Meeting Title: Javy-Data-Engineering-Weekly Date: 2024-11-12 Meeting participants: Luke Daque, Uttam Kumaran, Aman Nagpal, Payas Parab, Robert Tseng
WEBVTT
1 00:03:50.890 ⇒ 00:03:52.019 Aman Nagpal: Hey? How’s it going? Guys?
2 00:03:59.030 ⇒ 00:03:59.636 Luke Daque: Hey, everyone.
3 00:04:00.430 ⇒ 00:04:01.430 Payas Parab: What up guys.
4 00:04:09.010 ⇒ 00:04:09.820 Uttam Kumaran: Hey, everyone.
5 00:04:14.390 ⇒ 00:04:15.890 Payas Parab: How are you guys doing today?
6 00:04:17.769 ⇒ 00:04:18.979 Aman Nagpal: Good! How are you?
7 00:04:18.980 ⇒ 00:04:20.190 Payas Parab: Good.
8 00:04:21.980 ⇒ 00:04:23.360 Luke Daque: Doing well, doing well.
9 00:04:23.910 ⇒ 00:04:32.930 Payas Parab: Nice I think Robert should be joining. I think there’s some stuff in the slack that he wanted. I’m on. You wanted to discuss right regarding like real.
10 00:04:33.290 ⇒ 00:04:34.260 Payas Parab: Oh.
11 00:04:34.910 ⇒ 00:04:35.570 Payas Parab: just.
12 00:04:35.570 ⇒ 00:04:37.090 Aman Nagpal: Where? Sorry? Go ahead?
13 00:04:37.090 ⇒ 00:04:38.410 Payas Parab: No! Go ahead! Go ahead! You go!
14 00:04:38.590 ⇒ 00:04:42.810 Aman Nagpal: I was. Gonna say, while we’re waiting is it Luke or Ryan? What do you prefer.
15 00:04:43.920 ⇒ 00:04:45.760 Luke Daque: Yeah. Just call me Luke.
16 00:04:45.760 ⇒ 00:04:49.520 Uttam Kumaran: We have 2, Ryan. We have 2. Ryan’s on the team, so
17 00:04:49.540 ⇒ 00:04:54.201 Uttam Kumaran: Ryan, one change his name to his middle name. But yeah.
18 00:04:54.590 ⇒ 00:04:56.419 Aman Nagpal: The second one should have changed his name.
19 00:04:57.180 ⇒ 00:05:03.129 Uttam Kumaran: Yeah. Well, that’s cause, Ryan, one is, is obviously nicer and more empathetic. So.
20 00:05:05.507 ⇒ 00:05:15.290 Aman Nagpal: Well, anyway, thanks for hopping on the call earlier I spoke with our Cx. Vp, were you able to find anything about the macros or no.
21 00:05:15.840 ⇒ 00:05:23.560 Luke Daque: No, not still, not like I can’t. I can’t like figure out how way to like, integrate them to the tickets table, or join them.
22 00:05:24.020 ⇒ 00:05:33.129 Aman Nagpal: Yeah. So I did get some info one. We do definitely need the macros used. I just remembered how we were doing before. And
23 00:05:33.310 ⇒ 00:05:57.090 Aman Nagpal: I mean honestly, the the plan would be to get it from the Api, if possible, just because it’ll constantly sync that data. But what’s happening is when they use the macro. It doesn’t appear anywhere on that ticket. It may not even. I’m not sure but it seems like it may not even get attached to the ticket at all, like we were discussing. What happens is
24 00:05:57.100 ⇒ 00:05:59.830 Aman Nagpal: they use the macro. It shoots the message.
25 00:05:59.850 ⇒ 00:06:07.100 Aman Nagpal: and you know I would hope it saved somewhere, so we can pull it from the Api. But if not, what we were doing before was setting up a web hook
26 00:06:07.160 ⇒ 00:06:11.190 Aman Nagpal: to each macro, or to certain macros, and I guess what
27 00:06:11.250 ⇒ 00:06:19.660 Aman Nagpal: one option is. What we would need to do is set up a web hook with every macro and fire that every time the macro is used.
28 00:06:19.740 ⇒ 00:06:31.660 Aman Nagpal: and record that somewhere, whether it’s, you know, in the data Warehouse or before we were doing amplitude directly. That’s 1 option. Another option is, I don’t know if this is easier or harder.
29 00:06:32.391 ⇒ 00:06:36.209 Aman Nagpal: Just to when the macro is used, try to group
30 00:06:36.540 ⇒ 00:06:38.720 Aman Nagpal: the similar messages
31 00:06:38.760 ⇒ 00:06:51.950 Aman Nagpal: altogether. So I you know, if the Cx rep is using a macro for cancellation right? The message is probably the same every time, and maybe they make some tweaks here, there. But if there’s certain sections
32 00:06:51.970 ⇒ 00:07:05.619 Aman Nagpal: within the message, or like maybe like an invisible Id or something that within the message that we can grab and kind of group together as that Macro was used that way. I think that’s an option as well. I don’t know what you think is is better or easier.
33 00:07:06.830 ⇒ 00:07:19.560 Luke Daque: Yeah, we can do both like like for the the 1st one, though, the web hooks one that would be like a moving forward one. Right? We can’t like, put the web books, the historical messages anymore. And that’s like the problem. With that
34 00:07:19.690 ⇒ 00:07:21.110 Luke Daque: we can try to
35 00:07:21.840 ⇒ 00:07:29.139 Luke Daque: do the second option that you mentioned where it’s like. Mostly it’s the same message. But maybe there’s just a couple of
36 00:07:29.637 ⇒ 00:07:37.360 Luke Daque: variables there, like the name, or whatever but the the message would be the same mostly, and we can compare it with what the set
37 00:07:37.440 ⇒ 00:07:41.260 Luke Daque: messages in the macro. Maybe that’s something I can look into.
38 00:07:41.970 ⇒ 00:07:48.719 Aman Nagpal: Okay, yeah, no, that that sounds great. I actually just remembered I did email them in the past
39 00:07:48.990 ⇒ 00:07:50.970 Aman Nagpal: about asking if
40 00:07:52.690 ⇒ 00:07:54.759 Aman Nagpal: the macro is available.
41 00:07:54.780 ⇒ 00:07:56.369 Aman Nagpal: Let me check my email.
42 00:07:56.830 ⇒ 00:08:00.959 Aman Nagpal: So I asked them if we would like to record what macros are being used.
43 00:08:01.240 ⇒ 00:08:06.810 Aman Nagpal: One rep at gorgeous said, I’m afraid that there’s not an easy way to send over the macro used
44 00:08:06.860 ⇒ 00:08:08.970 Aman Nagpal: inside the Json body.
45 00:08:09.707 ⇒ 00:08:14.120 Aman Nagpal: Since the value macro value is not present under the ticket object.
46 00:08:14.220 ⇒ 00:08:18.859 Aman Nagpal: What you could do as a workaround is, use this endpoint that they link to
47 00:08:20.970 ⇒ 00:08:28.060 Aman Nagpal: for listing the ticket messages for a specific ticket id and check if the macro object is present.
48 00:08:29.559 ⇒ 00:08:32.490 Aman Nagpal: So let me forward this over to you. I don’t know if you wanna.
49 00:08:33.010 ⇒ 00:08:33.650 Luke Daque: Okay.
50 00:08:33.659 ⇒ 00:08:43.659 Aman Nagpal: Reach out to them. But hopefully this is helpful. If not, maybe they can give us an answer and worst case scenario. If neither of those routes help. Then maybe we can do the message grouping like you, said.
51 00:08:44.230 ⇒ 00:08:45.130 Luke Daque: Sounds good.
52 00:08:45.810 ⇒ 00:08:47.590 Aman Nagpal: Perfect. Let me
53 00:08:47.770 ⇒ 00:08:49.300 Aman Nagpal: forward this to you. Now.
54 00:08:52.040 ⇒ 00:08:53.520 Aman Nagpal: what’s your email?
55 00:08:56.160 ⇒ 00:09:00.289 Luke Daque: It’s ryan dot dake DAQU e
56 00:09:00.380 ⇒ 00:09:02.100 Luke Daque: at brainforge.ai
57 00:09:03.240 ⇒ 00:09:04.000 Luke Daque: right.
58 00:09:07.940 ⇒ 00:09:09.380 Aman Nagpal: Cool, just forward it to you.
59 00:09:09.740 ⇒ 00:09:10.939 Luke Daque: Cool sounds, good.
60 00:09:15.150 ⇒ 00:09:20.000 Aman Nagpal: Yeah. Did you guys hear back from 5 tran about the charges.
61 00:09:20.970 ⇒ 00:09:31.670 Uttam Kumaran: Yeah, Nico is leading that he’s just added, yeah, he’s just on vacation this week. But I’m I think
62 00:09:31.780 ⇒ 00:09:34.519 Uttam Kumaran: I can just follow up and try to get a hold of that.
63 00:09:34.949 ⇒ 00:09:37.349 Uttam Kumaran: Yeah, we. We had a conversation with them.
64 00:09:38.140 ⇒ 00:09:43.279 Uttam Kumaran: We had a conversation with them about a month ago before the charge hit confirmed the rate.
65 00:09:43.790 ⇒ 00:09:51.260 Uttam Kumaran: Then, like a different number happens. So I think they’ll smooth it over and then, basically, it’s kind of like what I mentioned, where.
66 00:09:51.410 ⇒ 00:09:57.710 Uttam Kumaran: as we sync everything, we get a sense of, like what we need for our Kpis, and then we just cut everything else off
67 00:09:58.132 ⇒ 00:10:03.297 Uttam Kumaran: and then, ideally, what we can do is work with them on like a discount schedule.
68 00:10:03.700 ⇒ 00:10:05.899 Uttam Kumaran: that’s usually kind of like how it works.
69 00:10:07.650 ⇒ 00:10:12.719 Uttam Kumaran: So again, we, it’s hard for us unless we didn’t know exactly like how many
70 00:10:12.740 ⇒ 00:10:27.099 Uttam Kumaran: transactions and how much data was coming in, especially given like Tiktok. And everything is coming through. Shopify as well. But I think a lot of that will stabilize closer to like a thousand a month. We cut some stuff off stuff off last week. So we’re waiting.
71 00:10:27.320 ⇒ 00:10:29.889 Uttam Kumaran: probably for this month to see like how it impacts.
72 00:10:32.530 ⇒ 00:10:36.251 Aman Nagpal: Yeah, that makes sense. I think there was just some
73 00:10:36.740 ⇒ 00:10:40.119 Aman Nagpal: I don’t know if miscommunication is the right word. But I guess, like you said maybe
74 00:10:40.613 ⇒ 00:10:44.720 Aman Nagpal: we just didn’t discuss it enough about the costs of the
75 00:10:44.850 ⇒ 00:10:47.150 Aman Nagpal: various platforms
76 00:10:47.410 ⇒ 00:10:51.090 Aman Nagpal: cause. I had no idea it would even be a thousand or 1,500 a month.
77 00:10:51.701 ⇒ 00:10:58.330 Aman Nagpal: So if you know I think you guys are already working on kind of a list of what expectations could be
78 00:10:59.110 ⇒ 00:11:03.380 Aman Nagpal: in terms of costs for each platform.
79 00:11:03.750 ⇒ 00:11:07.579 Aman Nagpal: But yeah, I guess we’ll take a look at that, and then kind of take it from there.
80 00:11:09.420 ⇒ 00:11:22.880 Aman Nagpal: yeah, I mean, I I thought you know, 5 trend would be at the beginning. Almost nothing. So it’s just shocking to see that number, right? But yeah, see what you guys can do with the discount. And then you know, we can kind of take it from there.
81 00:11:23.620 ⇒ 00:11:32.319 Uttam Kumaran: Yeah. And we’re saving costs in a couple of other areas. Like snowflake, we’re running pretty lean and Dbt, we’re basically
82 00:11:32.420 ⇒ 00:11:38.939 Uttam Kumaran: trying to run for free. But yeah, hopefully, we can give you that estimate as soon as we get it.
83 00:11:39.100 ⇒ 00:11:45.350 Uttam Kumaran: And then, you know, depending on what that is, there’s a couple of other competitors. 5 train that we can go discuss with and move
84 00:11:45.410 ⇒ 00:11:48.540 Uttam Kumaran: and and move off of, and things like that. So if we need.
85 00:11:49.800 ⇒ 00:11:50.530 Aman Nagpal: Okay.
86 00:11:52.524 ⇒ 00:11:56.560 Uttam Kumaran: Cool, I guess. Pies. Do you want to go? Or we? We can talk about the real stuff
87 00:11:56.910 ⇒ 00:11:59.519 Uttam Kumaran: to I’m happy to chat about that.
88 00:11:59.520 ⇒ 00:12:19.000 Payas Parab: Yeah, I’m on. I curious what? I sent a message this morning in the other chat. I know there’s a couple of chats now with like regarding the assumptions sheet that we Ryan, has built the connector to Ryan Dash. Luke has built the connector for 5 Tran. So we have the sheet set up based on
89 00:12:19.160 ⇒ 00:12:25.279 Payas Parab: like, how I’m blittered was doing it before. We don’t know if that’s the best way we want to set up the spreadsheet, but I pulled the most
90 00:12:25.290 ⇒ 00:12:35.589 Payas Parab: commonly used skews from the last few months, and like they’re ordered on there by that, and then I have like a default assumption. I can share that spreadsheet. I don’t know if you saw it. I can pull it up real quick if that’s helpful.
91 00:12:35.940 ⇒ 00:12:39.280 Aman Nagpal: I can you pull it up? I saw your message. I haven’t looked at the sheet yet.
92 00:12:39.280 ⇒ 00:12:41.679 Payas Parab: No worries. Let me just pull that up here.
93 00:12:46.980 ⇒ 00:12:54.999 Payas Parab: Yeah. So the only reason I just wanted to flag this earlier rather than later is like we’re building views off of this right like. Now, I’ve incorporated this
94 00:12:55.290 ⇒ 00:13:00.406 Payas Parab: cogs these cogs metrics into our like Meta base dashboard.
95 00:13:01.480 ⇒ 00:13:10.346 Payas Parab: like, if we have to change the columns. Adjust the structure of this like, it’s just kind of like, we just have to like redo the connection every time. So I just wanna make sure
96 00:13:10.770 ⇒ 00:13:33.420 Payas Parab: we sort of get it right as much as possible. These are the most commonly used skews. And then these were the attributes that we were using an amplitude. So I believe they were like some type of logic that we found that was filling it into amplitude. But for 3 pl. Like pick and pack is here, and then product cost. We have here. We also have product costs that comes directly from shopify. So we can
97 00:13:33.570 ⇒ 00:13:37.658 Payas Parab: either. If you want to like, manually update these cogs here.
98 00:13:38.690 ⇒ 00:13:45.810 Payas Parab: or if we want to do it somewhere else, like we’re welcome. We can do that as well. But we have this connected to 5 train, and then we can.
99 00:13:45.970 ⇒ 00:13:48.680 Payas Parab: We have it pulling into the orders.
100 00:13:48.690 ⇒ 00:13:50.320 Payas Parab: Order database.
101 00:13:50.694 ⇒ 00:14:04.189 Payas Parab: we’re working on that now, so that, like every orders cost based on this sheet, will be filled into our orders. Table. But we want to make sure this is the right format. And like we have the right understanding of like, we want to look at it by skew.
102 00:14:04.200 ⇒ 00:14:11.399 Payas Parab: The sheet is something that like you want to update and maintain. I know, Jared said. Google. Sheets is the primary method that would be best. But
103 00:14:11.510 ⇒ 00:14:15.869 Payas Parab: it’s still like it’s not like a. It’s not a trivial thing. So we wanna make sure that this is like.
104 00:14:15.880 ⇒ 00:14:21.459 Payas Parab: the way you guys want to kind of do this. Yeah, I yeah, just let me know.
105 00:14:22.910 ⇒ 00:14:42.930 Aman Nagpal: Yeah, no, this. This looks great. The only thing that comes to my well, so shopify. Like I mentioned, we were inputting the cogs there every time I created a new product. But definitely not for old products before I joined the team in March. So there’s definitely a gap there. But Jerry said he wanted to do cogs within
106 00:14:43.544 ⇒ 00:14:49.145 Aman Nagpal: the product cost within the sheet, so that that works fine. We probably just stop pulling from shop file.
107 00:14:49.440 ⇒ 00:14:50.130 Payas Parab: Bye.
108 00:14:50.130 ⇒ 00:14:51.790 Aman Nagpal: The manual.
109 00:14:51.790 ⇒ 00:14:52.360 Payas Parab: Yeah.
110 00:14:53.540 ⇒ 00:14:55.840 Aman Nagpal: How is it pulling from shopify? By the way.
111 00:14:56.982 ⇒ 00:15:00.780 Payas Parab: So every I order order has
112 00:15:01.370 ⇒ 00:15:07.510 Payas Parab: like multiple items in there and then that item is tied to a skew which has a product cost saved
113 00:15:07.650 ⇒ 00:15:11.910 Payas Parab: in shopify, which we we kind of viewed in the portals, like what you guys are.
114 00:15:11.910 ⇒ 00:15:15.479 Aman Nagpal: Yeah, man, how how is is this getting updated from shopify, too or no? This is.
115 00:15:15.480 ⇒ 00:15:17.319 Payas Parab: This is just like these are dummy numbers, I put in.
116 00:15:17.320 ⇒ 00:15:23.539 Aman Nagpal: Oh, okay, okay, yeah. Sorry. I misunderstood. Yeah, I think we can just pull it from here. The 3 pl assumptions.
117 00:15:26.520 ⇒ 00:15:31.159 Aman Nagpal: is it. So all the numbers you have here I miss. Is it dummy data, or is it different.
118 00:15:31.160 ⇒ 00:15:38.669 Payas Parab: It’s so. It’s remember, like amplitude had that like like this little like, there’s this like bit of code. I can’t remember the exact
119 00:15:38.930 ⇒ 00:15:44.630 Payas Parab: like for every event it would like. Add these fields in automatically based on the product.
120 00:15:44.640 ⇒ 00:15:46.990 Payas Parab: We pulled it from there, for now.
121 00:15:49.110 ⇒ 00:16:01.900 Aman Nagpal: Okay cause the flow that we had that sends order data to amplitude. The flow within shopify it had that whole formula right where it’s based on weight. And then it’s X amount. Whatever
122 00:16:01.980 ⇒ 00:16:03.190 Aman Nagpal: is it?
123 00:16:04.620 ⇒ 00:16:13.090 Aman Nagpal: The what you have here could be the correct way to do it? I guess it depends on what Jonathan and Jared want to do. I was under the assumption. We would just put
124 00:16:13.180 ⇒ 00:16:15.980 Aman Nagpal: pretty much whatever’s in the flow.
125 00:16:16.080 ⇒ 00:16:19.769 Aman Nagpal: Move it over here in a sheet and just update those numbers so that.
126 00:16:19.770 ⇒ 00:16:25.987 Payas Parab: Yeah, that. That’s what this is. This. This is like more detailed breakdown of like what was actually inside there. And Lou,
127 00:16:26.520 ⇒ 00:16:32.160 Payas Parab: you pulled these right from that logic like these fields. These columns came directly from there right.
128 00:16:32.160 ⇒ 00:16:33.879 Luke Daque: Right. Yep, and then we join.
129 00:16:33.880 ⇒ 00:16:38.470 Aman Nagpal: But like to update base cost. I would have to do it on a per item basis this way, right?
130 00:16:39.420 ⇒ 00:16:41.210 Payas Parab: Yeah, so.
131 00:16:41.210 ⇒ 00:16:42.170 Aman Nagpal: Right, so.
132 00:16:42.170 ⇒ 00:16:42.809 Payas Parab: Am I sure? But.
133 00:16:42.810 ⇒ 00:16:55.195 Aman Nagpal: We were to just have one row that says base cost. And if that’s the same for every product, if then, it’d be easier to just have one row right and just for every individual like
134 00:16:55.870 ⇒ 00:17:00.669 Aman Nagpal: you know, basically, all the assumptions that we had in that flow moved over exactly as one row each.
135 00:17:00.670 ⇒ 00:17:01.520 Payas Parab: Yeah, yeah.
136 00:17:02.049 ⇒ 00:17:24.779 Aman Nagpal: And see if you know, Jonathan, I I feel like that’s easier to maintain rather than this. If they say No, this is good. Then we can go this route right? I I’m assuming they’ll go the other way where you know, if base is the same, we’ll just put have one base cell. If it’s, you know, X amount per weight will include the per weight, you know. Cost maybe like 15 rows there.
137 00:17:25.296 ⇒ 00:17:27.909 Aman Nagpal: That way. It’s a lot easier to handle
138 00:17:28.409 ⇒ 00:17:32.609 Aman Nagpal: but yeah, let’s let’s get you connected. In a slack message.
139 00:17:33.287 ⇒ 00:17:39.650 Aman Nagpal: With Jonathan and Jared, and maybe even Justin, would it be quick to just
140 00:17:41.460 ⇒ 00:17:42.949 Aman Nagpal: mock up a few
141 00:17:43.370 ⇒ 00:17:47.719 Aman Nagpal: rows with the other method, and just let them pick. Which way is that quick.
142 00:17:48.490 ⇒ 00:17:51.360 Payas Parab: Yeah. Other method being like this sort of like
143 00:17:51.470 ⇒ 00:17:54.500 Payas Parab: sort, some sort of like logic that basically
144 00:17:54.740 ⇒ 00:17:57.340 Payas Parab: like based on weight based on whatever that.
145 00:17:57.340 ⇒ 00:18:07.170 Aman Nagpal: Not even the logic, I think I mean a little bit on the weights, but pretty pretty much every column you have here, box, price bubble. Just that’ll be its own row. And they just put input it once.
146 00:18:07.850 ⇒ 00:18:14.609 Aman Nagpal: Yeah. So something easy like that without executing it, and then let them pick which way, and if we go that way, then we execute like
147 00:18:14.710 ⇒ 00:18:16.340 Aman Nagpal: per weight times what.
148 00:18:16.340 ⇒ 00:18:23.450 Payas Parab: Sure that that I can do that. That’s like an Excel formula, too. I just want to confirm utam. And Ryan, if I put in a formula in there.
149 00:18:23.620 ⇒ 00:18:27.090 Payas Parab: The 5 tran just pulls the values right? Or would that break it?
150 00:18:27.090 ⇒ 00:18:27.680 Luke Daque: Yeah, it should.
151 00:18:27.680 ⇒ 00:18:29.880 Uttam Kumaran: No, it’s gonna pull. It’s gonna pull the values. Yeah.
152 00:18:29.880 ⇒ 00:18:31.029 Payas Parab: Okay, so, yeah, so.
153 00:18:31.030 ⇒ 00:18:41.879 Uttam Kumaran: You can mock this up. Yeah, you can mock this up, however, is easiest using formulas, but it’ll just pull the value. You can also pie. So you can make one. That’s like the input. And then you can make one for the Etl.
154 00:18:42.030 ⇒ 00:18:43.499 Uttam Kumaran: Also, you know what I mean.
155 00:18:43.500 ⇒ 00:18:46.270 Payas Parab: Probably do. Yeah, that’s probably what I’ll do. Okay.
156 00:18:47.230 ⇒ 00:18:52.349 Aman Nagpal: So we can add an option to this. Let them compare this sheet versus that sheet. Let them pick.
157 00:18:52.350 ⇒ 00:19:04.530 Payas Parab: Sure. Yeah, yeah, we can have like a default price. Or we can do like by weight. We can have like a grading like, I can make a little table that’s like weights, and then there’ll be formulas that fills in the sheet that actually connects to the Etl.
158 00:19:04.772 ⇒ 00:19:22.459 Payas Parab: Yeah, I think I think that’s doable. That’s that. I can do that right now and then I’ll send that to you. I’m on right after this call, and just be like, hey, here’s a simplified version of those assumptions. It feeds into this other table which you can view, and that’s what’s connected to 5 tran we can do that. That would be probably the best way. Yep. Sounds good.
159 00:19:22.670 ⇒ 00:19:27.740 Aman Nagpal: Thank you. That sounds great. Yeah, I I wouldn’t spend too much time just a quick mock up, I guess, for them. And if they.
160 00:19:27.740 ⇒ 00:19:28.060 Payas Parab: Yeah.
161 00:19:28.060 ⇒ 00:19:28.990 Aman Nagpal: We can go through.
162 00:19:28.990 ⇒ 00:19:34.390 Payas Parab: And and so a lot of these things are relatively fixed. Right? It’s just like at the product level. There’s something specific.
163 00:19:35.610 ⇒ 00:19:47.720 Aman Nagpal: The cogs is was my understanding of it. And then, you know, shipping would be dependent on the weight of the item which I’ve been inputting into shopify, too, but maybe we will want the weights in this sheet as well.
164 00:19:48.003 ⇒ 00:19:57.900 Aman Nagpal: Per product. So I’ll kind of leave that up to them how they want to decide to do it. But yeah, that those are the only 2 items that I would think should be on a per
165 00:19:58.000 ⇒ 00:19:59.340 Aman Nagpal: product basis.
166 00:19:59.340 ⇒ 00:20:03.010 Payas Parab: Got it. And the rest is, yeah, kind of fixed. Yeah, okay.
167 00:20:03.170 ⇒ 00:20:04.670 Payas Parab: makes sense. Okay.
168 00:20:04.898 ⇒ 00:20:08.439 Aman Nagpal: Robert, I I saw your message. Do you want to go into the real convo now? Quickly.
169 00:20:09.690 ⇒ 00:20:11.119 Robert Tseng: Yeah, sure, we can do that.
170 00:20:11.770 ⇒ 00:20:12.690 Robert Tseng: So.
171 00:20:12.690 ⇒ 00:20:17.180 Aman Nagpal: So I just wanted to. One get a sense of you know what we think
172 00:20:17.440 ⇒ 00:20:20.440 Aman Nagpal: rail would cost on a monthly basis. But
173 00:20:20.885 ⇒ 00:20:25.610 Aman Nagpal: yeah, my, my whole point wasn’t necessarily the real aspect of that. It was just
174 00:20:25.840 ⇒ 00:20:43.860 Aman Nagpal: whatever we’re using, what we, whether it’s on the table side 1st or wherever it is. Should we just try to get all the data in and then organize and decide later on what we want to show rather than only, you know, pulling certain data and trying to organize it. Cause. I know
175 00:20:43.880 ⇒ 00:20:56.460 Aman Nagpal: Ryan mentioned that it can get cluttered, but I feel like, earlier in the stages in the process. It’s I would think it’s okay if it’s cluttered as long as in amplitude. You know, we’re not going to be cluttered over there. For example, if we end up there.
176 00:20:56.710 ⇒ 00:21:02.809 Uttam Kumaran: Yeah, Ryan, can you specify what you meant by like, it’s gonna get cluttered. I guess I was just not following on what you meant by that.
177 00:21:03.070 ⇒ 00:21:10.680 Luke Daque: Yeah, I think I just I would just meant like there, there’d be a lot of dimensions to look into, like all the if we add all the fields, from all the
178 00:21:11.190 ⇒ 00:21:14.260 Luke Daque: tables, like from the tickets, messages.
179 00:21:14.470 ⇒ 00:21:17.180 Luke Daque: macros, everything recents.
180 00:21:18.170 ⇒ 00:21:21.880 Luke Daque: Yeah, just just that aspect where it’s like there’s a million dimensions.
181 00:21:21.880 ⇒ 00:21:23.300 Uttam Kumaran: 30 or 40. Okay.
182 00:21:23.440 ⇒ 00:21:29.160 Uttam Kumaran: yeah, yeah, we could. I mean, yeah, I mean, I think we could go either way pretty much. What we’ve been doing is just like
183 00:21:29.230 ⇒ 00:21:42.879 Uttam Kumaran: we, we get like the 5 or 10 questions among that you had initially. And then we kind of form the data model that way. But if we want to buy store just like, get everything in there, then do a review with you, and then be like cut back from there, because some of the stuff is like
184 00:21:43.100 ⇒ 00:21:47.835 Uttam Kumaran: is really either like so narrow, or it’s just may not be useful.
185 00:21:48.220 ⇒ 00:21:50.990 Uttam Kumaran: But if we want to take the stance of like.
186 00:21:51.090 ⇒ 00:21:53.809 Uttam Kumaran: create the widest data model and then
187 00:21:53.830 ⇒ 00:21:56.520 Uttam Kumaran: cut it out from there. We could do that, too.
188 00:21:57.270 ⇒ 00:21:58.499 Uttam Kumaran: I feel like that’s
189 00:21:58.790 ⇒ 00:22:01.420 Uttam Kumaran: that doesn’t change much on the process side.
190 00:22:01.690 ⇒ 00:22:05.989 Aman Nagpal: Yeah, if you guys are good with it. I mean, my thought process is.
191 00:22:06.010 ⇒ 00:22:12.379 Aman Nagpal: let’s take all the data that’s available to us, even if we don’t use half of it. Because at the end of the day.
192 00:22:12.940 ⇒ 00:22:25.249 Aman Nagpal: once we hit amplitude or whatever it is, whatever the end result is, we’re going to pick and choose in our charts and dashboards what events to focus on. So that it’s okay if we don’t use half of them.
193 00:22:26.600 ⇒ 00:22:39.340 Aman Nagpal: because, you know, we’ll be able to organize later on as long as it doesn’t affect cost. Right? So if we’re already pulling that data in 5 trend. Then I’m sure there’s not gonna be any cost difference on that side. But is there
194 00:22:39.510 ⇒ 00:22:48.470 Aman Nagpal: more work to be done on your side? Where, you know, is it? Just? Here’s all the data, and it throws it into tables. And then you work on the ones that we’re going to use? Or
195 00:22:48.730 ⇒ 00:22:51.149 Aman Nagpal: do you have to work on all of them right?
196 00:22:51.150 ⇒ 00:22:52.630 Uttam Kumaran: It’s yeah. It’s
197 00:22:52.770 ⇒ 00:22:55.850 Uttam Kumaran: more towards the latter, like we can make some.
198 00:22:56.530 ⇒ 00:23:06.409 Uttam Kumaran: We can. It’s it’s more towards the latter, which is like, if we bring in all those fields, we do have to model all of them like, for example, if there are, if they’re just like 5 fields.
199 00:23:06.450 ⇒ 00:23:24.110 Uttam Kumaran: we have to figure out like what they are and like clean it up and make it available and real. So that’s the real like time thing. So, and and and I think that’s kind of like what Robert was hitting on is like, basically, what we do is we get up all this data from gorgeous. We have to join. We have to clean up columns. But then make sure you have like
200 00:23:24.110 ⇒ 00:23:45.519 Uttam Kumaran: this final reporting table. Then we kind of hand that supplies for modeling and Meta base. There are stages before that that we. And when we were just bringing in these connectors, we make some decisions on like, okay, let’s just like, go towards the key things. We’ll if we miss a couple of things over here, we’ll add them. As we find that out, we can go really wider. But it does take us to go model that data.
201 00:23:46.860 ⇒ 00:23:49.240 Uttam Kumaran: you know. And it’s not all cut and dry. It’s it’s
202 00:23:49.360 ⇒ 00:23:56.649 Uttam Kumaran: it changes based on the data source. Also, some of the stuff we’re really, really familiar with some of the stuff. Of course, there’s stuff that’s that’s
203 00:23:56.930 ⇒ 00:24:03.360 Uttam Kumaran: specific to you guys that we’ve had to figure out. So it’s not as cut and dry. But there is some more work associated.
204 00:24:04.348 ⇒ 00:24:23.740 Aman Nagpal: No, that that makes sense, and I guess that’s you know. Now I understand a little bit more. What about what Robert was hitting on, too. So maybe what we should do is obviously we don’t wanna waste a ton of hours on things that we don’t need if for every source. We have a sort of list at the beginning, or maybe even now that hey, this is all the data that’s available.
205 00:24:23.740 ⇒ 00:24:42.359 Aman Nagpal: We’ve picked these out of the ones we’re not doing. Do you want any of these as well? And that way, Justin and I can look at it and say, Hey, look! No, this might be useful, even though we haven’t discussed it. Let’s throw this in there also. So just having a list of all the available data and then letting us kind of have that final say of, let’s include this, or let’s not include this after you guys go through it
206 00:24:42.540 ⇒ 00:24:43.790 Aman Nagpal: is that pretty quick.
207 00:24:47.680 ⇒ 00:25:06.749 Uttam Kumaran: Yeah. And like, what did you think of like the meeting structure that you had with with Luke? Because I I sometimes I find just getting it able to just see the stuff allows you to say like we need that. We don’t need that. Oh, that should be cleaned up. So that’s why I was like I’d rather have that meeting, and maybe what we can do is try to have that meeting earlier on, or I’ll kind of let
208 00:25:07.127 ⇒ 00:25:19.209 Uttam Kumaran: Luke. You can decide kind of like when to have them, but maybe we do more of that instead of like refining and refining it, and then getting it to a final spot. So that’s a little bit of process change. And then, in terms of the the sources we do have.
209 00:25:19.550 ⇒ 00:25:29.809 Uttam Kumaran: Some of that already in in notion, which is like we bring in a source. And and we’re bringing in what information we get out of it. We can probably try to make that a little bit more clear.
210 00:25:30.830 ⇒ 00:25:48.188 Aman Nagpal: Yeah, no, I mean, if if I missed it that’s my bad. If if you guys can just link me to it. And just a full list of what’s available. For, for example, in gorgeous like, we, we didn’t go too much in detail to into the tables today we kind of looked at real and we noticed, you know. Oh, we have agent.
211 00:25:49.330 ⇒ 00:26:12.099 Aman Nagpal: I think. Well, we have customer name, but not customer email. And I feel like that’s definitely an important one that you know definitely should be there. Right? So I’m wondering, oh, is there other fields that gorgeous offers that maybe you guys didn’t think is important. But Justin might say, Oh, no, we wanna do a chart for this, right? So happy to have a call for any sources, and just go over the tables.
212 00:26:12.337 ⇒ 00:26:19.470 Aman Nagpal: But I think just having that list to at least look, offline, and say, do our little check marks, and then maybe we can have that call, or whatever works.
213 00:26:19.960 ⇒ 00:26:47.559 Uttam Kumaran: Okay, yeah, I think, Luke, it’s probably some level of like when we get the 5 training, we have the Erd, we basically just can shove that into notion. Or you can just extract that and put all the columns in notion. Some of the columns aren’t as obvious as like customer email. It’s like things were like, Oh, this, we have to go look at the values and then find it out. But at least we can. We can have a place where you can go see everything, and then again, a lot like for for Amazon and shopify. There’s literally
214 00:26:47.730 ⇒ 00:26:51.750 Uttam Kumaran: like there’s literally like probably 10,000 fields. So
215 00:26:51.790 ⇒ 00:27:04.759 Uttam Kumaran: you know, it’s some of the stuff is is worth it. Some of it is like it’s like components of components of random things. So we do make assumptions there, but on something like gorgeous and even recharge. The data model is a little bit smaller.
216 00:27:04.880 ⇒ 00:27:16.517 Uttam Kumaran: so we can just have everything listed there, and you could take a look. And then, for the most part, if there’s anything in the tools that you’re like, how do we get this like. I think we’ll we’ll we can actually probably circle back and talk about
217 00:27:17.030 ⇒ 00:27:29.229 Uttam Kumaran: the macros. But if there’s anything like that where you’re like, it’s not here. How do we go get it? If it’s in the tool, they probably have it in the Api somewhere. So that’s how we that’s our mentality on like going and finding those.
218 00:27:30.120 ⇒ 00:27:35.559 Aman Nagpal: Sounds great. Robert, anything you wanna add to that or any questions on the real side.
219 00:27:35.840 ⇒ 00:27:49.829 Robert Tseng: Yeah, no. I think the guys summarize it. Well, but yeah, I think, yeah, definitely makes sense. If we can get ahead of the yeah. For on our end, when we’re ingesting a source just to have that conversation earlier. So we know, like.
220 00:27:50.200 ⇒ 00:28:09.219 Robert Tseng: Hey, like, these are the ones that we want to be focusing on. I think that that helps us, because definitely, I think well, the the work on recharge, which is kind of like took that in, and then have been like asking afterwards. But I think maybe like moving forward, we can have more of those proactive calls. If that’s if that’s easier than just like messaging back and forth
221 00:28:09.510 ⇒ 00:28:12.629 Robert Tseng: on on what? Which fields we want to be be using.
222 00:28:13.640 ⇒ 00:28:16.029 Aman Nagpal: Yeah, no, that works, too. I think
223 00:28:16.090 ⇒ 00:28:22.263 Aman Nagpal: I know some cases like you said with them, you know, if there’s 10,000 columns. It’s gonna be a little bit harder. But
224 00:28:22.550 ⇒ 00:28:27.849 Aman Nagpal: at the same time, you know, I mentioned, for Justin is extremely, you know.
225 00:28:27.900 ⇒ 00:28:46.019 Aman Nagpal: detailed data heavy, and he wants as much data as possible. So whatever we can have a list for if I go through at the beginning before you know, you guys jump into it. That’s that’s probably the the best solution. And in terms of real I know, Robert, you mentioned that Jared wasn’t crazy about using real. But
226 00:28:46.368 ⇒ 00:28:58.800 Aman Nagpal: I mean, it depends on cost, right? So if it if it’s if we have this self service tool, that maybe we can’t do in Snowflake. But it’s easy for me to see data in real, and it’s dirt cheap. Then we’ll keep it right. If it’s
227 00:28:58.820 ⇒ 00:29:13.500 Aman Nagpal: pricey, then we might not want to if it takes a lot of time to get data in there. So what’s that kind of like in terms of monthly costs on the real side. But also, how much time does it take to actually get data into real that could be used for something else?
228 00:29:15.750 ⇒ 00:29:25.379 Robert Tseng: So I mean, we can kind of talk about the the cost side. But I mean right now, we’re just we’re just running it like locally, right? So we’re not really or like, I guess we have a cloud dashboard right now.
229 00:29:25.380 ⇒ 00:29:54.274 Uttam Kumaran: We have a cloud dashboard. But like, I have a relationship with the real guys. So I’m basically like, Hey, we’re just getting you on boarded kind of the whole time we were saying, Okay, is it going to be Meta base, or this? I know the other crew wasn’t too jazz about it. But also I think you kind of see like how valuable it’s been, just because, like, it’s super easy to see all the columns instead of having to go write a sequel. Query? That being said it it is. It’s there’s a platform fee. It’s actually it’s actually more based on the
230 00:29:54.650 ⇒ 00:30:17.150 Uttam Kumaran: like amount of data, but it is quite cheap and then like in in the hundreds of dollars a month. And then but the real, the real thing that’s different is that we have to spend time maintaining so kind of like here. I mean, cause we have. We’ll have Meta Base real. The amplitude work. Dbt, and 5 trend right? So it just adds another tool that although I know we are doing it now.
231 00:30:17.420 ⇒ 00:30:36.170 Uttam Kumaran: all this work we do has to maintain and updates. So as we start to have more dashboards and things, it’s not gonna all be as net new as it is now. It’s gonna be a lot of maintenance and additions. So we’re we just probably with the allotted like 10 h we have right now. It’s gonna it’ll end up being a challenge for us to
232 00:30:36.250 ⇒ 00:30:38.409 Uttam Kumaran: continue to do real
233 00:30:39.250 ⇒ 00:30:49.940 Uttam Kumaran: in addition, with with everything else that we’re doing. So so for me, the biggest thing would be if we want to maintain that it would take at least a few more hours per week. To handle that
234 00:30:50.319 ⇒ 00:30:53.209 Uttam Kumaran: but it, I would say, like. I don’t know my
235 00:30:53.730 ⇒ 00:31:03.720 Uttam Kumaran: bias opinion, or whatever I think. It’s a great tool, I think, even for us to do debugging. It’s a really really amazing tool like. We will probably maintain some version of it
236 00:31:03.900 ⇒ 00:31:05.110 Uttam Kumaran: regardless.
237 00:31:05.413 ⇒ 00:31:13.900 Uttam Kumaran: Just cause. It helps us, like Debug, between pious and and Luke, and just like understand with the shape of the data. But of course, it’s another bi tool to maintain.
238 00:31:14.900 ⇒ 00:31:22.320 Aman Nagpal: Yeah, I mean even taking your hours out of it, if it’s if it was like 50 bucks a month. Fine, you know it is what it is. If it’s a couple of 100, then that’s a little bit more to
239 00:31:22.360 ⇒ 00:31:26.079 Aman Nagpal: excuse me. Think about but then you add, in the time you’re spending
240 00:31:26.759 ⇒ 00:31:36.459 Aman Nagpal: that. Obviously, we would prefer to go to our quote unquote, final solution rather than this intermediary that we’re kind of just using just to poke around for ourselves.
241 00:31:36.918 ⇒ 00:31:47.309 Aman Nagpal: If if we don’t have real. And I wanted to. Just look at the data, you know. There’s no way snowflake for me to do something similar. Right? I would need to write sequel query.
242 00:31:47.990 ⇒ 00:32:01.670 Uttam Kumaran: Yeah, you, I mean, you could click on the table and basically see the contents fairly easily. Or, again, if you hop on a call with one of us, we can put that together. It’s not that hard to write a simple like select query. And just look at a table.
243 00:32:01.890 ⇒ 00:32:02.480 Aman Nagpal: Yeah. But of course.
244 00:32:02.480 ⇒ 00:32:03.820 Uttam Kumaran: It’s not as
245 00:32:04.340 ⇒ 00:32:06.597 Uttam Kumaran: it’s not as easy as real makes it.
246 00:32:07.150 ⇒ 00:32:24.380 Payas Parab: There are some in even in Meta base, too. There are some like capabilities as well where you like, and I could quickly show you like, we’re trying to get away from having that built in sequel for you guys to ensure that like, it’s a click experience, you know. So there are ways to kind of do like, hey? I just want like
247 00:32:24.430 ⇒ 00:32:28.979 Payas Parab: a sum of this. And I want it grouped by this. And I want it joint with this data. It’s
248 00:32:29.030 ⇒ 00:32:34.790 Payas Parab: a click interface. At least, so it’s not like a SQL. Query. But the logic is very SQL like. So
249 00:32:34.880 ⇒ 00:32:42.600 Payas Parab: that’s also an option as well. There’s like some self serve components to metabase. But it’s typically better done by someone who understands the SQL logic.
250 00:32:43.390 ⇒ 00:32:51.519 Aman Nagpal: If you guys don’t mind. Could we take a quick look at Snowflake and Meta Base, and just see what that would be like just clicking around without writing anything.
251 00:32:52.620 ⇒ 00:33:00.400 Uttam Kumaran: Yeah, price. Maybe if you want to do that, I know, I definitely wanna also, I think we found some stuff on the macro side. So I want to go through that as well.
252 00:33:01.810 ⇒ 00:33:03.850 Uttam Kumaran: and so we can go through it. And maybe.
253 00:33:04.150 ⇒ 00:33:06.140 Uttam Kumaran: Luke, do you want to do the macro stuff.
254 00:33:06.140 ⇒ 00:33:06.680 Luke Daque: Yeah, sure.
255 00:33:06.680 ⇒ 00:33:11.270 Uttam Kumaran: First.st Just so we get that out of the way, and then if we have time, we can go through that, or we can grab some more.
256 00:33:13.100 ⇒ 00:33:17.200 Aman Nagpal: I have a hard cut off at 2 45. But let’s go through any of it.
257 00:33:17.200 ⇒ 00:33:17.670 Uttam Kumaran: Okay.
258 00:33:17.670 ⇒ 00:33:20.120 Luke Daque: Yeah, I actually just added it to the
259 00:33:20.150 ⇒ 00:33:22.389 Luke Daque: real locally, for now.
260 00:33:23.930 ⇒ 00:33:26.918 Luke Daque: yeah, let me share my screen. Can you see my screen?
261 00:33:27.780 ⇒ 00:33:33.749 Luke Daque: Yeah, I just added the ticket macros over here. We we noticed it’s actually one of the
262 00:33:35.000 ⇒ 00:33:44.120 Luke Daque: fields in messages like, there’s a macro field that has a list of the macro ids. So we use that to to to join
263 00:33:44.570 ⇒ 00:33:49.599 Luke Daque: to the messages. So if ever a ticket has a message that is tied to a macro.
264 00:33:50.100 ⇒ 00:33:53.969 Luke Daque: Then, yeah, you’ll be able to see like what Macro
265 00:33:54.350 ⇒ 00:33:56.059 Luke Daque: name was used.
266 00:33:56.740 ⇒ 00:34:08.480 Aman Nagpal: So if I go to ticket, you know by ticket Id. I find all the list of all the messages, and that messages, let’s say, object has the actual message, but it also has
267 00:34:08.510 ⇒ 00:34:14.979 Aman Nagpal: probably the date it was sent, but it also has the macro that was used for that message. If there was a macro used.
268 00:34:15.590 ⇒ 00:34:16.710 Luke Daque: Exactly.
269 00:34:16.710 ⇒ 00:34:19.320 Aman Nagpal: That’s that’s all we need. I think that takes care of it.
270 00:34:19.870 ⇒ 00:34:25.109 Uttam Kumaran: And then, yeah, Luke, you should. If we maybe we can clean up the actual field.
271 00:34:25.110 ⇒ 00:34:27.649 Luke Daque: Yeah, like the black one.
272 00:34:27.650 ⇒ 00:34:30.300 Uttam Kumaran: And I don’t know it’s coming in as like a list or something.
273 00:34:30.690 ⇒ 00:34:34.280 Luke Daque: Yeah, cause there could be like multiple macros in a a single ticket, though, that.
274 00:34:34.280 ⇒ 00:34:35.330 Uttam Kumaran: That’s why.
275 00:34:35.330 ⇒ 00:34:36.870 Luke Daque: I made it into list.
276 00:34:37.040 ⇒ 00:34:38.359 Luke Daque: Okay, okay.
277 00:34:38.360 ⇒ 00:34:39.350 Uttam Kumaran: Okay. Cool.
278 00:34:40.290 ⇒ 00:34:44.279 Luke Daque: Yeah, I did see a couple of tickets that were like that hadn’t portable
279 00:34:44.500 ⇒ 00:34:45.520 Luke Daque: macros.
280 00:34:45.820 ⇒ 00:34:48.149 Aman Nagpal: Each message only has one macro right?
281 00:34:48.600 ⇒ 00:34:49.850 Luke Daque: There.
282 00:34:50.989 ⇒ 00:34:56.800 Luke Daque: I’m not sure like this one. Yeah, this one, this. This is just one message, but it has like 2 macros in it.
283 00:34:57.360 ⇒ 00:35:02.029 Aman Nagpal: That’s interesting cause. I know. Obviously each ticket can have multiple macros. But I wonder how
284 00:35:02.330 ⇒ 00:35:03.889 Aman Nagpal: one message
285 00:35:04.270 ⇒ 00:35:09.410 Aman Nagpal: can have multiple macros unless message Id refers to the ticket, and not one.
286 00:35:09.410 ⇒ 00:35:11.899 Luke Daque: No, this is this is a.
287 00:35:11.900 ⇒ 00:35:12.230 Uttam Kumaran: Maybe.
288 00:35:12.230 ⇒ 00:35:12.685 Luke Daque: Terrific.
289 00:35:13.400 ⇒ 00:35:15.339 Luke Daque: Let me add the ticket. Id.
290 00:35:16.280 ⇒ 00:35:18.410 Aman Nagpal: Ticket. Id is separate. Message. Id, right? Yeah.
291 00:35:18.410 ⇒ 00:35:19.630 Luke Daque: Yeah, it’s a different.
292 00:35:20.440 ⇒ 00:35:22.950 Uttam Kumaran: You should pull it up in gorgeous. We could look at this one.
293 00:35:24.200 ⇒ 00:35:26.499 Luke Daque: Yeah, I don’t have access to gorgeous, though.
294 00:35:26.775 ⇒ 00:35:28.430 Aman Nagpal: Share my screen again. Let’s see.
295 00:35:28.430 ⇒ 00:35:30.429 Uttam Kumaran: Yeah, maybe just send this one in the chat.
296 00:35:30.630 ⇒ 00:35:32.660 Uttam Kumaran: This I, the Id. We’re looking at.
297 00:35:32.660 ⇒ 00:35:34.489 Luke Daque: That ticket. Id, I guess right.
298 00:35:34.490 ⇒ 00:35:35.539 Uttam Kumaran: With the 3.
299 00:35:36.040 ⇒ 00:35:38.590 Luke Daque: Yeah, where was that? This one?
300 00:35:38.960 ⇒ 00:35:41.550 Luke Daque: So this is the ticket. Id. Let me send it here.
301 00:35:41.550 ⇒ 00:35:42.660 Uttam Kumaran: Has 2
302 00:35:43.520 ⇒ 00:35:46.470 Uttam Kumaran: or may, maybe they sent. Yeah, let’s see.
303 00:35:49.710 ⇒ 00:35:50.950 Aman Nagpal: Let me pull this up.
304 00:35:51.580 ⇒ 00:35:52.860 Luke Daque: There’s 2
305 00:35:54.017 ⇒ 00:35:57.120 Luke Daque: let’s see if there’s anything. Yeah, this one’s.
306 00:35:57.550 ⇒ 00:35:58.779 Luke Daque: Oh, this is one.
307 00:35:58.780 ⇒ 00:36:01.229 Aman Nagpal: It says I can’t share my screen.
308 00:36:01.480 ⇒ 00:36:03.063 Aman Nagpal: Oh, let me stop sharing.
309 00:36:04.960 ⇒ 00:36:05.870 Aman Nagpal: There we go.
310 00:36:09.120 ⇒ 00:36:10.560 Aman Nagpal: I can find it.
311 00:36:13.160 ⇒ 00:36:14.070 Aman Nagpal: Okay.
312 00:36:14.240 ⇒ 00:36:15.520 Aman Nagpal: You guys see.
313 00:36:16.400 ⇒ 00:36:17.115 Luke Daque: Yep.
314 00:36:17.830 ⇒ 00:36:21.109 Aman Nagpal: Okay. So this should be the ticket we’re looking at, is it? Deborah
315 00:36:21.470 ⇒ 00:36:22.350 Aman Nagpal: Hareth?
316 00:36:24.720 ⇒ 00:36:28.419 Luke Daque: I don’t know. But yeah, it should be the ticket number. It’s the same ticket number.
317 00:36:28.870 ⇒ 00:36:30.060 Aman Nagpal: Cool, so
318 00:36:32.000 ⇒ 00:36:33.000 Aman Nagpal: I assume.
319 00:36:33.690 ⇒ 00:36:34.399 Luke Daque: Yeah, I can see.
320 00:36:34.400 ⇒ 00:36:38.689 Aman Nagpal: Very least. This is probably a macro, this message.
321 00:36:43.110 ⇒ 00:36:46.209 Aman Nagpal: I wonder if there’s a way to see more info on this?
322 00:37:14.990 ⇒ 00:37:17.419 Aman Nagpal: Okay, I can search by.
323 00:37:28.410 ⇒ 00:37:30.599 Aman Nagpal: Yeah, okay, so this is this macro
324 00:37:31.577 ⇒ 00:37:33.640 Aman Nagpal: I assume this is what was used here.
325 00:37:35.190 ⇒ 00:37:38.150 Aman Nagpal: but I don’t know how that message would fall under 2.
326 00:37:39.100 ⇒ 00:37:43.320 Luke Daque: Yeah, the 1st macro is it’s that one the cancel
327 00:37:43.430 ⇒ 00:37:49.389 Luke Daque: unaware of sub. And then there’s a macro that was cancel 1st request
328 00:37:50.810 ⇒ 00:37:53.420 Luke Daque: 1st last in like one St.
329 00:37:53.640 ⇒ 00:37:55.300 Aman Nagpal: Oh, one.
330 00:37:57.210 ⇒ 00:37:58.599 Luke Daque: Yeah, that one. Yeah.
331 00:38:03.760 ⇒ 00:38:04.460 Aman Nagpal: Hmm!
332 00:38:04.950 ⇒ 00:38:06.540 Aman Nagpal: I don’t see this message.
333 00:38:07.440 ⇒ 00:38:08.580 Luke Daque: Interesting.
334 00:38:11.895 ⇒ 00:38:12.630 Luke Daque: Yeah.
335 00:38:13.050 ⇒ 00:38:16.089 Aman Nagpal: I wonder if it was clicked accidentally
336 00:38:16.520 ⇒ 00:38:17.900 Aman Nagpal: and then changed.
337 00:38:21.290 ⇒ 00:38:22.470 Aman Nagpal: Could that be it.
338 00:38:23.050 ⇒ 00:38:24.950 Luke Daque: Could be a possibility.
339 00:38:25.720 ⇒ 00:38:26.539 Uttam Kumaran: Or maybe
340 00:38:26.950 ⇒ 00:38:30.480 Aman Nagpal: Does it show the order in which the macros were used or no?
341 00:38:31.710 ⇒ 00:38:32.779 Luke Daque: Let’s check.
342 00:38:33.940 ⇒ 00:38:36.760 Uttam Kumaran: I don’t know if it does.
343 00:38:36.760 ⇒ 00:38:37.220 Luke Daque: Yeah.
344 00:38:38.000 ⇒ 00:38:43.200 Uttam Kumaran: So there’s a there’s a macro table and a macro action, Luke. You could probably see if one of those.
345 00:38:47.150 ⇒ 00:38:49.629 Luke Daque: Yeah, that’s a good point. Let’s try that.
346 00:38:52.980 ⇒ 00:38:55.380 Luke Daque: Just need the macro id, though.
347 00:38:58.090 ⇒ 00:39:01.600 Luke Daque: yeah, I’ll I’ll look into that. It might take a while to.
348 00:39:02.700 ⇒ 00:39:03.720 Aman Nagpal: Sure. Yeah.
349 00:39:03.720 ⇒ 00:39:04.600 Luke Daque: To get.
350 00:39:04.780 ⇒ 00:39:14.410 Aman Nagpal: Maybe we can also check if if there’s only like a few that have 2 macros per message, maybe those are just accidental clicks, and we don’t need to worry about it as much
351 00:39:15.447 ⇒ 00:39:21.559 Aman Nagpal: or you know, like you said, maybe the macro actions table has the correct need.
352 00:39:24.853 ⇒ 00:39:29.550 Aman Nagpal: But yeah, let’s if you can take a look at that. But I think we should be good there then. So
353 00:39:30.020 ⇒ 00:39:41.229 Aman Nagpal: now that we pretty much have that, the question is, now, where would we want this to live? Right? So I know Robert had to hop off. But this is something that I want to compare
354 00:39:41.616 ⇒ 00:39:44.390 Aman Nagpal: to churn. Let’s say right? So you know.
355 00:39:44.875 ⇒ 00:39:52.440 Aman Nagpal: what percent of which Macro used leads to higher churn. For example, another one would be
356 00:39:52.450 ⇒ 00:39:53.900 Aman Nagpal: which agents
357 00:39:54.020 ⇒ 00:39:55.180 Aman Nagpal: use
358 00:39:56.348 ⇒ 00:39:59.910 Aman Nagpal: which macros the most. So these charts
359 00:40:00.340 ⇒ 00:40:06.799 Aman Nagpal: our initial assumption would be, it would go within into amplitude. But, you know, is there something else. You guys are thinking.
360 00:40:09.680 ⇒ 00:40:14.359 Uttam Kumaran: I think it’s I think this should end up going into amplitude. But I don’t apply. It’s like, is this on the Meta base side.
361 00:40:15.590 ⇒ 00:40:17.830 Payas Parab: I I honestly think it’s it would.
362 00:40:19.980 ⇒ 00:40:22.240 Payas Parab: So we want to figure out like, just a
363 00:40:22.560 ⇒ 00:40:29.260 Payas Parab: so which Macro they’re using, and then tie that back to like customer that cancels their subscription or.
364 00:40:29.960 ⇒ 00:40:36.119 Aman Nagpal: Yeah, so there, there’s like a few. I think I I sent the list of Nico. But there’s a few things that we want to do right. One is
365 00:40:36.300 ⇒ 00:40:38.359 Aman Nagpal: which Macros.
366 00:40:38.570 ⇒ 00:41:01.349 Aman Nagpal: when they’re used, lead to the highest cancellation rate. Then we know, okay, let’s stop using this macro or let’s adjust it next could be, you know, which reps as we hire. New reps are just canceling immediately. And so maybe there, maybe we’re looking at the cancel, Macro, and if it’s extremely high for a certain agent, then we want to have a talk with them and figure out what’s going on there right.
367 00:41:01.350 ⇒ 00:41:01.830 Payas Parab: Sure.
368 00:41:01.830 ⇒ 00:41:02.590 Aman Nagpal: Policies.
369 00:41:02.590 ⇒ 00:41:03.100 Payas Parab: Yeah.
370 00:41:04.110 ⇒ 00:41:05.420 Aman Nagpal: And
371 00:41:05.460 ⇒ 00:41:11.210 Aman Nagpal: yeah, but the bigger thing is, you know, which Macros lead to higher churn. So
372 00:41:12.040 ⇒ 00:41:13.100 Aman Nagpal: I think that would be.
373 00:41:13.100 ⇒ 00:41:21.600 Payas Parab: That would be like somewhere where we can just query because that data that data set should have the status of whether they churned or not right like whether the customer
374 00:41:21.680 ⇒ 00:41:26.128 Payas Parab: I haven’t looked at the table directly, but there should be like a status right like it would almost be like.
375 00:41:26.800 ⇒ 00:41:55.259 Payas Parab: hey? This was resolved. We lost the customer whatever, and then, like a group by of that, to me, feels like a Meta base type deal that doesn’t necessarily need to be amplitude. The only time we would need to do that is, if it’s connected to some other components of the user journey that have to do with like people on your site, or, like the other, the other events that we have captured in there. Otherwise there’s really no need. If if the data set itself can, like you can resolve stuff within it. Then there’s no need to push it back into amplitude.
376 00:41:56.110 ⇒ 00:42:12.389 Aman Nagpal: I guess I know we keep going back to this conversation. But you know, I thought we were just gonna use Meta base, or whatever it is for financial data and everything else would live with an amplitude, because that’s what we’re used to. We signed a big contract with them for a lot of events. So I mean, is there any benefit
377 00:42:12.500 ⇒ 00:42:18.080 Aman Nagpal: to doing it in metabase as opposed to amplitude? I know we just made another project with them as well.
378 00:42:19.600 ⇒ 00:42:20.620 Payas Parab: Yeah,
379 00:42:23.020 ⇒ 00:42:26.820 Payas Parab: I I may need to just circle back with Robert on that. I don’t want to speak incorrectly. I’d rather
380 00:42:27.200 ⇒ 00:42:35.959 Payas Parab: with him. He’s just more of an expert on amplitude and like data in data out there. And I don’t want to misspeak here. So let me let me write that as a item that I will take yeah.
381 00:42:36.380 ⇒ 00:42:39.470 Aman Nagpal: Yeah, no, that that sounds good. Yeah. I think
382 00:42:39.530 ⇒ 00:42:46.090 Aman Nagpal: the main thing is we’re used to it. All of our data is already going there. We already have so much data there, especially
383 00:42:46.150 ⇒ 00:42:50.459 Aman Nagpal: like all of our clicks and landing page views and all that data, etc.
384 00:42:52.080 ⇒ 00:42:55.961 Aman Nagpal: you know, my assumption was all of that would just go there. How about the
385 00:42:57.110 ⇒ 00:42:59.950 Aman Nagpal: gross margin? I think we’ll we’ll end up in metabase, or whatever.
386 00:42:59.950 ⇒ 00:43:00.550 Payas Parab: Yeah, I am.
387 00:43:00.550 ⇒ 00:43:10.849 Aman Nagpal: Financial tool is right. What about the subscriber dashboard? Were you guys planning on doing that in metabase as well? Or is that something you want to run by, Robert, because I figured that would
388 00:43:11.080 ⇒ 00:43:13.619 Aman Nagpal: maybe be good, and amplitude.
389 00:43:13.620 ⇒ 00:43:31.369 Payas Parab: That one is good in amplitude. Yeah. So I think the subscriber and like retention, just because it’s like the user activity right on the platform is just best captured by those events, and like they have those user flows. And and I believe, like that was the other deliverables as well. We? We have completed right? I believe the subscriber
390 00:43:31.940 ⇒ 00:43:34.300 Payas Parab: analysis in amplitude
391 00:43:34.370 ⇒ 00:43:35.500 Payas Parab: from maybe a month.
392 00:43:35.500 ⇒ 00:43:37.220 Aman Nagpal: I don’t remember where we last left off on that.
393 00:43:37.220 ⇒ 00:43:48.759 Payas Parab: Okay, let’s check back on that. But I that that is a better like if it’s like, drop off in like subscribers, or like we’re trying to kind of track that user journey that does need to sit in amplitude. It’s just kind of more with this, like
394 00:43:48.930 ⇒ 00:44:03.850 Payas Parab: various static data that like also needs to be able to like, move around. And we need to be able to adjust with our own assumptions and run our own calculations. That that stuff is just. Amplitude is not made for that. It’s so. Yeah, the subscriber retention.
395 00:44:04.180 ⇒ 00:44:15.970 Payas Parab: These types of things, I think, will sit in amplitude. But like the reason this, like this, customer service, like what agents are doing, is different is because, like that data set has all the information we need. So we don’t wanna like
396 00:44:16.100 ⇒ 00:44:22.419 Payas Parab: move it into back into amplitude. And then, like connecting that to amplitude right like, how do we tie from
397 00:44:22.450 ⇒ 00:44:30.139 Payas Parab: whatever that customer Id is in that data to back to amplitude, like if we don’t have to do that, let’s not do that right. That’s that’s the other thing.
398 00:44:30.670 ⇒ 00:44:43.029 Aman Nagpal: Okay, yeah, let’s I guess. Let’s we’ll figure out the details on that. We can circle back to it. And I only have a couple of minutes. But is there anything else before we quickly jump into the Meta base or snowflakes?
399 00:44:43.310 ⇒ 00:44:43.590 Payas Parab: Have.
400 00:44:43.590 ⇒ 00:44:45.110 Aman Nagpal: Anything else. You guys wanted to go over.
401 00:44:46.240 ⇒ 00:44:46.910 Payas Parab: I think.
402 00:44:46.910 ⇒ 00:44:47.790 Uttam Kumaran: That’s it. I think
403 00:44:49.470 ⇒ 00:44:52.769 Uttam Kumaran: I think that’s it you can go ahead with if you want to show the Meta base.
404 00:44:52.770 ⇒ 00:44:53.330 Payas Parab: Yeah.
405 00:44:53.330 ⇒ 00:44:53.889 Uttam Kumaran: Oh yes!
406 00:44:53.890 ⇒ 00:44:55.020 Payas Parab: Quickly flash that up there.
407 00:44:55.020 ⇒ 00:44:57.020 Uttam Kumaran: How to do the exploration. Yeah.
408 00:44:57.020 ⇒ 00:44:59.188 Payas Parab: Yeah. So like, this is like.
409 00:44:59.830 ⇒ 00:45:06.240 Payas Parab: yeah, the snowflake is a little bit involved. But snowflakes the best place to kind of like. See what data you have. You can also do that in
410 00:45:06.370 ⇒ 00:45:20.510 Payas Parab: like metabase as well. So if I go in here, I’m in my Meta base dashboard, I’m like, hey? I want to just know what data we have in here. Right? You just stick to prod like, okay, cool. I know I want to do some analysis related to like messages, and I just want to quickly see
411 00:45:20.560 ⇒ 00:45:28.780 Payas Parab: what that data looks like. It’ll sort of like preview it for you here wanted to like sort of summarize. Some of this data you like wanted to see like
412 00:45:28.800 ⇒ 00:45:32.180 Payas Parab: number of messages like, let’s say.
413 00:45:32.290 ⇒ 00:45:38.740 Payas Parab: I want an average you know, total messages in a chat. Right?
414 00:45:39.950 ⇒ 00:45:50.030 Payas Parab: And I want to group it by the date, right? For whatever reason, like, you can sort of do that on your own right without needing to
415 00:45:50.570 ⇒ 00:46:02.790 Payas Parab: like by by week. I want to see what the average number of total messages is like a click interface that you can just kind of like play around and run some of that analysis right? Like it was a few clicks relatively intuitive. You wanna like
416 00:46:03.230 ⇒ 00:46:07.059 Payas Parab: cut the data into like a smaller timeframe like, I just want to see.
417 00:46:07.150 ⇒ 00:46:08.190 Payas Parab: you know.
418 00:46:09.000 ⇒ 00:46:11.240 Payas Parab: whatever previous.
419 00:46:12.590 ⇒ 00:46:14.030 Payas Parab: Sorry. One second
420 00:46:14.670 ⇒ 00:46:27.709 Payas Parab: you just want to see the previous 30 days. Only you can like filter that data easily and see it by week, play around. So like there is some capability here where you can do that like the most common thing is, you just go to this top right? Make a question.
421 00:46:27.720 ⇒ 00:46:36.420 Payas Parab: If you have a sense of like, what data is there right? Like fact? Orders is your main data set right? And you’re like, I want to add a filter, and I only want to see
422 00:46:36.520 ⇒ 00:46:39.249 Payas Parab: orders from Amazon
423 00:46:39.350 ⇒ 00:46:40.610 Payas Parab: and
424 00:46:41.970 ⇒ 00:46:43.460 Payas Parab: add filter.
425 00:46:43.530 ⇒ 00:46:49.299 Payas Parab: I can pick a metric, and I can just see like, let’s say, I want to see that total gross revenue which is some line items price.
426 00:46:49.620 ⇒ 00:46:51.990 Payas Parab: And I want to group it by
427 00:46:52.950 ⇒ 00:46:55.799 Payas Parab: whatever the time period of the order, right
428 00:46:56.070 ⇒ 00:46:57.660 Payas Parab: by month of order.
429 00:46:57.860 ⇒ 00:47:01.619 Payas Parab: you can quickly like visualize that pretty easily.
430 00:47:01.730 ⇒ 00:47:04.820 Payas Parab: You can change the chart type to be like a bar chart.
431 00:47:05.130 ⇒ 00:47:14.690 Payas Parab: So like there are some like, this is pretty self. Serve. I don’t want to like get into. I’m just like kind of playing around and voicing over. But like this is something that’s like a little bit like doable. If you wanted to
432 00:47:15.270 ⇒ 00:47:18.319 Payas Parab: kind of figure it out, you know. Hey? I want to see.
433 00:47:18.580 ⇒ 00:47:21.760 Payas Parab: you know, certain metrics. There’s pivot tables. There’s like
434 00:47:21.980 ⇒ 00:47:25.500 Payas Parab: couple charts. The filter and summarization is
435 00:47:25.770 ⇒ 00:47:28.589 Payas Parab: at least a click interface rather than
436 00:47:29.030 ⇒ 00:47:29.560 Payas Parab: a sequel.
437 00:47:29.560 ⇒ 00:47:41.260 Aman Nagpal: Sorry. I gotta. I gotta hop soon. But this this looks great. I mean, even just the fact. I didn’t even know that the charts. I could do all that easily, but even just need to see all the data. That I think in itself is great.
438 00:47:41.655 ⇒ 00:47:44.710 Aman Nagpal: Is there anything else besides what you showed me that
439 00:47:44.940 ⇒ 00:47:49.700 Aman Nagpal: we were using, we would use real for outside of these items that you just showed me
440 00:47:50.300 ⇒ 00:47:51.430 Aman Nagpal: on metabase. Yeah.
441 00:47:51.970 ⇒ 00:47:53.970 Payas Parab: Yeah, no, not really. Yeah.
442 00:47:55.240 ⇒ 00:47:59.410 Aman Nagpal: Based on that alone can we do any of this in Snowflake without sequel or no. We would need to use SQL.
443 00:47:59.410 ⇒ 00:48:01.279 Payas Parab: You kind of need to use. SQL, it’s a little.
444 00:48:01.280 ⇒ 00:48:04.050 Aman Nagpal: Even just viewing all the data that’s coming in from Fivetran.
445 00:48:04.340 ⇒ 00:48:04.810 Payas Parab: Yeah.
446 00:48:04.810 ⇒ 00:48:07.630 Uttam Kumaran: As a table, but you can’t do like the viz. Really easily.
447 00:48:08.070 ⇒ 00:48:08.969 Uttam Kumaran: but I can.
448 00:48:08.970 ⇒ 00:48:09.930 Aman Nagpal: The data that’s coming in.
449 00:48:09.930 ⇒ 00:48:10.480 Uttam Kumaran: Yeah.
450 00:48:10.480 ⇒ 00:48:11.700 Payas Parab: I can see the data. Yeah.
451 00:48:11.700 ⇒ 00:48:28.439 Aman Nagpal: I think that’s easy. So I’ll I’ll confirm. But I think for now maybe we don’t need to spend too much time on real. Then, if that’s the case, and if that’s what Jared’s saying, and if you guys are on the same page, then maybe we just don’t waste hours on that, and we use the hours for something else, maybe in again, the end result being
452 00:48:28.826 ⇒ 00:48:35.159 Aman Nagpal: amplitude. And then, if we finalize on Meta Base, which we should say at this point, finalize what we’re doing.
453 00:48:36.128 ⇒ 00:48:40.749 Aman Nagpal: In terms of whether it’s metabase or something else for financial data. All that good stuff.
454 00:48:40.750 ⇒ 00:48:41.230 Payas Parab: Hmm.
455 00:48:41.551 ⇒ 00:48:46.690 Aman Nagpal: So we’ll see what Jared says. But yeah, no, thanks. This is this is super helpful.
456 00:48:47.220 ⇒ 00:48:50.740 Aman Nagpal: And just shoot me a slack for anything else. But
457 00:48:50.810 ⇒ 00:48:53.160 Aman Nagpal: I think, yeah, all this looks great.
458 00:48:53.620 ⇒ 00:48:54.749 Payas Parab: Awesome. Thanks. Mark.
459 00:48:55.050 ⇒ 00:48:56.210 Aman Nagpal: Thanks a lot guys.
460 00:48:56.450 ⇒ 00:48:57.900 Uttam Kumaran: Thanks, talk soon.
461 00:48:58.190 ⇒ 00:48:59.370 Luke Daque: Thanks, bye, bye.