Meeting Title: Dashboard-UX Date: 2024-02-12 Meeting participants: Uttam Kumaran, Bencohen, Daniel Schonfeld, Patrick Trainer
WEBVTT
1 00:02:19.830 ⇒ 00:02:20.800 Patrick Trainer: Yo.
2 00:02:20.920 ⇒ 00:02:21.880 Uttam Kumaran: you know.
3 00:02:22.640 ⇒ 00:02:37.070 Patrick Trainer: I had to click off like at Zoom just updated and had to click off like all the thousands of like. Look, we’ve got a new chat box just like you’ve had a fucking chat box every time you open this app. It’s like.
4 00:02:37.160 ⇒ 00:02:40.289 Patrick Trainer: I don’t. I just I just don’t know
5 00:02:40.700 ⇒ 00:02:44.529 Uttam Kumaran: what’s better, because it’s a desktop client. But why do they
6 00:02:44.790 ⇒ 00:02:51.630 Uttam Kumaran: ship updates every hour like I don’t get it. It’s kind of like annoying like
7 00:02:52.000 ⇒ 00:02:54.239 Uttam Kumaran: I get it. You’re making updates, but
8 00:02:55.020 ⇒ 00:03:14.789 Patrick Trainer: just takes up so much
9 00:03:14.990 ⇒ 00:03:18.760 Uttam Kumaran: of my browser memory, and it never
10 00:03:18.780 ⇒ 00:03:28.719 Patrick Trainer: so so gpu acceleration or something. The trick is to use. So there’s a standalone app
11 00:03:28.960 ⇒ 00:03:48.370 Patrick Trainer: and like you can, it’s kind of like, like, how you can use Google services and standalone apps. It’s like like in like an applet, almost. And so it takes it takes it out of the browser. It’s I guess it’s technically still in its own little browser, but it’s it’s separate from, like
12 00:03:48.530 ⇒ 00:03:54.060 Patrick Trainer: your main tabs, and everything like you have a dedicated thing, and that’s
13 00:03:54.200 ⇒ 00:04:01.569 Patrick Trainer: seemed to help me. It also. It’s like you never lose the little red.in the sea of tabs like, that’s annoying.
14 00:04:01.760 ⇒ 00:04:02.769 Uttam Kumaran: I see.
15 00:04:04.600 ⇒ 00:04:07.730 Patrick Trainer: Yeah, no, I think the the qualities
16 00:04:07.950 ⇒ 00:04:13.899 Patrick Trainer: much better. Yeah, II, this is just like
17 00:04:14.060 ⇒ 00:04:24.579 Uttam Kumaran: it’s just needs. This is like a needs to work type of thing, you know. And it’s also like, yeah. And I, just yeah, I think sometimes.
18 00:04:30.160 ⇒ 00:04:31.030 Uttam Kumaran: hey, Ben.
19 00:04:35.960 ⇒ 00:04:38.149 bencohen: hey, guys, how’s it been?
20 00:04:38.870 ⇒ 00:04:50.069 bencohen: Can you hear me? Oh, perfect one sec. Let me turn off all of my telegram and all these other Jesus.
21 00:04:56.200 ⇒ 00:04:58.499 bencohen: Alright, I think I should be good.
22 00:04:59.410 ⇒ 00:05:02.880 Patrick Trainer: Nice to meet you, Patrick. Yeah. Likewise, how are you doing?
23 00:05:03.740 ⇒ 00:05:06.829 bencohen: I definitely can’t complain. It was a nice weekend.
24 00:05:07.780 ⇒ 00:05:12.649 bencohen: and really all good. That’s great. Where? Where are you?
25 00:05:12.670 ⇒ 00:05:14.749 bencohen: I’m in Westport, Connecticut.
26 00:05:14.900 ⇒ 00:05:15.850 Patrick Trainer: Okay.
27 00:05:16.260 ⇒ 00:05:20.499 Patrick Trainer: little bit north. I’m in New Orleans.
28 00:05:20.530 ⇒ 00:05:21.950 bencohen: Oh, nice!
29 00:05:22.580 ⇒ 00:05:29.009 Patrick Trainer: That sounds much more fun than I’m having. It’s Marty Mardi Gras in kind of like peak
30 00:05:29.110 ⇒ 00:05:46.569 Patrick Trainer: peak times right now it’s been going on for I mean the past 2 months. But like, this is kind of like the the apex. Yeah, yeah, it hasn’t stopped. But this is, this is definitely like the peak. So
31 00:05:48.140 ⇒ 00:05:51.120 bencohen: sounds pretty good. Was there
32 00:05:51.900 ⇒ 00:06:04.060 bencohen: right around Mardi Gras? Once it was okay, it was good enough for me it was. It was. It was great. I didn’t need my, I didn’t need any more convincing
33 00:06:04.280 ⇒ 00:06:17.950 Patrick Trainer: I that’s so. Every time I’ve had this conversation with people they like, I always get something similar of like, Oh, yeah, I went to Mardi. Gras did that like it was crazy. Yeah, it’s it’s I love those stories
34 00:06:18.670 ⇒ 00:06:19.840 bencohen: definitely.
35 00:06:19.850 ⇒ 00:06:26.670 bencohen: definitely alright. Well, let’s you obviously know, Tom, who we like a lot. So any friend of his
36 00:06:27.210 ⇒ 00:06:29.389 bencohen: probably gonna be a friend of ours
37 00:06:29.600 ⇒ 00:06:30.710 Patrick Trainer: awesome.
38 00:06:30.740 ⇒ 00:06:43.330 Uttam Kumaran: I don’t know about that.
39 00:06:43.550 ⇒ 00:07:12.539 Uttam Kumaran: That takes a good friend of mine, and also is extremely talented. A rare scale that I feel like even in my years of working with a lot of data. People like a lot of engineers aren’t very good at and you know me and Pat talk a little bit about 2 weeks ago. Kind of the stuff that we’ve been doing, and I think we just really impressed me with a lot of his work so far in like information, architecture, dashboarding. And I think what we did over the last 2 weeks was kind of just
40 00:07:12.560 ⇒ 00:07:22.199 Uttam Kumaran: through the existing dashboards and give feedback. And I’ll kind of let him run and give a little bit of his background, and then just jump right into kind of that exercise, and then.
41 00:07:22.400 ⇒ 00:07:23.809 Uttam Kumaran: you know, we could just go from there.
42 00:07:25.550 ⇒ 00:07:31.030 Uttam Kumaran: He has contacts, too, on, you know, the whole business and everything. So I think we can just run right in.
43 00:07:31.220 ⇒ 00:07:55.650 Patrick Trainer: Awesome. Yeah. So like I, said Patrick. Down here in New Orleans, I’ve been working as a data engineer for few different companies for I mean the past number of years, kinda all my, all my career. But so my kind of like journey and data, it’s like it is started kind of like in in high school, then into college did econometrics, and then
44 00:07:55.670 ⇒ 00:08:13.199 Patrick Trainer: grad school was all industrial engineering and supply chain, and so that kind of like gave me this very like process, focus, kind of mindset. And in that process, focus becomes like a lot of like
45 00:08:13.200 ⇒ 00:08:26.350 Patrick Trainer: how do you tell the story, and as little amount of like space as possible, just because of like the nature of these businesses are so kinda like information dense it can get super overloading.
46 00:08:26.500 ⇒ 00:08:40.629 Patrick Trainer: So I’ve like, I said, I’ve kind of run the gamut in the data sphere, going from like traditional engineering to kind of like data science, and like forecasting all the way
47 00:08:40.799 ⇒ 00:08:57.739 Patrick Trainer: onto the other side of the spectrum in like dashboarding, visualization and kind of like information architecture it. That is like, I have a very technical brain, but also like this, scratches that like artistic itch as well. But it’s it’s
48 00:08:57.760 ⇒ 00:09:00.069 Patrick Trainer: a a lot of.
49 00:09:00.110 ⇒ 00:09:07.250 Patrick Trainer: There’s a lot of small nuance to it that that makes like the small things good so that’s that’s kinda where I’m coming from
50 00:09:08.360 ⇒ 00:09:19.579 Patrick Trainer: awesome great background. Thank you. Cool, cool, and so I think what we can do. I can share this screen pretty soon. But I went through
51 00:09:19.840 ⇒ 00:09:25.110 Patrick Trainer: what the 4 main kind of like dashboards and views
52 00:09:25.150 ⇒ 00:09:27.469 Patrick Trainer: that y’all are.
53 00:09:27.520 ⇒ 00:09:30.029 Patrick Trainer: Sorry I’m looking for the
54 00:09:30.580 ⇒ 00:09:33.390 Patrick Trainer: share button. But
55 00:09:37.580 ⇒ 00:09:38.280 sorry
56 00:09:38.890 ⇒ 00:09:41.539 Daniel Schonfeld: did you make these new ones the
57 00:09:43.140 ⇒ 00:09:54.119 Daniel Schonfeld: So when I go to the the main dashboard, there’s 3 new ones links to other dashboard. Is that you, Tom? Or is that Patrick? Yeah. So the all the ones that existed are the ones that that I’ve done.
58 00:09:54.170 ⇒ 00:10:04.909 Uttam Kumaran: But what I pretty much had Patrick uses is, hey? Take a look at all the existing dashboards. Here’s all the context about the business. Just go through each, and pretty much just like run and audit
59 00:10:04.950 ⇒ 00:10:10.160 Uttam Kumaran: And you know, I think I when I when I saw, I think what Patrick’s gonna share today.
60 00:10:10.230 ⇒ 00:10:32.739 Uttam Kumaran: one, I was like, yeah, I’m I’m definitely not the best at this, but at the same time, when you see someone go through with the finds with comb and kind of understand the information display. But also, I think you really clearly put together, how do we jump to what’s actually this visualization as part of this larger dash force trying to translate. I thought it was, you know, really cool. So I think, maybe, Pat, you just want to run through one.
61 00:10:32.850 ⇒ 00:10:51.749 Patrick Trainer: Yeah, yeah. Yeah. So as an example here. Like, I was going through this kind of like your as I understand it, kind of like your daily view. Right? Is that good? And you’ll see that? Yeah, I can also zoom in on
62 00:10:52.040 ⇒ 00:11:20.480 Patrick Trainer: zoom like you can pinch zoom and zoom. So, yeah, so this to me, what this looks like is kind of like your your high, level kind of like day to day, heartbeat of like what the business is doing and what I did is kinda like went through. Look through. I mean, kind of like the obvious things, like like formatting and kind of consistency making things like. So you’re not having to contact switch
63 00:11:20.720 ⇒ 00:11:36.300 Patrick Trainer: a whole, a whole bunch, but also like trying to reduce a lot of the visual clutter. That’s that’s coming around. And what you’re seeing. And so the I, the idea behind this, and kind of like the like, where my
64 00:11:36.570 ⇒ 00:11:42.989 Patrick Trainer: perspective and thoughts are coming from is like, I think of this stuff as kind of like a flywheel. And it’s like you’re
65 00:11:43.320 ⇒ 00:11:57.349 Patrick Trainer: you need to have this like daily update of something quick, distilled kind of like looking at the like. The back page of the Wall Street Journal, where you have a super information. Dense
66 00:11:57.450 ⇒ 00:12:05.969 Patrick Trainer: table. But you’re also looking at 1,500 stocks at at at one time, and so that’s kind of like my
67 00:12:06.620 ⇒ 00:12:25.430 Patrick Trainer: overall like, if I could transform all of this, it would be into something like that. So where you’re able to quickly notice trends like the I mean going up, down, but also those those relations between it. But then also be able to distill your your
68 00:12:25.430 ⇒ 00:12:48.940 Patrick Trainer: keep metrics and your kpis in kind of like one glance that you can digest without having to like. Click through one dashboard scroll all the way down. Think about it, and then wait. Where do I go for this and real? Relate it to another one, and then you just start. It’s like an information overload and so there’s there’s a lot of
69 00:12:49.350 ⇒ 00:12:53.469 Patrick Trainer: like room for for revising things.
70 00:12:53.490 ⇒ 00:12:59.039 Patrick Trainer: And then I’ve also gone through a lot of the
71 00:12:59.310 ⇒ 00:13:19.580 Patrick Trainer: some of the metrics that you’re looking at as well. There’s definite like opportunity for loop grouping like like metrics in in the same thing. So you’re not trying to relate. Say, you’re looking at like something that’s looking at a balance sheet, and then directly next to
72 00:13:19.680 ⇒ 00:13:23.759 Patrick Trainer: something on inventory turn. May just kind of
73 00:13:23.970 ⇒ 00:13:29.280 Patrick Trainer: confuses things a bit and then and then also gone into
74 00:13:29.630 ⇒ 00:13:31.160 Patrick Trainer: a
75 00:13:31.860 ⇒ 00:13:49.819 Patrick Trainer: kind of like best practices in how we view numbers and view kind of like percentages and weights of the of the actual like. Say, if you’re viewing a chart. And so in this, like a concrete example of what I’m talking about, is like
76 00:13:50.300 ⇒ 00:14:05.810 Patrick Trainer: these axis are like a continuous axis, but we have like discrete bars. So it’s like an example of what gets better is that is, is actually like lines, because those lines like hit
77 00:14:05.840 ⇒ 00:14:09.150 Patrick Trainer: the the continuous of the like
78 00:14:09.770 ⇒ 00:14:13.379 Patrick Trainer: the difference between the dollar amount rather than
79 00:14:13.400 ⇒ 00:14:14.660 Patrick Trainer: having, like
80 00:14:14.730 ⇒ 00:14:23.339 Patrick Trainer: 100 discrete customers or 1,000 discrete customers. You kind of have like a mix in between, like you’re never gonna have half of a customer, but
81 00:14:23.350 ⇒ 00:14:25.730 Patrick Trainer: you have sense and
82 00:14:26.170 ⇒ 00:14:36.149 Patrick Trainer: quarters, nickels and dimes, and all all of that and then there’s also like some opportunities, too, that are just kind of like visual cues of like.
83 00:14:37.090 ⇒ 00:14:40.260 Patrick Trainer: I’ll use this again as an example, like
84 00:14:40.350 ⇒ 00:14:44.930 Patrick Trainer: humans are intrinsically bad at
85 00:14:46.550 ⇒ 00:14:56.279 Patrick Trainer: kind of like discerning percentage differences like, if you were to tell me like, what was Walmart on the 26 like, how much money was that it’s like I have. I have no idea.
86 00:14:56.290 ⇒ 00:15:03.099 Patrick Trainer: I can maybe say a little bit, but it’s like it’s kinda like in pie charts like it’s
87 00:15:03.160 ⇒ 00:15:20.460 Patrick Trainer: there’s there’s one that looks like 75%. And then there’s another one that looks like 25. But it it might be 19 which, when you’re dealing with a high enough like dollar amount like that, that 6 percentage points really adds up and can kinda like skew your views.
88 00:15:20.470 ⇒ 00:15:30.450 Patrick Trainer: but then again, at the at the end of the day, like the the goal. For kind of like these daily flywheels are like.
89 00:15:30.710 ⇒ 00:15:45.370 Patrick Trainer: the goal is like quick information, retrieval and like quick spotting. And so I think with these, at least, at like the high, like executive level, you wanna be able to zoom out a good bit and see the see the forest from the trees.
90 00:15:45.510 ⇒ 00:15:46.819 Patrick Trainer: and then
91 00:15:47.130 ⇒ 00:15:56.099 Patrick Trainer: the subsequent dashboards are derivatives of that. Where, when you’re when you want to ask additional questions you have like what
92 00:15:56.110 ⇒ 00:15:59.110 Patrick Trainer: XYZ is affecting this metric.
93 00:15:59.160 ⇒ 00:16:06.260 Patrick Trainer: then you have that opportunity to further drill down and get to it when you have like more time. And you’re not just
94 00:16:06.380 ⇒ 00:16:09.689 Patrick Trainer: trying to kind of like, understand what’s going on right now?
95 00:16:11.610 ⇒ 00:16:14.590 Patrick Trainer: So at a
96 00:16:14.950 ⇒ 00:16:18.260 Patrick Trainer: again at a high level, I did this for
97 00:16:18.590 ⇒ 00:16:28.869 Patrick Trainer: most of these actual, all 4 of these dashboards, and there’s similar similar sentiment around the same. But it all kinda
98 00:16:28.980 ⇒ 00:16:35.490 Patrick Trainer: folds together. So I think I think kind of like my, that the end of it is like there’s a there’s a ton of opportunity
99 00:16:35.580 ⇒ 00:16:39.800 Patrick Trainer: to. I mean, just make your
100 00:16:40.070 ⇒ 00:16:43.140 Patrick Trainer: like gut intuition a lot quicker.
101 00:16:45.240 ⇒ 00:16:47.950 Daniel Schonfeld: Yeah. And I think
102 00:16:48.240 ⇒ 00:17:03.409 Daniel Schonfeld: a lot of the important things you you hit on the head, especially the groupings. You know, you can get a little bit of a paralysis by analysis with all this data. But I think you know, to Utam’s credit our first instruction was, let’s get all the data
103 00:17:03.440 ⇒ 00:17:08.749 Daniel Schonfeld: without even asking for any visuals. But let’s just get all the data in, especially on the revenue.
104 00:17:08.800 ⇒ 00:17:14.369 Daniel Schonfeld: the cost, and then that quickly moved to shipping because we recognized the large opportunity utam and bended
105 00:17:14.520 ⇒ 00:17:26.529 Daniel Schonfeld: and actually saved us a ton of money by by kind of shifting focus to the shipping side. But now that we’ve got data and we have a good feel for the accuracy and timeliness of it.
106 00:17:26.859 ⇒ 00:17:29.880 Daniel Schonfeld: Which will continue to watch and monitor.
107 00:17:30.030 ⇒ 00:17:31.290 Daniel Schonfeld: yeah, I think
108 00:17:31.300 ⇒ 00:17:39.989 Daniel Schonfeld: now, grouping it and doing it more in a linear fashion, like, if I just wanna dive into the revenue side of the business, and it’s all kind of
109 00:17:40.060 ⇒ 00:17:43.190 Daniel Schonfeld: in the same area, because right now, kind of
110 00:17:43.280 ⇒ 00:17:45.710 Daniel Schonfeld: all over the page and in different sections.
111 00:17:45.730 ⇒ 00:17:51.829 Daniel Schonfeld: But I do think at some point soon, before we start any work is that you should do kind of like a
112 00:17:51.880 ⇒ 00:18:00.530 Daniel Schonfeld: a bit of a working session or brain dump with Ben and then myself, cause we both have a bit of. We’ll use this differently.
113 00:18:00.850 ⇒ 00:18:06.690 Daniel Schonfeld: whereas he might be more into the weeds and more tactical
114 00:18:06.850 ⇒ 00:18:13.309 Daniel Schonfeld: and so there might be different needs. But we probably could accomplish it, you know.
115 00:18:13.830 ⇒ 00:18:15.829 Daniel Schonfeld: you know, with with the same
116 00:18:16.210 ⇒ 00:18:24.700 Daniel Schonfeld: kind of out out like it just might change the way you configure it a bit meet both needs. We’re we’re pretty much on the same page. Just
117 00:18:24.910 ⇒ 00:18:27.520 Daniel Schonfeld: I’m I’m a little more higher level.
118 00:18:27.660 ⇒ 00:18:37.369 Patrick Trainer: And then he’s he’s gonna get a little more into the details of it, right? Right?
119 00:18:37.510 ⇒ 00:18:41.780 Patrick Trainer: It hints at like future iterations of like
120 00:18:41.890 ⇒ 00:18:54.040 Patrick Trainer: as we go through these kind of like exercises, we’re gonna discover more and more like kpis and things that you want to pay attention to. And of course, like, like all of these
121 00:18:54.290 ⇒ 00:19:01.240 Patrick Trainer: dashboards, models, charts, whatever they rely on the underlying models and data sets. And like.
122 00:19:01.720 ⇒ 00:19:16.109 Patrick Trainer: if they’re kind of like they go hand in hand but like it’s your your colleague, Ben. There, it’s like I’m I’m sure we’re eventually gonna be talking about like kpis around like inventory turnover stock out rates
123 00:19:16.210 ⇒ 00:19:21.430 Patrick Trainer: accuracy rates those sorts of things. And
124 00:19:22.060 ⇒ 00:19:29.729 Uttam Kumaran: yeah. So we’ll be able to leave those in as well. Yeah, I think I think I think Daniel nailed it like we have now. And you know.
125 00:19:29.750 ⇒ 00:19:41.240 Uttam Kumaran: through a lot of the data models is that we have now all the data across. you know, all the way from like gross sales all the way down to refunds and discounts and everything in between. And so
126 00:19:41.430 ⇒ 00:19:55.310 Uttam Kumaran: the the data that you see right now is pretty much our first passage is showing, you know, kind of the depth, but I think now the all the stuff is available. And now the packaging, I think, is the most important part.
127 00:19:55.360 ⇒ 00:20:07.969 Uttam Kumaran: All the modeling is is frankly pretty much done. There are like small things we’re working on, but from inventory side all the way to sales, to marketing, to refunds, to discounts, to shipping. We have everything from every source.
128 00:20:08.040 ⇒ 00:20:11.379 Uttam Kumaran: And so I’m really excited that I think we could.
129 00:20:11.430 ⇒ 00:20:22.230 Uttam Kumaran: The iteration cycles of the visualizations are much quicker than the iteration cycles on the on the modeling side. And so it’s pretty much just getting that putting into a visualization that makes sense.
130 00:20:22.920 ⇒ 00:20:30.790 Daniel Schonfeld: yeah, Patrick, do you? Do you have any training or or any background in in incorporating AI into any of this this data modeling. And
131 00:20:31.620 ⇒ 00:20:34.580 Patrick Trainer: so like, I
132 00:20:35.320 ⇒ 00:20:49.279 Patrick Trainer: have a lot of like like personal pet projects, kind of on the side that are using like AI and Gen. AI. I’ve worked with companies that I mean, their main product is is built around AI as far as like
133 00:20:49.560 ⇒ 00:20:52.510 Patrick Trainer: data modeling and and dashboarding. There’s
134 00:20:52.770 ⇒ 00:21:02.870 Patrick Trainer: I’d say the the routes, or that AI takes place in. It’s more from a like a statistical approach in in like forecasting
135 00:21:02.930 ⇒ 00:21:06.260 Patrick Trainer: and so like, if
136 00:21:06.850 ⇒ 00:21:12.939 Patrick Trainer: I think the like, the term AI can be a bit of a misnomer. In the sense of like.
137 00:21:13.820 ⇒ 00:21:25.619 Patrick Trainer: it’s they’re they’re both statistical models. But those forecasts and statistical models are definitely things that are incorporated in many
138 00:21:25.650 ⇒ 00:21:39.810 Patrick Trainer: dashboards, etc. Yeah, I think if we get to the end, the end state would be, I mean, ideally, and I’m not gonna get too much into the weeds here. But the way I was thinking about it recently is, if we had these by groupings, let’s say there was a revenue dashboard or section
139 00:21:40.100 ⇒ 00:21:51.019 Daniel Schonfeld: something. I don’t care what you want to call it AI machine learning data, whatever. Something that was like an alert center or a learning center like on the right side
140 00:21:51.160 ⇒ 00:22:02.419 Daniel Schonfeld: that would just point out anomalies or trends that it sees or understands based on historical data. Maybe we can kind of like point it in a direction. So it’s not giving us
141 00:22:02.440 ⇒ 00:22:22.480 Daniel Schonfeld: a flux of information that’s irrelevant. So we say, look for trends in this. So certain areas, we can kind of hone in on. If we came in there would be some kind of an alert saying, we’ve detected an anomaly for better or for worse, and that can immediately trigger rather than me, spending 4 h shifting through data. It’s alerting us to it.
142 00:22:22.720 ⇒ 00:22:39.870 Patrick Trainer: I again, to give you a concrete example of like something that I’ve done in the past. It’s like I’ve worked with databases, my entire career, and one of those the things that you need to do to maintain like uptime is like
143 00:22:39.870 ⇒ 00:22:57.360 Patrick Trainer: monitoring database load. And so you have like an average, and that’s good. But what you really want to understand, is like the distribution of load, like on the on a bell curve. And so you have these like and 99% tiles.
144 00:22:57.470 ⇒ 00:23:19.219 Patrick Trainer: and use those as kind of like your your litmus test of like, when are things really like out of that normal distribution range and going into those like 95 or 90 95 99 ranges, and it’s like those are. That’s your that’s your anomaly detection.
145 00:23:19.490 ⇒ 00:23:28.570 Patrick Trainer: You find the average. And if something is like 2 standard deviations outside on either way, Trigger alarm. It’s not actually AI,
146 00:23:28.610 ⇒ 00:23:32.040 Patrick Trainer: yeah, you’re you’re you’re telling the system what to look for, and
147 00:23:32.070 ⇒ 00:23:41.849 bencohen: we don’t need to do more than that. That’s enough. The alert is enough. And that’ll be just. We are in, you know. Yeah, that’s and yeah, from from the alert. Then
148 00:23:41.950 ⇒ 00:23:43.160 Patrick Trainer: you action.
149 00:23:43.840 ⇒ 00:23:55.050 Daniel Schonfeld: Yeah. And I would even say further, like we we probably need in the beginning, can focus in on things like shipping and refunds cause revenue. We pretty much know what we’re doing to obtain that revenue. I hope
150 00:23:55.050 ⇒ 00:24:15.530 Daniel Schonfeld: expenses are quite constant. There will be deviations and add spend. But we’re not gonna like freak out every time there’s a deviation on a daily basis from Google or Facebook. But when shipping rates start to get out of whack or refunds, that’s a huge one. We wanna keep an eye on that. Sometimes it’s very difficult across all the platforms.
151 00:24:15.620 ⇒ 00:24:24.130 Daniel Schonfeld: So like that’s an area where we would want to know pretty quickly, cause we’ve lost. I mean, last year on like product like ladders. I’ll use an example.
152 00:24:24.440 ⇒ 00:24:40.419 Daniel Schonfeld: It took a it took a month or 2 to realize that we’re really losing money on shipping returns refunds because things were happening very quickly, and we sold them rapidly. We, if we lower the price point 20 on any one skew. We’re gonna get a huge influx of sales.
153 00:24:40.500 ⇒ 00:24:46.539 Daniel Schonfeld: and it might take a few days to catch up. But if there was a programs like flashing alerts.
154 00:24:46.630 ⇒ 00:24:52.420 Daniel Schonfeld: saying that all right. I always forget to show up my messages now they’re flying in
155 00:24:52.780 ⇒ 00:25:09.919 Daniel Schonfeld: Should I lost my my train of thought. Yes, that we want to catch those almost immediately, and it would be worthwhile to to understand those deviations almost semi real time to set those alerts like blaring like, you guys are getting crushed
156 00:25:10.250 ⇒ 00:25:13.430 Daniel Schonfeld: on refunds and shipping. Take action. Now.
157 00:25:13.490 ⇒ 00:25:22.280 Daniel Schonfeld: revenue. Yeah, it’s up or down. We’re not gonna go. There’s no emergency there. There’s like kind of a priority of deviations, if you will.
158 00:25:22.620 ⇒ 00:25:28.359 Uttam Kumaran: It’s also the levers, right? Just as you mentioned like. If you know that. Okay, we lower the price. This is like.
159 00:25:28.390 ⇒ 00:25:48.510 Uttam Kumaran: what price change impact will have. And so you can say, Hey, here your options are like, lower the price. Or you know, that’s that’s the stuff as we’re as even. I’m going through discounts this week pretty heavily. The other area is, I’m I’m really reading through tickets. And the the thing I’m gonna actually
160 00:25:48.510 ⇒ 00:26:12.300 Uttam Kumaran: use AI for this week is going through and summarizing a ton of tickets. And I’m gonna do that in Snowflake to pretty much give us like a one line indication of Hey, there’s 10 different responses between Cody and this person. What was the gist like, was it? Was it? The thing was broken? And again, if I were to write that sequel, Pat, it’s it’s kind of complicated, and it’s it’s gonna
161 00:26:12.340 ⇒ 00:26:26.219 Uttam Kumaran: be a nightmare. But pretty much I’m like, take all this text and summarize what was the key issue? That’s a great area for us to go from like a 20 back and forth email to understanding. Oh, something was broken and then flicking that directly into skew. So
162 00:26:26.520 ⇒ 00:26:36.920 Uttam Kumaran: so both on that sort of like textual understanding. But then, definitely, I think there’s something to do on like identifying those trends. And how does refunds affect?
163 00:26:37.180 ⇒ 00:26:57.889 Daniel Schonfeld: I like, how does refunds, and what is like price action affect those refunds or revenue. There’s probably some Si statistical modeling there, too, is there? I’m I’m sure there is. If it’s not developed already, but like sort of AI, that summarizes what’s going on in the business and kind of emails you and tells you, here’s an overview what’s happening in the refund area, like in in
164 00:26:58.220 ⇒ 00:27:00.019 Daniel Schonfeld: you know, kind of paraphrase
165 00:27:00.330 ⇒ 00:27:05.059 Daniel Schonfeld: whittling you’re talking about. There must be plugins or something people are doing
166 00:27:05.220 ⇒ 00:27:08.850 Daniel Schonfeld: to kind of summarize on daily basis what happened in the business?
167 00:27:09.010 ⇒ 00:27:11.189 Daniel Schonfeld: Just so you can get a quick email.
168 00:27:11.540 ⇒ 00:27:16.480 Uttam Kumaran: I think, within domains. There are like, for example, I’m sure
169 00:27:16.550 ⇒ 00:27:30.750 Uttam Kumaran: Zendesk offers like a tell me what the total summary of tickets are. There are a lot of I mean me and Pat, I would say, we look at like all these new tools every day. There’s probably a lot of people that will promise you that they can look at all your data and kind of do that.
170 00:27:31.560 ⇒ 00:27:57.880 Uttam Kumaran: I mean, even just like thinking about that. Prom. It’s it’s quite difficult. But again, it’s like our first board is like this dashboard. This vital signs dashboard is like the manifestation of that today. It’s like, okay, how do we get that? So we’re not reading it. It’s then outputting something. And then how do how do we? How are we able to actually summarize and have layers so like? Here are the key movements and then summarize those movements into like a flash email. That’s easy to read thinking about those.
171 00:27:57.970 ⇒ 00:28:14.689 Uttam Kumaran: I think that step by step is good way. But again, this is stuff that most companies stop at this point. And I think they lose out because they don’t realize that this is just the representation of the data. It’s actually the action. And like the distillation into like the 4 things that is actually the goal. Not
172 00:28:14.900 ⇒ 00:28:16.539 Uttam Kumaran: like, you know. So
173 00:28:16.600 ⇒ 00:28:22.269 Daniel Schonfeld: yeah, no, that’s exactly right. I mean. Look, it led to you doing this, the whole shipping thing just by us, bringing in the data
174 00:28:22.320 ⇒ 00:28:47.659 Daniel Schonfeld: through osmosis whatever you want to call it, like we realize holy shit, there’s a problem with the shipping. Let’s fix it. That’s what I’m trying to get to without us doing too much intelligence, sifting through thousands, or, if not, millions of data points to figure that out is, can we use programs off the shelf for the smoke that would at least alert us immediately to some kind of an anomaly or issue, and then distill it down to what matters over time. Is that?
175 00:28:48.060 ⇒ 00:29:13.520 Uttam Kumaran: Yeah. It’s a really good example, for example, that that maybe took a month or 2 to kind of go through everything and see that we were to take that as a base case, what opportunities there, if and again, we have to not go, do that for discounts, for refunds to every single domain. Okay, how do we make that happen in 2 weeks instead of whatever it was we don’t know. We don’t know. And the data will tell. Yeah, we might find that we should only be selling cover pumps in the south southeast.
176 00:29:13.790 ⇒ 00:29:28.920 Daniel Schonfeld: And something would say, Listen, you’re just highly unprofitable. Instead of us cutting off a skew which sometimes Ben. And I do say this skew sucks, let’s just cut it. But in reality, when we sift through the data like, Oh, shit! Actually, in these months, in these zones. It actually works really well.
177 00:29:28.930 ⇒ 00:29:45.700 Daniel Schonfeld: And if we had a DC. There and just put this one skew there, you would crush it, and then we don’t have to worry about anything else and cut the add spend in all these other areas, so that those are like the nuggets I’m really looking for. And that’s how we really is not taking these broad strokes or making assumptions that something’s not working, but
178 00:29:45.840 ⇒ 00:29:54.249 Daniel Schonfeld: going through that data and understanding. If a skew works in a certain area at a certain time. Certain place. It’s tough. There’s all too much data to even go through
179 00:29:54.320 ⇒ 00:29:57.849 Daniel Schonfeld: as a human being. I mean it would take months just to figure that out. But
180 00:29:57.870 ⇒ 00:30:12.299 Daniel Schonfeld: as an intelligent program would see that in seconds. If it can crunch that data, shipping zones returns. Looking at did we discount heavily and looking at time of year? I imagine it’s several commands, but it can crunch through. That data
181 00:30:12.370 ⇒ 00:30:22.689 Daniel Schonfeld: would take us months in probably minutes. So that that’s kind of stuff I’m I’m hoping for in the future, not at this this stage, but that we’d get to at some point
182 00:30:22.920 ⇒ 00:30:46.869 Patrick Trainer: that that makes sense. And I’ve heard a story, and it’s it’s they call it the Turkey Popper problem. And so, if you know, like the turkey problem er the turkey popper at Thanksgiving like you put it in the turkey, and when the turkey’s gone in the oven the little red thing pops up. Well, so the creator of that. They they were going. They’re shipping, they they put it out into market, and it was doing good. And then.
183 00:30:47.540 ⇒ 00:30:54.539 Patrick Trainer: months a year later, whatnot they kind of get these alert, and they realize like holy shit, like people are returning
184 00:30:54.820 ⇒ 00:31:01.380 Patrick Trainer: Turkey poppers at this incredible rate like these turkey poppers are like this is, we gotta stop.
185 00:31:01.500 ⇒ 00:31:04.859 Patrick Trainer: But after looking more into it they realized that
186 00:31:05.000 ⇒ 00:31:29.719 Patrick Trainer: all our sales are through the roof. So, of course, like on a nominal basis like our our turkey poppers are going to like, you’re going to have more returns after app when you sell more volume. And so that’s kind of like adjacent to to to what I’m what I’m hearing from from you. There, you’re exactly right, especially when it comes to 3 $4,000 heat pumps, and we’re like holy shit, right? So many returns. Now we’re warranties.
187 00:31:29.810 ⇒ 00:31:48.680 Daniel Schonfeld: We don’t know when the warranties are from. Like. That’s a great example. Actually. I’ll never forget the turkey popper now that is a good one. But it’s very analogous to to our situation, especially as it relates to not 2 products, but $3,000 that we need to understand.
188 00:31:48.700 ⇒ 00:32:09.260 Daniel Schonfeld: If we suddenly get 12 returns on heat pumps. And we’re like, Wow, we gotta shut this down. But in reality they’re coming back over 3 years, and we find that someone’s reclaiming a warranty over X amount of time. It’s actually not a huge deal. In relation. So that’s a that’s a great analogy, actually. And that’s something that is our Blind Spot at the moment.
189 00:32:13.150 ⇒ 00:32:18.170 Daniel Schonfeld: And that’s what time I know you were looking into that as far.
190 00:32:18.490 ⇒ 00:32:24.719 Daniel Schonfeld: That’s a perfect example is how the data prompted an action is. I actually went in. I was like holy shit. The discounts are through the roof.
191 00:32:25.090 ⇒ 00:32:36.609 Daniel Schonfeld: And then I realized that someone had $0. They were be given they were given away for free. I’m like, I know we’re not giving free stuff away. They must be warranties. I asked. Cody’s. Yes, those are. Those are warranties
192 00:32:36.790 ⇒ 00:32:47.409 Daniel Schonfeld: and which further prompted us to look into. Why are we giving the entire unit away versus parts? And so again, all this stuff is meant to lead to conversation, ultimate action.
193 00:32:47.810 ⇒ 00:33:03.740 Daniel Schonfeld: A bunch of these diagrams, and the one that Patrick pointed out in the beginning is yes. and Ben might argue, though differently. But for me personally, when I see, like Walmart in a little tiny little bar, I have no action personally that I can take
194 00:33:03.860 ⇒ 00:33:11.710 Daniel Schonfeld: cause. I don’t really understand what that means to the business, or what that real slipper represents in relation to the business.
195 00:33:11.980 ⇒ 00:33:20.209 Daniel Schonfeld: It’s just a nice to know, like, oh, okay, Walmart’s doing okay. And I see the bar going up, and that’s fantastic. But I don’t really have an action to take from there.
196 00:33:20.260 ⇒ 00:33:33.689 Daniel Schonfeld: whereas the more tactical is something telling us, hey, something’s out of whack. You need to look into this now, and I can just drill right into the data and figure it out in 2 s. Right? Right?
197 00:33:33.780 ⇒ 00:33:38.170 Patrick Trainer: That are successful. Data. Visualization is like it’s not
198 00:33:38.780 ⇒ 00:33:47.339 Patrick Trainer: the visualization itself that is valuable. It’s the action and kind of like the like. It’s a tool.
199 00:33:47.400 ⇒ 00:33:50.540 Patrick Trainer: it it’s like another tool in your toolbox to
200 00:33:50.990 ⇒ 00:34:06.110 Patrick Trainer: for you to to assess and to discern, and then, like what you said, carry on a conversation, and then, of course, correct from there. Yeah, yeah. Visuals, I’m personally a visual guy. You know, everyone’s got their different mode of how they learn best
201 00:34:06.190 ⇒ 00:34:14.229 Daniel Schonfeld: that prompts like you said it prompts to it keeps going down the line until you figure out what the problem is. That’s the first thing is like, oh, wow! There’s some kind of an anomaly there.
202 00:34:14.540 ⇒ 00:34:19.290 Daniel Schonfeld: Okay, I think we’re all on the same page. What are? What are the next steps for? What are the next steps?
203 00:34:20.239 ⇒ 00:34:42.299 Uttam Kumaran: Yeah, I think one is Dan. You mentioned having a conversation between you and Ben. I don’t know what’s best. I think I want Pat to go headfirst onto some of these dashboards and really work directly with you on. Hey? What are the steps I look at, or what are the levers that I could pull on a daily, weekly, monthly basis? And then how do we
204 00:34:42.350 ⇒ 00:35:01.630 Uttam Kumaran: get you to make or not make a decision via visualizations. I think that’s like a really great like thing to tackle you. Let me know what’s best. Is it worth to just get all that information initially from you and Ben. And then think about, okay, how will this kind of work with your project? Or is it worth just going directly into.
205 00:35:01.930 ⇒ 00:35:08.250 bencohen: I think we need to be really careful with me and Dan directing Patrick to too much.
206 00:35:08.320 ⇒ 00:35:17.349 bencohen: I think we we should do is, have, like, figure out the economics of how this engagement goes. So we understand what you know time and money, and then
207 00:35:18.500 ⇒ 00:35:24.770 bencohen: kick off. Call Dan Patrick, speak for 30 min tells him how he thinks about, and he’s already kind of done it. But
208 00:35:24.830 ⇒ 00:35:30.709 bencohen: an a another 30 min block to download. How Dan likes to observe and think
209 00:35:30.890 ⇒ 00:35:33.800 bencohen: I need to do one. It’s a little bit different from how Dan things.
210 00:35:34.030 ⇒ 00:35:37.989 bencohen: And then I think Patrick should come to us and say, here’s my plan of attack.
211 00:35:38.160 ⇒ 00:35:39.609 bencohen: Here’s what I’m gonna do.
212 00:35:39.940 ⇒ 00:35:46.389 bencohen: and then we do it. And then, you know, we we check in after he’s gotten to a certain level. I don’t. I don’t. Wanna
213 00:35:47.720 ⇒ 00:35:52.070 bencohen: well, you know, I want. I want him to to make some decisions.
214 00:35:52.080 ⇒ 00:35:58.919 bencohen: with his experience. If if we’re directing too much. It’s gonna end up like
215 00:35:59.050 ⇒ 00:36:01.020 bencohen: what happened with Kenny and Ravine?
216 00:36:02.120 ⇒ 00:36:10.139 Daniel Schonfeld: There’s no sense in it, you know. I agree, I agree. And then he could. Even if he wants or Ben, maybe you can do it a debrief with
217 00:36:10.240 ⇒ 00:36:20.909 Daniel Schonfeld: maybe like a wants or nice to haves with Kim, with Cody. Just so they’re in the loop and they don’t build something or like, I don’t even know how to use this. So be good to get everyone’s input all, all the users of this
218 00:36:21.250 ⇒ 00:36:24.770 Daniel Schonfeld: of who are going to be actively using the system.
219 00:36:24.920 ⇒ 00:36:34.539 Daniel Schonfeld: To put maybe 2030 min it’s worth. It’s a worthy investment for him to meet with these folks. And Ben can direct it. Maybe Ben meets with them and
220 00:36:34.620 ⇒ 00:36:45.410 Daniel Schonfeld: shoots it uphill. But I think it’d be good for him to meet with all the actors that are gonna be using this. I think that’s a good great approach, Ben, I think doing those 30 min blocks. And then we’d like to get out of your way.
221 00:36:45.580 ⇒ 00:36:54.570 Patrick Trainer: Yeah, that and that all makes sense to me, too. Like I was, I was gonna say, from my perspective,
222 00:36:54.830 ⇒ 00:36:57.680 Patrick Trainer: like what I would like to know is kind of like
223 00:36:58.740 ⇒ 00:37:00.759 Patrick Trainer: when you think about the business.
224 00:37:00.980 ⇒ 00:37:22.189 Patrick Trainer: What are like, what comes? What pops into your mind? And and where are the levers? So what? Yeah, what are the strings that you pull when you notice something and then we can focus on those as kind of like tier one and then, like each tier subsequent is like
225 00:37:22.340 ⇒ 00:37:25.670 Patrick Trainer: derived and based off of that. Yeah, that
226 00:37:26.030 ⇒ 00:37:28.210 Daniel Schonfeld: that’s the way to do it.
227 00:37:29.510 ⇒ 00:37:47.240 Daniel Schonfeld: yeah, you’re gonna hear different priorities hitting from different people. But at the end of the day. The goal is to maximize revenue and profit and lower expenses, so like the same as any other business, you know, that’s not backed by a Major Vc. And they don’t care. Ii think you’d be surprised at how many places I work where they
228 00:37:47.240 ⇒ 00:38:03.849 Daniel Schonfeld: they never even say that out loud. And I’m like, guys, what? What are we talking about? Yeah, there’s yeah. And now you’re seeing the ramifications of that through failures in the Vc world. Those are all. Vc, that’s what Vc is a growth at all at all costs. Yeah.
229 00:38:03.930 ⇒ 00:38:06.730 bencohen: and that that all costs is usually.
230 00:38:06.760 ⇒ 00:38:14.669 Patrick Trainer: how do we operate at the biggest loss possible. It’s insane. It’s it’s it’s funny, like, I mean, my experience of work with startups.
231 00:38:14.670 ⇒ 00:38:42.079 Patrick Trainer: I mean my entire career. And it’s like it’s grow, grow, grow, grow, grow, grow, grow, spend as much as possible, and then, like, Okay, let’s Ipo and like, hit the break. And like, all right. Now let’s be profitable. And it’s it’s such a shift and it’s it’s like trying to stop a freight train like, yeah, it’s it’s not their fault, either, because they it’s they’ve been conditioned by Vcs to do that. They’re saying, we only want to buy companies at this growth rate
232 00:38:42.100 ⇒ 00:38:49.079 Daniel Schonfeld: until, like fuck it. Who cares about pro? If they don’t give up profit, we don’t care if we’re gonna get 30 million dollars check, and
233 00:38:49.120 ⇒ 00:39:13.839 Daniel Schonfeld: you know, buy out at that rate, why go for profit? But now, all of a sudden, you have to be saying Whoa! We need to buy profitable cash flow businesses, and it’s a whole cycle. It’ll go back to the same way. It was ultimately because they’ll all find the profitable businesses, and they’ll say, Shit, we gotta grow. How are. We gonna make money for lps and just a big cycle? It’s like real estate or anything else.
234 00:39:13.980 ⇒ 00:39:18.480 Daniel Schonfeld: But at the end of the day the common denominator of all good businesses are making profit.
235 00:39:18.520 ⇒ 00:39:21.049 Daniel Schonfeld: That’s right next. And so
236 00:39:21.330 ⇒ 00:39:24.050 Daniel Schonfeld: I do think about things. And
237 00:39:24.510 ⇒ 00:39:34.420 Daniel Schonfeld: you, you, you know, and duality with that like I do think about growth. Also, like Ben will be more focused on the profit side. And say, Let’s just cut these skew. Let’s move on, and I’ll kind of
238 00:39:34.450 ⇒ 00:39:39.949 Daniel Schonfeld: say, well, hold on. We still need to show growth. We don’t want to cut these areas off, because one day
239 00:39:40.230 ⇒ 00:39:54.930 Daniel Schonfeld: we could be ultra profitable in these areas. Case in point, the cover pumps and other ones. Our goal in the beginning was just grow, grow grow. We did have that mindset, and we actually told our agency, don’t worry about profit, for now let’s just dominate the category, and we’ll figure it out.
240 00:39:55.170 ⇒ 00:39:59.429 Daniel Schonfeld: Which has somewhat been the case. But now it is time to get more tactical and
241 00:39:59.550 ⇒ 00:40:02.120 you know. Cut those expenses, and
242 00:40:03.310 ⇒ 00:40:13.440 Daniel Schonfeld: you know you you gotta you gotta. You gotta kind of juggle those 2 worlds because we are ultimately looking for a raise at some point. And so we do have to show that growth. But
243 00:40:13.650 ⇒ 00:40:28.239 Daniel Schonfeld: you know, minded, we, we have to be profitable to get there. So yeah, I think this is gonna work out. Great. I’m really excited. Patrick, really excited to have you you know, join the team and get get to work on this. We need the help for sure, likewise
244 00:40:29.030 ⇒ 00:40:48.179 Uttam Kumaran: cool. So Pat, maybe me and you will regroup, and then we’ll send a note over about maybe doing a download with everybody, maybe this week or next week, just like quick sessions. I mean, again, I’ll pretty much give Pat as much context I have. So hopefully, if those conversations need to happen after that, they’re they’re pretty specific, and then we’ll just let them run.
245 00:40:48.320 ⇒ 00:40:58.479 bencohen: And then maybe I can catch up with you. Yeah, just, you know, just let me know the economics and stuff. Just so we can make the plan on our end. Obviously, you here, we’re we’re focused on
246 00:40:58.550 ⇒ 00:41:05.310 bencohen: things being inbound. So we just want to know we’re getting into, and then otherwise easy to work with. So we’ll rock and roll here.
247 00:41:05.740 ⇒ 00:41:12.209 Daniel Schonfeld: Yeah, let me know. Patrick, what’s your availability? I I’m having surgery tomorrow on my ankle. I’m like a week.
248 00:41:12.400 ⇒ 00:41:16.339 Daniel Schonfeld: 2 weeks like kind of out of the loop of it. But
249 00:41:16.860 ⇒ 00:41:42.969 Daniel Schonfeld: I was thinking, maybe I could even do something later today, or we could do it for later next week, when I’m like next week, you’re gonna be dying to go on phone calls. I’ve been even sooner the first 2, 3 days you’re not into it. And after that you’re like, alright. I need something to do so. Of course, there’s like a major snowstorm tonight and into tomorrow morning. And they’re like, we don’t know if you’re gonna have it. I’m like already so nervous.
250 00:41:43.030 ⇒ 00:41:48.140 Patrick Trainer: They’re like, we’ll let you know in the morning. I’m like, I’m gonna be up all fucking night like.
251 00:41:48.230 ⇒ 00:41:54.049 bencohen: yeah. And you can’t eat either. You need to need to prep your body
252 00:41:54.320 ⇒ 00:41:57.110 bencohen: after midnight, I think. Right.
253 00:41:57.720 ⇒ 00:42:16.800 Daniel Schonfeld: Mushrooms.
254 00:42:17.080 ⇒ 00:42:20.179 Uttam Kumaran: Alright. Guys, awesome. Thank you so much.
255 00:42:20.510 ⇒ 00:42:25.100 Patrick Trainer: Yeah. Likewise talk to you soon.
256 00:42:25.350 ⇒ 00:42:26.539 Daniel Schonfeld: Bye, guys, bye.