Meeting Title: Sucari’s Zoom Meeting Date: 2025-02-26 Meeting participants: Bo Yoon, Nicolas Sucari, Payas Parab
WEBVTT
1 00:01:28.520 ⇒ 00:01:30.120 Payas Parab: Hey, Nico! How are you?
2 00:01:32.320 ⇒ 00:01:34.990 Payas Parab: Anybody asking? All good? How are you?
3 00:01:35.330 ⇒ 00:01:38.060 Payas Parab: Good? Sorry. I keep forgetting to send the zoom link. Beth.
4 00:01:38.430 ⇒ 00:01:50.229 Payas Parab: Google actually did such a smart thing by like, when you use Google Calendar and you use G meets it like you get so used to it automatically adding it that you forget what it’s like to like. Have to add it back and like that alone. I’m like.
5 00:01:50.690 ⇒ 00:01:58.359 Payas Parab: honestly makes it hard to like implement zoom, because you’re like, I’m just so used to every calendar invite I send from Google Calendar is like defaulted
6 00:01:58.930 ⇒ 00:01:59.330 Payas Parab: hopefully.
7 00:01:59.330 ⇒ 00:02:00.329 Nicolas Sucari: Yeah, I know.
8 00:02:00.530 ⇒ 00:02:01.570 Payas Parab: Email,
9 00:02:03.820 ⇒ 00:02:05.739 Nicolas Sucari: You know, I’m not using Google Cloud.
10 00:02:06.420 ⇒ 00:02:06.880 Payas Parab: I’m using.
11 00:02:08.919 ⇒ 00:02:09.969 Payas Parab: Oh, nice. Okay.
12 00:02:10.910 ⇒ 00:02:21.920 Nicolas Sucari: So as I’m using vim call like it lets me like, compare different time zones. And that stuff, and I add, like I had connected with zoom, too. So it automatically adds, my zoom link there.
13 00:02:22.840 ⇒ 00:02:24.840 Nicolas Sucari: So yeah, it’s kind of the same like, it’s.
14 00:02:24.840 ⇒ 00:02:25.290 Payas Parab: I need.
15 00:02:25.290 ⇒ 00:02:25.670 Nicolas Sucari: I.
16 00:02:25.670 ⇒ 00:02:26.570 Payas Parab: No problem.
17 00:02:27.480 ⇒ 00:02:32.110 Nicolas Sucari: If I needed to add it manually. Yeah, I’ll forgot. But it’s easier like this way.
18 00:02:32.610 ⇒ 00:02:40.039 Payas Parab: Smart. I gotta switch to vim, Cal. Actually, Tom told me, that’s the right. Now I use reclaim. So it like syncs my calendar from work. Other work.
19 00:02:40.430 ⇒ 00:02:44.619 Payas Parab: Yeah. So yeah, what’s up? Beau.
20 00:02:45.060 ⇒ 00:02:47.180 Bo Yoon: Hey? How’s it going? Guys?
21 00:02:47.680 ⇒ 00:02:48.740 Payas Parab: Pretty good.
22 00:02:50.520 ⇒ 00:02:51.760 Nicolas Sucari: Yep, all good.
23 00:02:52.240 ⇒ 00:03:10.529 Payas Parab: Awesome. Well, thank you guys for pushing a little bit. Set some logistics issues this morning. Wanted to, just like, basically like we want to just prepare these like 2 to 3 sentence elevator pitches. So we like, I feel good about where we are at the skew stuff right? Like we’ve hashed that out. We’ve like gone as far as we can go without Dan. The next thing is sort of like figuring out how to tee up
24 00:03:10.650 ⇒ 00:03:15.431 Payas Parab: additional work, right? So we can get more hours for everyone. So
25 00:03:16.120 ⇒ 00:03:43.649 Payas Parab: I, I did some work around the discounts and warranties which I can share out basically the data and kind of share, like what I’m finding and like, do we think this is a meaning? Basically, the conclusion of like this group of 3 is like the 2 areas we explored a little bit like, is this a meaningful thing worth bringing up to Dan? Right? That’s the the conclusion we want out of here, and it’s like 2 to 3 sentence elevator pitch of. Here’s what we see. Here’s what we could do. And here’s what.
26 00:03:43.900 ⇒ 00:04:07.949 Payas Parab: Why it matters right like, why, it matters. For like an executive like that’s basically that’s the goal of this call. Essentially so I’m happy to go 1st kind of share my data or both. You want to go first.st I’m I’m I’m open to either. We can keep it to like just 5, 10 min, just going over what you research what you found in the data like what you think might be an opportunity. And then we can just together draft up those like those elevator pitch.
27 00:04:08.360 ⇒ 00:04:11.060 Payas Parab: Thanks. Yeah, you can. You can go first.st
28 00:04:11.460 ⇒ 00:04:14.980 Payas Parab: Sure. Awesome. So the area I explored was,
29 00:04:15.760 ⇒ 00:04:20.479 Payas Parab: around. Let me share my Google Doc here screen 2.
30 00:04:21.899 ⇒ 00:04:40.942 Payas Parab: So I again, in an effort to like, not do too much work until we know this is something else I just like have, like my snowflake queries, and just like my observations here. And Bo, I hope that was clear to you as well. It’s just like meant to be an exploration. It wasn’t meant to be like we need a client ready, deliverable. Maybe I didn’t make that clear. But
31 00:04:41.930 ⇒ 00:04:43.949 Payas Parab: did you? Did you think that.
32 00:04:44.280 ⇒ 00:04:49.755 Bo Yoon: Oh, no, no, I was didn’t even write it down. It’s it’s everything is on my head.
33 00:04:50.060 ⇒ 00:05:05.550 Payas Parab: Oh, that’s fine! That’s fine, too, that that works like that that good! This is meant to be a thought exercise more than it’s meant to be like a yeah. So the core problem here, right is like, so warranties, what are warranties right to give some background warranties? And I was jotting all this down chicken scratch. But like warranties.
34 00:05:05.550 ⇒ 00:05:23.080 Payas Parab: they’re basically like, Hey, we guarantee the product will work well. If it doesn’t right, then we can send it back right? And I believe what happens is that pool parts basically loses that revenue right? They send it back. We have to confirm that they, in fact, lose it? Or do they send it back to the manufacturer? That I’m not sure
35 00:05:23.090 ⇒ 00:05:26.639 Payas Parab: my guess part of the my guess is, they do, because.
36 00:05:26.720 ⇒ 00:05:47.610 Payas Parab: like sending it back. Seems like it would be hard based on what we knew about how hard it is to get these skews here, I imagine, like sending it back. And, like, you know, processing like a restocking, it’s like not really worth their time. So there’s 2 elements that we have to think about. One is like they actually sell warranties right? It’s like a revenue stream for them and based on
37 00:05:48.574 ⇒ 00:05:51.000 Payas Parab: yeah, apparently I can’t zoom in
38 00:05:53.980 ⇒ 00:05:56.560 Payas Parab: based on what I saw like, it looks like.
39 00:05:57.680 ⇒ 00:06:01.879 Payas Parab: you know, I’m just gonna I’m just gonna share my screen to snowflake. It’ll be easier to look at data that way.
40 00:06:02.350 ⇒ 00:06:04.729 Payas Parab: I have these key ones up.
41 00:06:06.980 ⇒ 00:06:11.200 Nicolas Sucari: So selling a warranty is like like Applecare, right? Like something like that.
42 00:06:11.200 ⇒ 00:06:14.590 Payas Parab: Yeah, you can basically pay for like a 2 year.
43 00:06:14.800 ⇒ 00:06:21.349 Nicolas Sucari: 2 year warranty. If something breaks they will replace it at no cost. That’s why insurance policy name something.
44 00:06:21.350 ⇒ 00:06:29.809 Payas Parab: Exactly. Yeah, and they’ll refund you. I think they’ll like refund you, or they’ll like, Give it like, or they’ll give you a new one. But either way, it’s like.
45 00:06:30.160 ⇒ 00:07:00.129 Payas Parab: assume that you eat the cost right like you. They have to eat the cost. So here’s what I noticed is like. So they’re 1st of all like in their orders themselves. They’re like selling warranties, right? So they sell them, and if you look at them it’ll be like for Black and Decker whatever. 2 year warranty, 4 year warranty. It looks like it’s like somewhat of an income stream. They sell about 20 to 40 of them a month, and they’re like making 5 to 10 K. On it. I imagine that doesn’t like they don’t really think of it as a revenue stream right? Like it’s like.
46 00:07:00.360 ⇒ 00:07:04.630 Payas Parab: it’s just like a nice like additional money they make. Now, the question is, is like.
47 00:07:04.930 ⇒ 00:07:13.090 Payas Parab: are they getting cooked on these right? If if they’re like giving, if they’re sending a lot like if they’re getting a lot of warranty claims.
48 00:07:13.220 ⇒ 00:07:23.670 Payas Parab: and then they’re like refunding the money, then they fully lose the money, and I believe they have to eat the cost of the product right? They already ordered it. They already shipped it here from Asia. They’re like.
49 00:07:23.670 ⇒ 00:07:26.709 Nicolas Sucari: All of those costs. Maybe they are down for them.
50 00:07:27.090 ⇒ 00:07:48.819 Payas Parab: They? They have to eat it. So what I notice is like they they start, they they sold less in 2024 in the second half, and later or earlier this year, they’ve sold less warranties right? But they’re selling 2 to 4 year warranties like, which means that they could be on the hook for like a while. You know what I mean. It’s like basically opening them up to like a giant risk of like, well, what if
51 00:07:48.900 ⇒ 00:08:14.050 Payas Parab: everyone suddenly puts in the warranty right? I tried to figure out like, Okay, warranty claims like how many we got like, there’s these like, there’s this table that was made. I think someone’s like done sort of an analysis like this, because there was a dimension table done already. That’s like, Hey, like by warranty date, we can figure out, like how many claims people made for warranty. So this is like, you know, we have 50 to 200. If the average good size is like
52 00:08:14.290 ⇒ 00:08:25.620 Payas Parab: couple of 100 bucks like, we’re getting like like this, can this can add up really quickly, right? It can be like a 10,000 $20,000 cost really, really quickly. And in order to figure that.
53 00:08:25.620 ⇒ 00:08:29.150 Payas Parab: yeah, sorry is that the dollar amount in the total, hey?
54 00:08:29.150 ⇒ 00:08:30.270 Bo Yoon: The number of.
55 00:08:30.270 ⇒ 00:08:32.449 Payas Parab: Number of claims. Yeah, number of claims.
56 00:08:32.450 ⇒ 00:08:36.649 Bo Yoon: Oh, then, isn’t that like way too much compared to.
57 00:08:37.690 ⇒ 00:08:39.760 Nicolas Sucari: How many warranties are selling per month?
58 00:08:39.760 ⇒ 00:08:40.850 Nicolas Sucari: Yeah, that’s 2.
59 00:08:42.470 ⇒ 00:08:45.500 Payas Parab: I see what you’re saying. Yeah. So that’s actually a good. That’s a good call out. This is why
60 00:08:45.950 ⇒ 00:08:48.650 Payas Parab: they’re selling like 20 to 50.
61 00:08:48.650 ⇒ 00:08:51.590 Nicolas Sucari: And they are receiving a hundred claims. That’s crazy.
62 00:08:51.860 ⇒ 00:08:52.390 Payas Parab: Yeah.
63 00:08:53.920 ⇒ 00:09:07.459 Payas Parab: that that’s a good point. So right there, that’s like, okay, I gotta go look at that. Yeah. So they’re like, you’re you’re right there. They shouldn’t be. So I think, okay. So there’s 1 thing to note here is, I think they sell warranties. And then certain products also just have warranties.
64 00:09:07.600 ⇒ 00:09:08.170 Bo Yoon: Terrific.
65 00:09:08.170 ⇒ 00:09:19.059 Payas Parab: Applied to the manufacturer. So black. And Decker, I know, is a big one, and that’s actually most of the warranty claims here. If you look at the order level, they’re black and Decker, right? So I assume that’s like a manufacturer level guarantee.
66 00:09:19.430 ⇒ 00:09:20.110 Bo Yoon: Oh, yeah.
67 00:09:20.379 ⇒ 00:09:28.459 Payas Parab: And and that’s what we may have to ask. Right is like, well, what happens when the black and Decker, like a claim, is made against Black and Decker warranty
68 00:09:28.580 ⇒ 00:09:55.050 Payas Parab: like, what do you do next? Right like, are you guys shipping that shit back to Black and Decker? Are they sending you a new one. How much that really cost your business? Because, you know, there’s hundreds it the nice thing is it like fell off in the last second half, QH. 2, 2024, early 2025. It kind of fell off, but in that, like mid of 2024 it was like a fuck ton of them. And I tried to figure out like how much it impacted right? And it’s like.
69 00:09:55.140 ⇒ 00:09:59.459 Payas Parab: it’s like, not. It’s a non trivial amount. So it’s like, if you looked at like
70 00:10:00.193 ⇒ 00:10:18.669 Payas Parab: the impact of warranty discounts the way I did this right? So their data, as we know, is a mess right. And one of the things that you’ll see is like they’re sort of tracking their claims separately in that data table. But then, one way that they track warranties, which is like not great at all. But like is what they do
71 00:10:18.820 ⇒ 00:10:22.749 Payas Parab: is they discount the original order a hundred percent.
72 00:10:23.060 ⇒ 00:10:32.220 Payas Parab: and then just write the word like warranty or write like a issue, a discount code that basically says, like, put in this discount code, and you’ll get the new one for free type deal.
73 00:10:32.370 ⇒ 00:10:32.900 Bo Yoon: It’s like.
74 00:10:32.900 ⇒ 00:10:48.860 Nicolas Sucari: There is a new order, but it’s with without a price, because they have like a big discount. So the total discount amount is going up the total sold amount or or the revenue amount is going down because they are selling something on a 0 price because of that discount right?
75 00:10:49.150 ⇒ 00:10:49.700 Payas Parab: Yeah.
76 00:10:49.700 ⇒ 00:10:52.979 Bo Yoon: Yeah, it’s lowering the aob, I guess. Then
77 00:10:54.190 ⇒ 00:10:58.899 Payas Parab: Yeah, they, they basically like they go in and like refund the like. They they basically put like.
78 00:10:59.610 ⇒ 00:11:15.860 Payas Parab: I, I can show you like a row. But it’s like they basically have the price of the item. And then they like just discount it a hundred percent. So that’s like the free warranty one, and then they like have a discount code, or like in the memo for the order, right like the word warranty. Somewhere in there. It’ll be like warranty with some weird.
79 00:11:16.220 ⇒ 00:11:18.800 Nicolas Sucari: And everything goes in a new order, right?
80 00:11:19.340 ⇒ 00:11:27.080 Payas Parab: Yes, I should confirm that that I’m not I. So the new order is the new order, do you think is reflecting? And we can look at like.
81 00:11:28.570 ⇒ 00:11:43.290 Nicolas Sucari: What I’m trying. Yeah, what I’m trying to figure out is like like the full process of I’m buying like a pump. Right? I buy the pump a black decor pump, and it’s it comes with the warranty, let’s say right and
82 00:11:43.450 ⇒ 00:12:00.430 Nicolas Sucari: in a year something breaks, and I like want to use that warranty I like, I bring up the claim to both parts, and they say, Okay, they give me a discount code or something, so that I can place a new order like that’s the that’s what’s happening.
83 00:12:02.550 ⇒ 00:12:05.720 Nicolas Sucari: So we have, like 2 orders with a new.
84 00:12:05.720 ⇒ 00:12:12.880 Payas Parab: Should be 2 orders right? We have to figure out. That’s that’s a good point. I think. I think I didn’t. I kind of like overshot that. I didn’t really think about.
85 00:12:13.620 ⇒ 00:12:20.199 Payas Parab: Okay, there’s there should be 2 aligned orders, the original order and the order that
86 00:12:22.260 ⇒ 00:12:25.110 Nicolas Sucari: That was used to just replace that broke.
87 00:12:25.110 ⇒ 00:12:28.799 Payas Parab: Replace that one. And then the replacement order. The discount is a hundred percent.
88 00:12:29.660 ⇒ 00:12:37.060 Payas Parab: And they or they put the word like warranty or something like it’s like one or the other kind of thing is like what I’ve noticed. So sometimes it’s like
89 00:12:37.390 ⇒ 00:12:49.550 Payas Parab: they don’t discount it all the way. I think they charge them like a like. My guess is like a restocking or like. This thing isn’t always 100 is what I noticed is like. Sometimes it’s like 95 90% discount
90 00:12:50.080 ⇒ 00:12:54.710 Payas Parab: But this, this basically tells us, like, okay, like, what are the
91 00:12:55.000 ⇒ 00:13:02.850 Payas Parab: yeah. But but what I like, basically like, in order to build the elevator pitch is like, the question is, is like, is this actually matter to them? Right like is this matter? And like.
92 00:13:02.850 ⇒ 00:13:03.210 Nicolas Sucari: Yeah.
93 00:13:03.210 ⇒ 00:13:08.590 Payas Parab: My working theory is like, yes, but I like would need some like in.
94 00:13:08.590 ⇒ 00:13:22.749 Nicolas Sucari: If we if we check, like the the amount, like the percentage amount that the warranties or total claims is from the total amount that they are getting from revenue. How much is that.
95 00:13:23.512 ⇒ 00:13:25.200 Payas Parab: It’s right here. It’s
96 00:13:25.859 ⇒ 00:13:29.900 Payas Parab: this is by quarter. So I just did it by quarter to keep it simple.
97 00:13:30.170 ⇒ 00:13:32.820 Nicolas Sucari: It’s like 10% or or less.
98 00:13:33.070 ⇒ 00:13:35.129 Payas Parab: It’s it’s less. It’s like 5 to 10%.
99 00:13:35.380 ⇒ 00:13:42.129 Nicolas Sucari: And how much like in total, in like in numbers? How how much dollars is that like 20.
100 00:13:43.240 ⇒ 00:13:47.360 Payas Parab: That would be. It’d be about 350,000. 300,000.
101 00:13:47.360 ⇒ 00:13:51.390 Nicolas Sucari: So maybe it’s worth to take a look at that. I don’t know. We should.
102 00:13:51.390 ⇒ 00:13:56.469 Payas Parab: Yeah, this, I mean, this kind of adds up right if you like. If we add up these core. This is like
103 00:13:57.080 ⇒ 00:14:03.359 Payas Parab: a quarter of a million per fiscal quarter. If we look at it annual, it’s like about a million dollars a year.
104 00:14:03.360 ⇒ 00:14:03.860 Nicolas Sucari: Yeah.
105 00:14:04.106 ⇒ 00:14:11.239 Payas Parab: The question is, is like, do they actually have any power to change that right like? Do they have any power to change that. So that’s why it’s like.
106 00:14:11.560 ⇒ 00:14:39.589 Payas Parab: I think part of it is like the black and Decker guarantees right and like, I think, part of their contract with them, and I. I don’t know where I saw this or where we talked about it, but it’s like there’s some reason they have to offer a warranty right. And they sort of like as the distributor have to like. Take up the risk. That’s the sense I get. But like what I’m seeing is like, it is a very nontrivial impact to their overall bottom line, right? Cause they assuming they have to fully eat the cost right. If Black and Decker eats the cost, and we just have to figure out like.
107 00:14:39.590 ⇒ 00:14:40.220 Nicolas Sucari: Yeah, yeah.
108 00:14:40.220 ⇒ 00:14:44.909 Payas Parab: Which skews are black and decker, and not black and decker. Then it doesn’t matter right? So it’s like.
109 00:14:45.260 ⇒ 00:14:50.820 Payas Parab: but like overall like discounts where we did this warranty discount thing
110 00:14:51.180 ⇒ 00:14:54.080 Payas Parab: as total potential revenue is like
111 00:14:55.500 ⇒ 00:15:05.060 Payas Parab: like, like last quarter, it was 4%, which is not bad, right? But, like the previous quarters. It was 9% 13 and 9%.
112 00:15:06.240 ⇒ 00:15:12.560 Payas Parab: And if they keep selling them, then they’re like, and they’re like 2 to 4 years, they could be on the hook for a bunch of them right like, if like
113 00:15:13.050 ⇒ 00:15:28.500 Payas Parab: at Year 3, these guys decide like, Hey, this thing isn’t working like I thought, and then then pull parts has to eat the cost. They’re kind of like hoping so like, my question is like, I think we like. I think there is some some impact here, right? And it’s like the discounts
114 00:15:29.000 ⇒ 00:15:39.559 Payas Parab: I think discounts. In general. I re like this thing was like discounts and warranties, right because they got mixed up. But, like true discounts, I don’t think really matter as much right like we’re looking
115 00:15:40.030 ⇒ 00:15:45.199 Payas Parab: like 20,000, and like as a percentage of revenue. It’s like 5 to 10%. So that’s just like them
116 00:15:45.420 ⇒ 00:16:00.529 Payas Parab: selling right? Like, that’s okay, like, that’s them selling, giving a discount. I don’t think there’s that much juice in terms of like discounts. I will say these are, by the way, these are true discounts, right. These are things that weren’t warranty, or weren’t like erasing the entire price.
117 00:16:01.147 ⇒ 00:16:06.059 Nicolas Sucari: You are looking for like the percentage when it is not 100% or.
118 00:16:06.060 ⇒ 00:16:06.829 Payas Parab: Yes, exactly.
119 00:16:06.830 ⇒ 00:16:09.090 Nicolas Sucari: A warranty in the in the demo right.
120 00:16:09.090 ⇒ 00:16:17.171 Payas Parab: Exactly. It is rising. It did rise a little bit in the like beginning of last year. So it’s like that is a direct revenue impact. Right?
121 00:16:17.830 ⇒ 00:16:20.040 Payas Parab: 5, 10% like
122 00:16:20.180 ⇒ 00:16:30.369 Payas Parab: I did like this, I think it is rising. But if I just look at the magnitude right like, if I look at the magnitude. It’s like costing them 30, $30,000, even at the Peak. It’s like
123 00:16:31.570 ⇒ 00:16:36.810 Payas Parab: whatever at the Peak. It’s like 40,000 20,000. I don’t think that’s as significant as like.
124 00:16:37.120 ⇒ 00:16:43.400 Payas Parab: okay, like, we’re these warranties are costing a couple of 100 K right? Of potential revenue. So
125 00:16:44.991 ⇒ 00:16:49.299 Payas Parab: I think the warranty is a more important problem. I think.
126 00:16:49.300 ⇒ 00:16:55.710 Nicolas Sucari: Is there a way to to check? How many active warranties we have like? How many.
127 00:16:56.180 ⇒ 00:16:56.650 Payas Parab: Right.
128 00:16:56.650 ⇒ 00:16:58.069 Nicolas Sucari: Okay, we don’t know that.
129 00:16:58.070 ⇒ 00:16:59.719 Nicolas Sucari: Yeah, like the orders.
130 00:16:59.940 ⇒ 00:17:06.640 Payas Parab: Yup, and that’s where, like I think I could show you real quick is if I pull.
131 00:17:07.260 ⇒ 00:17:08.890 Nicolas Sucari: Yeah. So that’s what you’re saying, Laura.
132 00:17:08.890 ⇒ 00:17:13.780 Payas Parab: That’s part of what I’m saying, right is like, you’re opening up to a gigantic financial risk, and it’s like
133 00:17:13.900 ⇒ 00:17:22.739 Payas Parab: some of your customers have the opportunity to play a game right? Like, some guys play games with this where they’re like, okay, 3 years. Let’s just like, submit the warranty, whatever. Right?
134 00:17:23.690 ⇒ 00:17:24.260 Payas Parab: Yeah.
135 00:17:25.240 ⇒ 00:17:36.670 Payas Parab: So if you if we look here, right? So this is a scenario where? Yeah, these are warranties. And you basically erase out the entire cost of the good right? So you have. Items were costing 869,
136 00:17:36.820 ⇒ 00:17:39.836 Payas Parab: they cancel it out, and then
137 00:17:40.630 ⇒ 00:17:45.769 Payas Parab: sorry. This is the wrong, not where discount code. It’s where product, what is it called product class
138 00:17:46.820 ⇒ 00:17:48.620 Payas Parab: confirm orders
139 00:17:49.170 ⇒ 00:17:55.959 Payas Parab: or order items or product class equals. And I think so. This is like the warranty. They sell right
140 00:17:57.330 ⇒ 00:18:04.841 Payas Parab: and the warranty they sell and also we’ll hop off this because I wanna make sure we have enough time to review both kind of ideas as well.
141 00:18:05.100 ⇒ 00:18:07.069 Bo Yoon: Oh, no, my mine is really short, so it’s.
142 00:18:07.070 ⇒ 00:18:15.149 Payas Parab: Sweet. Okay, so like, if we look at this one like you’ll see here. Right? So on
143 00:18:15.370 ⇒ 00:18:42.820 Payas Parab: 2020, they sold this 2 year warranty. This one’s now expired right? But they don’t really have a sense of like like. What we would have to do is like we’d have to go in. I think we just run it through like a chat, gpt to just like row clean. And it’s like it does. It? Does this thing spit out what year it like in almost all of them, it says, like 2 year warranty, 4 year warranty, 3 year warranty. But you see, like this, 3 year warranty was issued in 2023.
144 00:18:43.470 ⇒ 00:18:51.879 Payas Parab: They’re on the hook for the next couple of years, right? That, like one thing I did notice that’s like worth noting is like in the warranty claims.
145 00:18:52.220 ⇒ 00:19:13.840 Payas Parab: It seems that people claim the warranty pretty quickly, like it’s not like a like. I wait 2, 3 years, and then I’m like claiming the warranty. It’s kind of like fuck. This thing doesn’t work like I wanted it to. Oh, wait! This thing has a warranty. Let me go push that right like. That’s sort of how the behavior is. That’s my get, like I can confirm that. But that’s based on just a few pieces of data I was looking at.
146 00:19:14.050 ⇒ 00:19:29.819 Payas Parab: It seems that people sort of do it quickly. That’s another thing. We can check right? So I think, like 2 follow ups to check is like when an order is canceled out. Is it like? Is there 2 orders? Right? One for the warranty, one for the other one, and we have to make sure we’re not double counting the revenue impact. The second would be
147 00:19:30.540 ⇒ 00:19:31.700 Payas Parab: okay.
148 00:19:32.050 ⇒ 00:19:34.529 Payas Parab: We have these warranties that are sold.
149 00:19:35.210 ⇒ 00:19:37.089 Payas Parab: We need to figure out like
150 00:19:37.300 ⇒ 00:19:42.660 Payas Parab: we have the data in here to determine how long they’re supposed to be right.
151 00:19:43.240 ⇒ 00:19:48.089 Payas Parab: and we should get a sense somewhat of like how long it takes for people to claim them.
152 00:19:48.850 ⇒ 00:19:50.270 Payas Parab: claim the warranty.
153 00:19:52.310 ⇒ 00:19:52.960 Bo Yoon: Hmm.
154 00:19:53.330 ⇒ 00:19:59.100 Payas Parab: So then that’s that’ll tell you. Like, okay, all these warranties you issued in the last couple of years like.
155 00:19:59.960 ⇒ 00:20:07.899 Payas Parab: you know, what? What? What are you potentially on the hook for, or like? How many could you get hit with? Right? We could figure that out.
156 00:20:12.800 ⇒ 00:20:17.599 Bo Yoon: Would that also be like a prediction of how many claims we’ll be getting in the future?
157 00:20:18.020 ⇒ 00:20:19.470 Payas Parab: Yeah, yeah, exactly.
158 00:20:19.470 ⇒ 00:20:36.729 Nicolas Sucari: Trying, like with the behavior that we can see from the past warranties or the past claims on the amount of claims, on the amount of warranties that are issued and are like, kind of active. How? What we can predict that will happen in the following years. Right? That’s what you’re you’re saying.
159 00:20:36.730 ⇒ 00:20:37.480 Payas Parab: Yeah.
160 00:20:37.940 ⇒ 00:20:50.336 Payas Parab: yeah. So like, what I, what I want to know, right is like in the like, what we notice is like, Hey, it’s actually having a pretty big impact like it’s a sizable portion of revenue is just like some of these warranty refunds where you’re giving these products for free.
161 00:20:50.920 ⇒ 00:21:03.499 Payas Parab: You know, we’re looking at like a magnitude of like 5% of revenue impacted. So like the question, like, the the actual thing is like, is it time to reevaluate the warranty? Do we stop selling them right to like
162 00:21:03.720 ⇒ 00:21:06.869 Payas Parab: reduce the amount of risk we’re exposing ourselves to?
163 00:21:07.010 ⇒ 00:21:14.530 Payas Parab: That’s kind of what I’m thinking is like the pitch, right is like, we’re seeing a bunch of them. We’re seeing a lot of impact. It’s the primary
164 00:21:14.860 ⇒ 00:21:19.619 Payas Parab: thing in your discounts, right? It’s like just sending people the warranty replacement.
165 00:21:22.150 ⇒ 00:21:27.120 Payas Parab: and there’s quite a few you’ve sold in the last 2 to 3 years that like are still active.
166 00:21:27.270 ⇒ 00:21:30.410 Payas Parab: even though people only claim them, and, like quickly
167 00:21:30.520 ⇒ 00:21:35.819 Payas Parab: generally, we still think it like opens you up to a bunch. So the goal is like, can we figure out
168 00:21:35.940 ⇒ 00:21:38.500 Payas Parab: how many outstanding warranties there are.
169 00:21:38.680 ⇒ 00:21:46.480 Payas Parab: how much it could cost you, and then give you a recommendation on like, is it worth reconsidering? Some of this warranty stuff?
170 00:21:47.400 ⇒ 00:21:51.600 Nicolas Sucari: Yeah, and also trying to ask about the black and decker ones and try to see exactly.
171 00:21:51.840 ⇒ 00:21:54.859 Nicolas Sucari: You need to clean the data from those ones, too. Yeah.
172 00:21:55.160 ⇒ 00:22:04.139 Payas Parab: Exactly so that that to me is like this does feel like a viable project. It’s something that I what I imagine is like it. Also, like they they are used to like
173 00:22:04.520 ⇒ 00:22:16.070 Payas Parab: like they’re getting. My guess is like they’re getting feedback from customers right where they’re like fuck like customer service is costing us so much. And why? And it’s like, well, we just have to like eat the entire cost of this entire product
174 00:22:16.400 ⇒ 00:22:18.929 Payas Parab: because someone made a warranty claim.
175 00:22:19.780 ⇒ 00:22:29.299 Payas Parab: And maybe it’s like we have to be harsher with warranty claims. Maybe we have to like stop issuing warranties right? That’s sort of like the decision we’d want to help them make.
176 00:22:30.270 ⇒ 00:22:32.430 Nicolas Sucari: Yeah, or charge more for the warranties.
177 00:22:32.620 ⇒ 00:22:33.850 Payas Parab: Exactly. Yeah.
178 00:22:34.150 ⇒ 00:22:46.330 Payas Parab: So and I think it’s even like it’s worth just like the way we could even bring it up is just like, hey, we wanna like talk to you. But like again that the 3 sentences like, Hey, we’re observing this. It’s pretty large, as like a percentage of all your discounts
179 00:22:46.460 ⇒ 00:22:51.960 Payas Parab: like, and you have a lot that you’ve sold in the last 2 to 4 years that like might still be active.
180 00:22:52.820 ⇒ 00:22:58.739 Payas Parab: We wanted to just check is like, is this something you want to look into right like? Is this like, do we want to like evaluate?
181 00:22:59.050 ⇒ 00:23:03.360 Payas Parab: Can we? Can we adjust warranty policy. Right? You know what I mean like, can we.
182 00:23:03.360 ⇒ 00:23:03.850 Nicolas Sucari: Yeah.
183 00:23:03.850 ⇒ 00:23:11.419 Payas Parab: Just whether we sell warranties and things like that. Can we be harder with the warranties? Those are things that like you could do.
184 00:23:12.810 ⇒ 00:23:14.650 Nicolas Sucari: Cool, like, it. Yeah.
185 00:23:14.650 ⇒ 00:23:24.079 Payas Parab: Awesome. Alright. Well, I think I’ll put together the 2 to 3 sentence, like, kind of like what I just described here. And then, yeah, I think I think this is. There’s something here. Worth exploring.
186 00:23:24.680 ⇒ 00:23:30.329 Nicolas Sucari: Yeah. And when we were when we were talking about exploring, we were talking about like going through that 2 week process right.
187 00:23:30.330 ⇒ 00:23:31.739 Payas Parab: Yeah, exactly. It’ll be like.
188 00:23:31.740 ⇒ 00:23:34.920 Nicolas Sucari: It’s not. It’s not gonna be like a long project that we’re.
189 00:23:34.920 ⇒ 00:24:01.949 Payas Parab: That’s what I exactly. It’s it’s like an analysis sprint, not like some data that we’re doing. It’s like, Hey, like I, I will spend 10 h in the next 2 weeks on this, just to like, see if this is something, and like we’ll do it in a 2 week sprint to make sure it. Like, basically, we put pressure on ourselves to get something out. There’s like, there’s like, you can always go down any fucking rabbit hole with all of this right like I can be like, oh, wait! We should find the manufacturer for each of these warranties, and break them down, and, like.
190 00:24:02.310 ⇒ 00:24:02.910 Nicolas Sucari: No, that’s fine.
191 00:24:02.910 ⇒ 00:24:29.059 Payas Parab: Deadline. It’s kind of like. Well, let’s get something. Let’s like like, what do you find so far? And it gives us an excuse to meet with them every 2 weeks. So cool I am. I’m good on this one, so I think warranty. Let me draft that couple sentence thing. We’ll send that Tom be like we do think there’s something here worth evaluating. I looked in the slack. It seems that they did some warranty work back in q. 1 of 2024. So it might be like due time to refresh this.
192 00:24:29.470 ⇒ 00:24:35.639 Payas Parab: So I’ll bring that up, and then I’ll flag that to Tom, as I did see that they did some analysis like this before, so.
193 00:24:35.830 ⇒ 00:24:40.259 Nicolas Sucari: I shared. I don’t know if it was about warranties, but I shared one pdf, right.
194 00:24:40.260 ⇒ 00:24:41.730 Payas Parab: Other reports. Right.
195 00:24:42.070 ⇒ 00:24:42.780 Nicolas Sucari: Yeah.
196 00:24:43.410 ⇒ 00:25:03.429 Payas Parab: Great awesome. Well, that’s all I had. Bo, if you wanna just share kind of like your Zendesk exploration kind of like what you had. And like, I said, goal is just to like, let’s get. Let’s get 3 sentences right. Let’s determine. Is this something worthwhile, and that’s 1 and 2 is, let’s get the 3 sentence summary of what we might do.
197 00:25:04.090 ⇒ 00:25:08.510 Bo Yoon: Yeah, well, seems like the only valuable
198 00:25:08.820 ⇒ 00:25:12.660 Bo Yoon: column that we can use is going to be this one which.
199 00:25:13.816 ⇒ 00:25:14.973 Payas Parab: Really.
200 00:25:16.130 ⇒ 00:25:22.199 Bo Yoon: Yeah, which is, I guess it’s a email that that they’re getting, that the customer service is getting.
201 00:25:22.950 ⇒ 00:25:23.939 Payas Parab: Oh, God! Okay.
202 00:25:23.940 ⇒ 00:25:24.770 Bo Yoon: You’re
203 00:25:25.660 ⇒ 00:25:36.330 Bo Yoon: some of them are well well organized, but most of them are are not. But yeah, I mean, it’s impossible to go over each of the emails. So
204 00:25:36.690 ⇒ 00:25:37.560 Bo Yoon: well, we can.
205 00:25:37.560 ⇒ 00:25:39.290 Payas Parab: We got we got AI baby.
206 00:25:39.290 ⇒ 00:25:40.790 Bo Yoon: Yeah, and we got a, yeah.
207 00:25:40.790 ⇒ 00:25:45.650 Payas Parab: Is this? Is this actually the only useful thing like there’s nothing else I.
208 00:25:45.650 ⇒ 00:25:48.199 Bo Yoon: Find anything else that that was.
209 00:25:48.810 ⇒ 00:25:51.219 Payas Parab: I guess they have type, right? Incident.
210 00:25:51.700 ⇒ 00:26:01.410 Bo Yoon: Type was I mean, there was also type, but most of them were empty, so we will probably have to redo it ourselves.
211 00:26:01.800 ⇒ 00:26:05.670 Bo Yoon: like, I, I think that’s like just summarizing or
212 00:26:06.950 ⇒ 00:26:12.310 Bo Yoon: or labeling the the emails, whether it’s a question, a problem.
213 00:26:12.533 ⇒ 00:26:17.009 Payas Parab: But I think that’s like a great like thing for us to do right then, like we can pitch it.
214 00:26:17.010 ⇒ 00:26:17.879 Bo Yoon: Yeah, I mean.
215 00:26:18.170 ⇒ 00:26:26.619 Payas Parab: Like we actually have to process all of the info. And like the way I would do this right is like, I’d be thinking like
216 00:26:26.890 ⇒ 00:26:41.879 Payas Parab: like, can we create categories? Right? And if we create categories like, what I’ve done with something like this is, I will like, have I use this thing called numerous AI, and it’s basically like a plugin that’s like a chat Gpt Plugin into excel.
217 00:26:42.030 ⇒ 00:26:42.560 Payas Parab: And you
218 00:26:42.970 ⇒ 00:26:48.349 Payas Parab: like, put in a formula, it’s like equals AI, and it’s like you can just link it to a cell and be like in this cell
219 00:26:48.570 ⇒ 00:27:00.510 Payas Parab: is like, what do you think this is like? What product category. Do you think this is like we’ve I’ve done that before where it’s like, or we like make it a Csv, and we write a script with a bunch of open AI calls just to like, tag the data.
220 00:27:01.210 ⇒ 00:27:02.000 Payas Parab: Yeah, that’s.
221 00:27:02.000 ⇒ 00:27:02.989 Bo Yoon: Doesn’t matter.
222 00:27:03.230 ⇒ 00:27:09.059 Bo Yoon: Yeah. I mean, I’m I’m not really sure how this type column is being made. Maybe
223 00:27:10.640 ⇒ 00:27:13.379 Bo Yoon: full time, or anyone else in the team.
224 00:27:13.380 ⇒ 00:27:20.270 Payas Parab: But do you know how much, how many nulls? So we have 200? It says there’s 208,000 in there. How many data points do we have as a whole.
225 00:27:21.190 ⇒ 00:27:22.940 Bo Yoon: How many data points we have.
226 00:27:22.940 ⇒ 00:27:25.260 Payas Parab: Yeah, like, how many rows is this entire sheet?
227 00:27:26.700 ⇒ 00:27:28.210 Bo Yoon: 74,000.
228 00:27:28.360 ⇒ 00:27:29.910 Payas Parab: No, no! Oh!
229 00:27:31.270 ⇒ 00:27:34.089 Payas Parab: Can you click the column with the the type.
230 00:27:34.730 ⇒ 00:27:35.849 Bo Yoon: The type. Yeah.
231 00:27:35.850 ⇒ 00:27:37.480 Payas Parab: Yeah. Like, V, column. V.
232 00:27:37.720 ⇒ 00:27:38.400 Bo Yoon: Oh, sir!
233 00:27:39.170 ⇒ 00:27:40.300 Bo Yoon: So that’s 20,000.
234 00:27:40.300 ⇒ 00:27:43.039 Payas Parab: 20,000. So the whole thing is around.
235 00:27:43.040 ⇒ 00:27:44.689 Bo Yoon: Around 24,000.
236 00:27:45.690 ⇒ 00:27:50.720 Payas Parab: Right, and there’s 74. So we have, like, 1 3rd of them, are categorized.
237 00:27:52.230 ⇒ 00:27:58.179 Payas Parab: Are they meaningful in any way like incident? I see incident, I see. Question.
238 00:27:58.640 ⇒ 00:28:03.310 Bo Yoon: Incident question ask. I I think there was also warranty
239 00:28:03.950 ⇒ 00:28:06.169 Bo Yoon: on there my incident problem question.
240 00:28:06.170 ⇒ 00:28:07.620 Payas Parab: Expression, task, okay.
241 00:28:09.630 ⇒ 00:28:23.510 Bo Yoon: So maybe we can, we can fill this up with another AI tool, one that you mentioned. And what what I was thinking was basically making 2 dashboards that we can show them.
242 00:28:23.680 ⇒ 00:28:24.790 Payas Parab: Yeah.
243 00:28:25.000 ⇒ 00:28:27.509 Bo Yoon: Post will be a time series. Dashboard.
244 00:28:27.920 ⇒ 00:28:28.830 Payas Parab: Okay.
245 00:28:28.830 ⇒ 00:28:37.730 Bo Yoon: One could be a sentiment kind of showing showing how customers are.
246 00:28:38.822 ⇒ 00:28:42.200 Bo Yoon: Reacting to the the customer service
247 00:28:44.920 ⇒ 00:28:47.740 Bo Yoon: it could. It could either be like good, good and bad.
248 00:28:47.900 ⇒ 00:28:52.820 Bo Yoon: like sentiment, analysis like or satisfied, not satisfied.
249 00:28:53.050 ⇒ 00:28:56.539 Bo Yoon: kind of side by side, chart over time.
250 00:28:56.670 ⇒ 00:28:59.850 Bo Yoon: See, like, for example, the the last month.
251 00:29:00.350 ⇒ 00:29:06.890 Bo Yoon: How many customers were satisfied, or how many customers were satisfied
252 00:29:07.420 ⇒ 00:29:16.510 Bo Yoon: that could be one of the dashboards, another one would be extracting the the product out of these emails.
253 00:29:18.200 ⇒ 00:29:25.010 Bo Yoon: If there’s a for the ones that have that have problems, so extract the the products that have
254 00:29:25.750 ⇒ 00:29:27.349 Bo Yoon: name out of the email
255 00:29:27.540 ⇒ 00:29:33.930 Bo Yoon: and show them like a graph, or like a bar graph or something, showing them how many.
256 00:29:34.350 ⇒ 00:29:40.729 Bo Yoon: how many of the of these products have had problems in the last month I mean over time.
257 00:29:41.810 ⇒ 00:29:42.270 Payas Parab: Got it.
258 00:29:43.770 ⇒ 00:29:47.050 Bo Yoon: Yeah, I was thinking of those 2 dashboards.
259 00:29:48.460 ⇒ 00:29:51.240 Bo Yoon: So I mean the the 1st step.
260 00:29:51.580 ⇒ 00:29:54.080 Bo Yoon: Obviously, it’s going to be extracting
261 00:29:54.210 ⇒ 00:29:56.509 Bo Yoon: the data out of this. This.
262 00:29:56.510 ⇒ 00:29:57.950 Payas Parab: Sure. Yeah, yeah.
263 00:29:58.670 ⇒ 00:29:59.400 Payas Parab: So yeah.
264 00:29:59.400 ⇒ 00:30:09.849 Payas Parab: if if you want to make like what we could do, like, what you just said is like, great, which is like, Hey, can we just make a list of like what data we want to extract? And we’re going to do that. And like, here’s what we’re gonna pull out right? The product.
265 00:30:10.090 ⇒ 00:30:19.869 Payas Parab: the incident, the like, everything. You just kind of said. Sorry I like like it was a bunch. So I I like, I think we like make like a hey like from all of these
266 00:30:20.220 ⇒ 00:30:28.730 Payas Parab: fields we can get what type of product it is, what type of thing is. Then we can set up by date. We can set up by product. We can set up by type.
267 00:30:29.866 ⇒ 00:30:31.529 Payas Parab: And then we can
268 00:30:32.470 ⇒ 00:30:40.690 Payas Parab: like give them like a dashboard. It’s like a view of my product by type of query, by yeah.
269 00:30:41.870 ⇒ 00:30:43.470 Payas Parab: mechanism. Even.
270 00:30:45.230 ⇒ 00:30:54.370 Nicolas Sucari: But when you’re talking about sentiment you are talking about like identifying the sentiment in in that description, or was or you have something else that you can use.
271 00:30:54.750 ⇒ 00:31:00.189 Bo Yoon: No. So so that will basically be just making another column and labeling the sentiment out.
272 00:31:00.190 ⇒ 00:31:04.159 Payas Parab: Could you like? How like are these like really detailed like, if I.
273 00:31:04.780 ⇒ 00:31:08.169 Bo Yoon: I mean some of them are like this one should. Oh.
274 00:31:09.350 ⇒ 00:31:10.499 Nicolas Sucari: Because I think it’s gonna be.
275 00:31:10.500 ⇒ 00:31:13.649 Bo Yoon: Some of them are really long. Some of them are like.
276 00:31:13.900 ⇒ 00:31:20.709 Payas Parab: Can you like? Can you make the the top part where you’re viewing that bigger? Yeah, like? Just so I could see what’s inside that cell when it’s like.
277 00:31:21.000 ⇒ 00:31:24.740 Payas Parab: So it doesn’t say what was the voicemail about it? Just it’s not like a.
278 00:31:25.310 ⇒ 00:31:25.890 Nicolas Sucari: No.
279 00:31:26.468 ⇒ 00:31:29.660 Bo Yoon: But some of them have, like a detailed.
280 00:31:30.860 ⇒ 00:31:31.830 Payas Parab: Hmm.
281 00:31:32.420 ⇒ 00:31:37.310 Bo Yoon: Oh, never mind, am I? Am I looking at the wrong? Oh, I’m I’m looking at the wrong one. Sorry.
282 00:31:37.310 ⇒ 00:31:39.370 Payas Parab: Oh, you’re looking at subject right? There’s.
283 00:31:39.950 ⇒ 00:31:42.070 Bo Yoon: Description versus description.
284 00:31:45.140 ⇒ 00:31:49.059 Payas Parab: Description has like a much more even including, okay, here’s the voicemail.
285 00:31:49.290 ⇒ 00:31:52.399 Payas Parab: Yeah. But then it’s like, if it’s like login to see the voicemail.
286 00:31:53.460 ⇒ 00:31:56.669 Bo Yoon: I mean here, there’s a body I went through.
287 00:31:56.670 ⇒ 00:32:00.079 Payas Parab: Yeah. And then there’s like, Okay, there’s like a product and stuff.
288 00:32:01.230 ⇒ 00:32:06.320 Bo Yoon: Yeah. So it’s gonna be mostly nlp, stuff. Nlp, analysis, that.
289 00:32:06.320 ⇒ 00:32:12.700 Payas Parab: Yeah, and that I’m like, like, I think we can build pretty like we can build that.
290 00:32:13.380 ⇒ 00:32:19.513 Payas Parab: my only concern is around the ones where it’s like the data isn’t actually there of like what it contains.
291 00:32:20.050 ⇒ 00:32:20.870 Nicolas Sucari: Yeah.
292 00:32:21.590 ⇒ 00:32:25.189 Bo Yoon: Yeah, I mean, this is some random stuff here.
293 00:32:26.840 ⇒ 00:32:27.590 Bo Yoon: Pretty?
294 00:32:29.460 ⇒ 00:32:33.980 Bo Yoon: Yeah. But you know, we got AI, yeah.
295 00:32:34.820 ⇒ 00:32:35.510 Payas Parab: But but I
296 00:32:35.510 ⇒ 00:32:44.020 Payas Parab: my, my big thing, is the ones that, like you’d have to click a link to go get. That’s the one I’m concerned about the ones there that I’m fine with, right? Like it’s the ones that are like.
297 00:32:46.530 ⇒ 00:32:47.290 Bo Yoon: Like this.
298 00:32:47.290 ⇒ 00:33:00.680 Nicolas Sucari: So the you you got these you got this information from Snowflake directly right from what we have there, because what we can do is like we can ask them to give us access to send this and see if we can get anything else from there.
299 00:33:01.080 ⇒ 00:33:05.190 Nicolas Sucari: I’m not sure if that’s gonna be possible. But that’s something that we can do
300 00:33:05.700 ⇒ 00:33:10.866 Nicolas Sucari: to try it and see if we can get like more details on that kind of tickets.
301 00:33:11.380 ⇒ 00:33:17.889 Nicolas Sucari: maybe we can get those links from there and directly transcribe them or do something.
302 00:33:18.770 ⇒ 00:33:19.330 Bo Yoon: Hmm.
303 00:33:19.330 ⇒ 00:33:20.950 Nicolas Sucari: I’m not sure I’m just.
304 00:33:21.253 ⇒ 00:33:44.890 Payas Parab: Think this is one where it’s like, it’s less like, Hey, here’s a problem that we know how to investigate. It’s like, Hey, like we found like this really cool data set. And we think it could give you into visibility into like, what products are going well like? Is the customer satisfied is like like, how are they sending in the concerns? Right? That type of thing like we think we can get that information for you
305 00:33:44.940 ⇒ 00:33:54.380 Payas Parab: from Zendesk. We have, like some of this data. But then we might need other data sources. But it’s like, if it’s not of interest, that we don’t want to unlock that right like like Nico. Like as like.
306 00:33:54.380 ⇒ 00:33:54.790 Nicolas Sucari: Yeah.
307 00:33:54.790 ⇒ 00:33:57.620 Payas Parab: You want these to be like my issue with Weeks? Right?
308 00:33:57.620 ⇒ 00:34:01.856 Nicolas Sucari: Yeah, exactly. Exactly. And my issue here is that we cannot like
309 00:34:02.490 ⇒ 00:34:26.400 Nicolas Sucari: like, quantify the opportunity that we have. So maybe selling these to them like I’m like in the pitch. It’s gonna be a little bit difficult. But if we can like have a clear a clear picture of what we’re trying to aim with this like looking a little bit more in the data. Maybe we can share that and see. Hey, we are trying to achieve these with these sets of data that we have here, and maybe that works with them.
310 00:34:26.750 ⇒ 00:34:30.790 Nicolas Sucari: we should we? We all need to be a little bit more specific. I think right?
311 00:34:31.719 ⇒ 00:34:33.049 Bo Yoon: Hmm, yeah.
312 00:34:33.050 ⇒ 00:34:51.790 Payas Parab: Yeah, my, my big thing for you, Bro is like, is there? Is there enough data that’s like, workable, right? Like that? That’s like that that would be like an important conclusion of this is like, okay, how much of it do we actually have things we can run an Lp on right. If it’s like login download, we can’t really do much with that right. But this thing we can right is like I received it.
313 00:34:52.838 ⇒ 00:34:55.351 Bo Yoon: Never got it, whatever it is. Right.
314 00:34:56.949 ⇒ 00:35:00.870 Bo Yoon: Hmm, yeah, maybe I should filter out.
315 00:35:00.870 ⇒ 00:35:07.089 Payas Parab: Could you do me a favor? Could you just try downloading from that? Send grid? I’m just curious. You see how how’s like download? And has that
316 00:35:07.570 ⇒ 00:35:09.430 Payas Parab: gigantic URL.
317 00:35:10.320 ⇒ 00:35:11.439 Bo Yoon: Yeah. Let me see.
318 00:35:11.440 ⇒ 00:35:15.020 Payas Parab: Go go to one where it says, Download not the URL.
319 00:35:15.280 ⇒ 00:35:16.340 Bo Yoon: Download. This is.
320 00:35:16.340 ⇒ 00:35:20.540 Payas Parab: Yeah, like one where not where it says login, one where it says like download.
321 00:35:21.240 ⇒ 00:35:25.739 Payas Parab: I’m wondering if it’s just like like, find a
322 00:35:27.370 ⇒ 00:35:31.969 Payas Parab: like, find one of the yeah one of the rows. It says the word login. But like, the second thing says, like, download.
323 00:35:31.970 ⇒ 00:35:34.129 Payas Parab: yeah, like there, yeah, right there
324 00:35:34.350 ⇒ 00:35:36.759 Payas Parab: like, I’m just wondering if, like, maybe they just.
325 00:35:38.680 ⇒ 00:35:41.750 Bo Yoon: Make that public because it’s like, okay, give me.
326 00:35:41.750 ⇒ 00:35:42.380 Payas Parab: I’m sure.
327 00:35:42.891 ⇒ 00:35:45.450 Bo Yoon: Audio file. Let me see.
328 00:35:46.080 ⇒ 00:35:49.989 Payas Parab: Were you able to get it without needing to log in.
329 00:35:50.550 ⇒ 00:35:52.430 Payas Parab: Also, we’re still seeing your excel.
330 00:35:52.430 ⇒ 00:35:55.350 Nicolas Sucari: That’s crazy. If we are able to get that, it’s crazy.
331 00:35:55.350 ⇒ 00:35:58.090 Payas Parab: I mean, sometimes they do it where it’s like they’re like.
332 00:35:58.280 ⇒ 00:36:04.149 Payas Parab: who on earth is gonna go on the Internet and find that URL in any meaningful way. So they’re like.
333 00:36:04.590 ⇒ 00:36:05.719 Payas Parab: is this.
334 00:36:05.720 ⇒ 00:36:07.079 Bo Yoon: The the link for that.
335 00:36:07.360 ⇒ 00:36:10.210 Bo Yoon: It’s just an audio file, but it’s not.
336 00:36:10.450 ⇒ 00:36:16.529 Bo Yoon: It’s like 2 seconds not showing anything. I think it’s just just a voicemail.
337 00:36:16.530 ⇒ 00:36:32.100 Payas Parab: Dial. Yeah, I mean, but like, see? Like, that’s like. But like, if we can access those right, that means that like again, as part of this script we can be like, let’s grab those. Let’s transcribe those, and let’s see what people are asking about right like to you like cause Dan doesn’t
338 00:36:33.230 ⇒ 00:36:39.249 Payas Parab: voice mails right can you send that one as well? I want to see if we can actually get one where someone actually left the voicemail.
339 00:36:39.550 ⇒ 00:36:40.700 Bo Yoon: This one.
340 00:36:41.070 ⇒ 00:36:45.739 Bo Yoon: This one is actually a 24 second voicemail.
341 00:36:46.240 ⇒ 00:37:08.333 Bo Yoon: Yeah. Heat pump from a pool recently. And you know, didn’t come with a check valve. So if you guys call me, give me a call back at (516) 848-3532. My name is Joe Omo, again, my phone number is (516) 848-3532. Let me know if you guys carry check valves. Okay for the heat pump for pool. Thank you.
342 00:37:08.620 ⇒ 00:37:20.020 Payas Parab: Yeah, that’s like this is. So we can actually analyze this right? Like, we can actually do something really clever here, right where it’s like. So here’s here’s what I say, like we, even though, are you like, do you? Do you? Do you code in python at all? Or.
343 00:37:20.200 ⇒ 00:37:22.220 Bo Yoon: Yeah, I, I, Tyson.
344 00:37:22.220 ⇒ 00:37:43.130 Payas Parab: Because there is. I’ve I’ve done something similar to this, where it’s like the trans I can. We can potentially like, do a quick demo right of like the tagging capabilities, and then show that to Dan, like, I think that could be really fun right? And it’s like, Hey, here’s a and it creates this opening for us to build for the AI front as well as the data front.
345 00:37:43.438 ⇒ 00:37:53.520 Payas Parab: What I what I think we still need before that. Like, I, I can help you make that demo. Right? Like, I actually did this for like, it’s basically, we’re helping podcasters transcribe their podcast
346 00:37:53.630 ⇒ 00:37:58.790 Payas Parab: so I have a script that like transcribes an audio file. And we can basically like.
347 00:37:59.150 ⇒ 00:38:28.619 Payas Parab: Call open AI to just like pump the message in and be like, can you tag it as one of these 5 categories? Right? If we meaningfully decide the 5 categories like, can you please tag it as like question, availability product, but product, not bug, I guess, like, you know what I mean, product default defect, right? Like, if we create a set of categories and we just have a script that like has a prompt that has that information. And it just pings openai a couple of times and gets us that like data.
348 00:38:28.980 ⇒ 00:38:31.150 Payas Parab: We could really, meaningfully, like
349 00:38:31.380 ⇒ 00:38:39.079 Payas Parab: like process, their entire Zendesk tickets to get like, what are the prominent issues? What are the problem products? That type of thing.
350 00:38:40.310 ⇒ 00:38:48.660 Nicolas Sucari: At least we can. Yeah, we’ll be able to share, like some insight on that sentiment or what we are getting for those. Yeah, that’s that’s something.
351 00:38:48.660 ⇒ 00:38:53.299 Bo Yoon: I mean, is there a way to download all these audio files.
352 00:38:54.690 ⇒ 00:38:55.080 Payas Parab: So what?
353 00:38:55.785 ⇒ 00:38:56.490 Payas Parab: Right?
354 00:38:56.490 ⇒ 00:39:10.049 Payas Parab: Right like a panda script right like it would just iterate through each row, and then it would look for, find a download link right? That could even be part of the prompt right? It’s like, if there’s a download link like, spit out the download link. And then you can actually just go ahead.
355 00:39:10.050 ⇒ 00:39:15.660 Bo Yoon: Save that in a in another column, and then use that column to how.
356 00:39:15.850 ⇒ 00:39:18.870 Payas Parab: You use the you use the request library in Python.
357 00:39:19.250 ⇒ 00:39:20.340 Bo Yoon: Oh, okay.
358 00:39:20.340 ⇒ 00:39:32.879 Payas Parab: Yeah, so you go to. You would do like request, Dot, get pump in that, URL. And then there’s like request, Dot, you would be able to like download that file there’s actually like in the if I’m sharing my screen here right like
359 00:39:33.762 ⇒ 00:39:36.803 Payas Parab: I can. If you want me to share real quick.
360 00:39:37.620 ⇒ 00:39:39.439 Payas Parab: so what we can do right is like
361 00:39:39.860 ⇒ 00:39:44.680 Payas Parab: step one extract, the URL, right step. One is extract. The URL
362 00:39:44.870 ⇒ 00:40:07.320 Payas Parab: second is like it will go to this screen right? So this URL, like the same way you pulled it out. We can just like have it go through each row and see if that download link exists. Also, like they’re using 2 main providers for voicemail. So it’s like there’s only 2 formats where this occurs. Right? So we know exactly where that thing should be. We can write a python script to like, grab that thing out.
363 00:40:07.420 ⇒ 00:40:16.080 Payas Parab: It will go to the website using request. Dot get this URL, and then we can actually just have it click and download it. That’s like a very doable thing.
364 00:40:16.450 ⇒ 00:40:17.590 Bo Yoon: Oh, we can make it.
365 00:40:17.930 ⇒ 00:40:29.200 Payas Parab: Yeah, there’s you. You basically would use selenium or beautiful soup to hit download. You’d create a temporary local file. You take that file and run it through Google Transcript.
366 00:40:29.570 ⇒ 00:40:30.839 Payas Parab: Pull it back
367 00:40:31.000 ⇒ 00:40:44.800 Payas Parab: and then send it to chat. Gpt to tag it, and then we’d have like tagged information. And I could make you a demo in like couple of hours, just to show you that if you want and we can do that, the the question, though the bigger question for you. Right is.
368 00:40:45.480 ⇒ 00:40:56.840 Payas Parab: is this worthwhile like there’s a lot we can do here, right? So the bigger question I would have is like, How do we frame it to Dan in a way to get an answer of like, is this a worthwhile thing for them
369 00:40:57.130 ⇒ 00:41:10.420 Payas Parab: like, Hey, we have 25,000, but we have so many voice mails and stuff that we can’t process. But we can actually go get that data for you right? Like Pius can quickly show you a demo where we like, grab a file, transcribe it.
370 00:41:10.730 ⇒ 00:41:13.040 Payas Parab: Do you want to do that right? That type of thing.
371 00:41:13.400 ⇒ 00:41:14.010 Bo Yoon: Hmm.
372 00:41:15.270 ⇒ 00:41:19.270 Nicolas Sucari: Is there an easy way to check? How many of those tickets have a voicemail?
373 00:41:19.640 ⇒ 00:41:23.010 Nicolas Sucari: Oh, I think, looking at?
374 00:41:24.380 ⇒ 00:41:27.719 Nicolas Sucari: Yeah, just with an excellent formula trying to find a link right?
375 00:41:28.660 ⇒ 00:41:31.069 Nicolas Sucari: I’m just counting the amount of.
376 00:41:31.230 ⇒ 00:41:35.740 Bo Yoon: There should be a way, I guess all the the login.
377 00:41:42.380 ⇒ 00:41:51.499 Nicolas Sucari: So if we can see how many tickets we are getting each month, for example, or in a period of time, and then we can. We can see how many of them has a had a voicemail.
378 00:41:51.900 ⇒ 00:41:59.479 Nicolas Sucari: Then we can try to use that if it is a high percentage to try to. Yeah, share that with Dan and see if that is worth
379 00:41:59.600 ⇒ 00:42:00.749 Nicolas Sucari: to do the right.
380 00:42:03.970 ⇒ 00:42:07.039 Bo Yoon: Yeah, I’m I’m not sure how to do it out in excel
381 00:42:14.360 ⇒ 00:42:17.278 Payas Parab: Yeah, also dense. I do have to run, but
382 00:42:17.570 ⇒ 00:42:18.270 Nicolas Sucari: Yeah.
383 00:42:18.270 ⇒ 00:42:19.060 Bo Yoon: Okay.
384 00:42:19.260 ⇒ 00:42:36.290 Payas Parab: But I think I can share in the chat our like our Tldr on the warranty and the discounts. We think there’s something here meaningful impact. Blah! Blah blah! What we would do next to like. Explore it. Bo, if you can get me again, you don’t need. We don’t need to like. Overthink it yet. Right? It’s just like, Can you get like.
385 00:42:36.410 ⇒ 00:42:46.799 Payas Parab: there’s a bunch of voicemails like we just need the 3 sentence version. We don’t need the. We don’t need a full. Hey? Here’s the percentage of data that can be processed by this by this, like some long prd, it’s just like
386 00:42:46.960 ⇒ 00:42:48.679 Payas Parab: few sentences of like.
387 00:42:49.030 ⇒ 00:42:58.010 Payas Parab: we know, there’s a bunch of stuff here we need to like, really dig in and like categorize and clean it up. But then we’ll be able to tell you
388 00:42:58.770 ⇒ 00:43:05.849 Payas Parab: product like, what will we be able to tell you right that he cares about Dan Level? If you can get me those 2 to 3 sentences. Then I think we’re golden, for now.
389 00:43:05.850 ⇒ 00:43:06.460 Nicolas Sucari: Yeah.
390 00:43:07.710 ⇒ 00:43:08.220 Payas Parab: Cool.
391 00:43:08.220 ⇒ 00:43:11.699 Nicolas Sucari: So payoff is shared like a warranty impact on.
392 00:43:11.700 ⇒ 00:43:14.290 Payas Parab: Yes, we just got your email.
393 00:43:14.290 ⇒ 00:43:16.019 Nicolas Sucari: Take a look at that. I didn’t know.
394 00:43:16.410 ⇒ 00:43:24.269 Nicolas Sucari: I’m not sure if that’s gonna be useful right now, but it’s something that we had there in evidence last year? I think
395 00:43:24.880 ⇒ 00:43:25.720 Nicolas Sucari: so. Yeah.
396 00:43:26.680 ⇒ 00:43:27.250 Payas Parab: Yeah, I know
397 00:43:27.250 ⇒ 00:43:34.939 Payas Parab: it’s crazy. Their takeaway was, it doesn’t present an overall sizable impact. So I gotta read this and figure out why they’re saying that.
398 00:43:35.240 ⇒ 00:43:37.949 Payas Parab: Yeah, but cool. Alright.
399 00:43:38.230 ⇒ 00:43:38.950 Nicolas Sucari: Okay.
400 00:43:39.050 ⇒ 00:43:40.640 Nicolas Sucari: Great guys great meeting.
401 00:43:41.840 ⇒ 00:43:42.519 Payas Parab: Alright. Thank you.
402 00:43:44.270 ⇒ 00:43:45.419 Bo Yoon: Yeah, thank, you.
403 00:43:46.360 ⇒ 00:43:47.160 Payas Parab: See you.