:24:12.720 Casie Aviles: Okay.
172 00:24:12.720 ⇒ 00:24:17.830 Amber Lin: I connect, I ask it to explore, I copy this name, because I explored this.
173 00:24:17.970 ⇒ 00:24:23.469 Amber Lin: Dataset, and it tells me… Alright, we can look at the new one.
174 00:24:24.060 ⇒ 00:24:32.120 Amber Lin: I did… Let’s see… Gray.
175 00:24:37.480 ⇒ 00:24:38.120 Amber Lin: Boom.
176 00:24:39.180 ⇒ 00:24:43.979 Amber Lin: And then load the column name so that you know what the values look like.
177 00:24:44.110 ⇒ 00:24:46.180 Amber Lin: You know what the column…
178 00:24:46.300 ⇒ 00:24:52.020 Amber Lin: field names are, so that’s helpful. After that,
179 00:24:52.500 ⇒ 00:25:08.070 Amber Lin: essentially what you do when you make Andy or make those stuff is just one step at a time, helping guide his thinking, so that it doesn’t jump steps. I don’t think I need to guide you on AI, but I think I can…
180 00:25:09.080 ⇒ 00:25:11.010 Amber Lin: Give some here.
181 00:25:11.490 ⇒ 00:25:17.869 Amber Lin: A lot of it is telling it to give me per… the count, and then percentage of…
182 00:25:17.910 ⇒ 00:25:35.659 Amber Lin: percentages of things, either by type, either over time, or either sometimes by cohort of people who joined at different times. You can look at it by day, by month.
183 00:25:36.290 ⇒ 00:25:38.359 Amber Lin: And then, usually, I ask it to…
184 00:25:38.560 ⇒ 00:25:45.850 Amber Lin: do this, and then ask it to give me, like, some ex… like, this, it automatically gave me, so it thinks…
185 00:25:46.200 ⇒ 00:26:01.739 Amber Lin: Slightly ahead of me, so it gives you some examples, and then you can also either copy this data and then put it in another cell to help it… to tell it to visualize it, or just say, also add a visualization, but it slows it down a little bit.
186 00:26:01.990 ⇒ 00:26:03.600 Amber Lin: Oh.
187 00:26:03.740 ⇒ 00:26:06.659 Amber Lin: And then… Let’s see…
188 00:26:07.000 ⇒ 00:26:14.449 Amber Lin: Yeah, so it gives me more visualizations than I need, but you can pick, oh, this looks interesting.
189 00:26:14.980 ⇒ 00:26:23.619 Amber Lin: Of what campaigns are most efficient. So you’re essentially slicing up things from different angles, and then you can
190 00:26:23.700 ⇒ 00:26:35.969 Amber Lin: Once you do the first pass of slicing it by one thing, you can start slicing it by two things. You can say, okay, now I want to slice it by different campaign types, and…
191 00:26:35.970 ⇒ 00:26:54.290 Amber Lin: over days of week. So you have… you can look at two things, and that might tell you, oh, this thing performs better on Sunday, versus that campaign type performs better on Wednesday. So those are just different ways to look at it, and you just look at where does this spike
192 00:26:54.310 ⇒ 00:26:55.799 Amber Lin: What is a fall?
193 00:26:56.710 ⇒ 00:27:02.209 Amber Lin: what’s the difference? So these are, like, the interesting parts where you can point out.
194 00:27:02.470 ⇒ 00:27:03.990 Casie Aviles: Yeah.
195 00:27:03.990 ⇒ 00:27:11.289 Amber Lin: And then the only caution is that sometimes you need to check… if it’s doing percentages, you need to check the count.
196 00:27:11.310 ⇒ 00:27:25.219 Amber Lin: Because there might be only one data point there. So, always do a verification, ask it to… ask it, what fields are you actually using, and how many data points are there. So those are things you need to go back and check.
197 00:27:25.300 ⇒ 00:27:30.710 Amber Lin: To make sure, like, this graph is actually valid. Because I think there’s only…
198 00:27:31.280 ⇒ 00:27:36.310 Amber Lin: This spike, there’s probably only, like, a few data points there.
199 00:27:36.440 ⇒ 00:27:38.009 Amber Lin: So that’s not helpful.
200 00:27:39.510 ⇒ 00:27:45.439 Casie Aviles: Okay, yeah, this is helpful. This is all in the repository, right? I can take a look.
201 00:27:45.660 ⇒ 00:27:51.099 Amber Lin: I will… yes, I will submit these PRs right now.
202 00:27:51.100 ⇒ 00:27:51.970 Casie Aviles: Okay.
203 00:27:52.100 ⇒ 00:27:53.400 Casie Aviles: Thank you, thank you.
204 00:27:53.690 ⇒ 00:27:57.130 Amber Lin: Okay, let me see, I haven’t done that.
205 00:27:57.700 ⇒ 00:27:59.020 Amber Lin: In a bit.
206 00:28:00.160 ⇒ 00:28:04.760 Amber Lin: Cool.
207 00:28:07.960 ⇒ 00:28:15.610 Amber Lin: Yeah, and then I think the next task that is coming in, we’re waiting on Matt to give us…
208 00:28:16.050 ⇒ 00:28:24.009 Amber Lin: So the context here, they’re cutting spending on some, like, Uber Eats, or they’re cutting spending on Uber Eats.
209 00:28:24.010 ⇒ 00:28:24.420 Casie Aviles: Yeah.
210 00:28:24.420 ⇒ 00:28:37.670 Amber Lin: on some stores, they said, okay, we don’t want to spend… we spent $0 there, but they selected it at random, and I think they don’t know what stores are actually being cut in spending, so he.
211 00:28:37.670 ⇒ 00:28:38.190 Casie Aviles: wants us to…
212 00:28:38.190 ⇒ 00:28:51.030 Amber Lin: look at the spend data and to see it by store, because he only sees it in aggregate. He wants to see it by store, and see if it’s affecting revenue. So I think that should be pretty easy to start with, just…
213 00:28:51.500 ⇒ 00:28:54.669 Amber Lin: Check the… check the, like, spending data.
214 00:28:55.050 ⇒ 00:28:55.480 Casie Aviles: You’re finished.
215 00:28:55.480 ⇒ 00:28:59.610 Amber Lin: Source equals zero, and then check if the revenue has changed compared to before.
216 00:29:00.490 ⇒ 00:29:07.460 Amber Lin: So, once he gives us the dataset, or if we can download it from somewhere, like, that’s something we can start with.
217 00:29:07.720 ⇒ 00:29:14.420 Amber Lin: Matt said… so Matt still hasn’t sent it to us. So, no rush there.
218 00:29:15.010 ⇒ 00:29:22.669 Casie Aviles: Yeah, okay, I mean, I believe I can export some data, I’m not… I’m just not sure if that will be all of it, all of what we need, but…
219 00:29:22.880 ⇒ 00:29:28.560 Casie Aviles: Yeah, I can take a look and just, do some exploration, initial exploration.
220 00:29:28.640 ⇒ 00:29:35.030 Amber Lin: Cool. And you can always add data here, and then copy the path asset to use that data set… data.
221 00:29:35.030 ⇒ 00:29:35.490 Casie Aviles: No way.
222 00:29:35.490 ⇒ 00:29:50.319 Amber Lin: Yeah, so that’s pretty easy. Cool. Let me submit these, so you can go in and check them. Some of them have errors, because sometimes it takes too long to fix them, so I just move on to the next one.
223 00:29:50.900 ⇒ 00:29:52.140 Casie Aviles: Yeah, no problem.
224 00:29:52.140 ⇒ 00:29:53.110 Amber Lin: Cool, okay.
225 00:29:54.290 ⇒ 00:29:56.760 Casie Aviles: Okay, yeah, that’s helpful. Thank you, Amber. Of course.
226 00:29:56.760 ⇒ 00:29:57.430 Amber Lin: Of course.
227 00:29:57.810 ⇒ 00:30:00.560 Casie Aviles: If I have any questions, I’ll just, yep, let you know.
228 00:30:00.560 ⇒ 00:30:05.529 Amber Lin: Ping me, ping me in the chat, and then when Matt gives the data set, I’ll let you know.
229 00:30:05.920 ⇒ 00:30:07.360 Casie Aviles: Okay, thank you.
230 00:30:07.360 ⇒ 00:30:08.979 Amber Lin: Cool, right, bye.
231 00:30:08.980 ⇒ 00:30:09.680 Casie Aviles: Bye-bye.