# S1 E17 - Snowflake - Prabhath Nanisetty

Episode 17 - Snowflake - Prabhath Nanisetty 

[00:00:00] Elgato Wave Neo-24: welcome to the middlemen. I'm Tom Limongello, and I'm here with Todd Sawicki. We, as middlemen, live at the intersection of media and e commerce, and we would like for you to join us in our discussions where we turn that chaotic intersection into your comfort zone  

[00:00:15] ​ 

[00:00:18] Tom (2): , we're going to do something different today. We are going to talk about what we don't understand. No matter where we look companies, whether it's retailers, brands. They all work with Snowflake and I am one of those people who pretended I knew what it was for but it wasn't until today's discussion where I think I, I finally nailed it. 

[00:00:40] Todd: I think what's fascinating to me about Snowflake, and I agree, like anybody in the ad tech world, or brand world or retail world is like, unless you're on the core it side says they know what the heck the snowflake is and does is lie. Let's be honest, there's an all by the way, there's lots of stuff that we talked about that we actually don't know what we're talking about. 

[00:00:56] Todd: So you know, that might we could potentially start there too. And, [00:01:00] but I think the important thing here is right. Snowflake is this hardcore foundational database platform technology thing. And then when it comes up and people are like, Oh yeah, I use snowflake for my clean room as someone who's helped build right ad tech companies and technology company, like snow, what snowflakes on a clean room. 

[00:01:17] Todd: Why are you talking about using snowflake for clean room? And it made no sense to me. And so it was interesting. It was like, God, everyone mentions they're thinking they're doing customer syncs and data syncs and, identity syncs That happened in clean rooms on snowflake. And I was just like, what the heck? 

[00:01:31] Todd: Where does it fit in? As people are building retail media networks. Do I really need snowflake? Or if I'm hearing in my company elsewhere that I'm using snowflake, and actually, I think going back to our conversation with Jonathan Mendes and Neuralift and talking about CDPs, people there are both CDPs being built on snowflake because again, snowflakes, the underlying data storage and retrieval layer underneath, right? 

[00:01:53] Todd: So it's like AWS, but just for data. Very specific. And so you could building a C. D. P. Like an [00:02:00] application on top of snowflake makes a ton of sense. And if you're going to build your own C. D. P. You could do it easily in snowflake. And obviously lots of companies, brands and big fortune 500 companies have built their own customer data management apps on things like snowflakes. 

[00:02:13] Todd: And I think what has happened is people then wonder, okay, if I have snowflake in this other part of my business, should I use it elsewhere? And I think that's what would Our conversation with Rabat from Snowflake led to our understanding of where does it fit in and why are people using it? 

[00:02:30] Todd: And very much is that actually one of the real interesting things that they did is how they connect data. I think this kind of speaks to your background and your own experiences having to do right, like what your experience at Ahold-Delhaize in the early days of like identity sinking really I think speaks to the problem that Snowflake is helping people solve today. 

[00:02:50] Tom (2): I think when you're a seller, when you're, even when you're in product, like I was you latch onto the things where there's a feature that you're thinking about. And so like [00:03:00] with the case of collaborating between brands and retailers. It was Oh, we need identity resolution. 

[00:03:06] Tom (2): We need to safely share identity between two parties. And so the concept of a clean room and live ramps capability to have keys that would, prevent. between the two parties in ways that weren't allowed. All of that sort of made sense to me, but the fact that it was a project didn't get off the ground was frustrating. 

[00:03:29] Tom (2): And I talked to, the head of technology Xavier, who we had on the podcast as well from Nielsen IQ. And he was talking to me about it saying, look, It's too expensive. The way it's done today at the time it was called live ramp safe haven, is that you have to make a copy of the data from each side and you have to run these copies and then there's issues with getting things out of sync and all, there's all these challenges. 

[00:03:53] Tom (2): And we were already using Snowflake, but Snowflake, what really wasn't at the point of having native [00:04:00] cleanrooms and capabilities there. Later came Habu, later came Samuha, but it was really an eye opener for me was, I always said, okay, yeah, we use Snowflake. 

[00:04:10] Tom (2): It's a data, Lake warehouse, whatever, all that stuff. I was like, okay, check the box. I don't care. Cause there's no sort of features that I am worried about. But it seems, and this is from the conversation we had with Prabhath that we're going to share today. I really got that it enables a better foundation for collaborating, it's more efficient it'll end up being cheaper. 

[00:04:33] todd-sawicki: I think another thing that was that'll be interesting that I don't think people realize is that applications are built on Snowflake. Like you mentioned Samoha and others, right? And in live ramp, like live ramp now has built a connection into Snowflake. Samoha's was natively built on top of Snowflake. 

[00:04:47] todd-sawicki: And so those are clean rooms, right? They're actually there's a clean room application that you use that just happens to be built on Snowflake. And if the yeah, If the two companies who are trying to sink their data are both on snowflake, then you [00:05:00] don't have these crazy costs about shipping data and sharing data. 

[00:05:04] todd-sawicki: It's just like it's a built into snowflakes, which is really an interesting from a core it standpoint and why companies are in this space are happy to leverage snowflakes and the apps built on snowflake is once their customer data is in snowflake. Enabling it for uses with others is really pretty straightforward. 

[00:05:20] todd-sawicki: And so I think that's one of the interesting things about this is Snowflake has apps. Third party apps that are built on top of it. So you're still using LiveRamp or Simohub for the cleanroom syncing. And it just happens to be on Snowflake. But since everybody's on Snowflake, it's super easy for these apps to work and for cleanrooms to work. 

[00:05:38] todd-sawicki: That's one of the fascinating things that comes out of this. And so when people say, oh yeah, I'm using Snowflake for cleanrooms. Yes, but through someone else who just happens to be built on Snowflake. And, I think one of the interesting aspects that you just mentioned from a product standpoint, from, in terms of the old school ways, how we used to ship snapshot in times of datasets to people and how quickly [00:06:00] they became out of date. 

[00:06:00] todd-sawicki: And you've ever looked at like cookie sinks and how quickly they devolve and age out and expire. I think it speaks to that, but if you have a system that's built on something like Snowflake and app built on top of that you're using, then the sinking can happen and be happening all real time. 

[00:06:15] todd-sawicki: So it's really fascinating to learn about this. And realize where Snowflake fits into the ecosystem is an example of a tech that I think we all talk about. But unless you're a hardcore data engineer or IT guy, you probably really have no clue what Snowflake does. And this is, I think, a great example of us trying to get to the, the weeds of how everything's working across, I think, sometimes an opaque and sometimes very technical and sophisticated ecosystem. 

[00:06:39] Tom (2): I guess it's time to rip off the band aid and hear from the source. .  

[00:06:43] Todd: Today we're starting with a great topic and to borrow the Chris Pratt meme, everyone's afraid to ask. And we're meeting with Prabhath from Snowflake. And that's because as Tom and I were talking about clean rooms and CDPs and DMPs, I feel like no one has a [00:07:00] freaking clue what actually Snowflake does. 

[00:07:02] Todd: We just wave our arms that Snowflake does something, and honestly, I think if I asked, my buds across the space, I would get a bunch of answers that would look just like a hallucination from, from Chat GPT. Honestly. No one has a clue. I thought That snowflake was just as like in the cloud raw database and as it's people started saying oh, yeah I'm using snowflake for my clean room. 

[00:07:23] Todd: I'm like, how the hell do you use a database for a clean room? That's not an application And so as tom and I talked about this we're like, you know It might be great to actually talk with someone like yourself for about who could actually tell us what the heck does snowflake actually do 

[00:07:39] Tom (2): So yeah, rather than throw him in the deep end, let's let him tell us who he is and where he came from. But yeah, let's get, we'll quickly get to the meat of it. But tell us, how did you get to Snowflake before we go into what is Snowflake? 

[00:07:52] Prabhath Nanisetty: Yeah, absolutely. Yeah. It's kind of an interesting career trajectory, but actually started out in consumer [00:08:00] goods. So I was at Procter and Gamble. We've done everything there brought from manufacturing R and D you know, creating products that are sold around the world to eventually landing in the insights and analytics space and where, you know, I've always been a data geek. 

[00:08:15] Prabhath Nanisetty: I love data. I love code. . So it was a natural fit to take this function for, of like market research and insights and actually start to modernize it. And then spent a little bit of time in Northwest Arkansas, actually working still for Procter and Gamble, but working with our Walmart merchants and and category managers to really help, , drive joint value. 

[00:08:37] Prabhath Nanisetty: And these are the early days. This was like the early 2010s. So this was before retail media existed and everyone still talked about. Shopper marketing, remember that and trade fund management, those were the the conversations of the day. And then started very early on in a company that was called InfoScout, it was one of the first mobile apps that [00:09:00] came out that asked people to scan in pictures of their receipts. We would go on to create a very large household panel. Many of the listeners might know it now and with its new name, which is called numerator. So a lot of you know, CPGs and consumer goods manufacturers are, are customers of numerator. But you know, we, we actually did a lot of work to make our application, right. 

[00:09:22] Prabhath Nanisetty: Serving up data about consumers is interesting. But. You know, going back 10 years, it was difficult to run those analytics and be able to scale that to hundreds and thousands of users without everything slowing down. And that's where I got my first taste of what Snowflake was. It's a data platform and it's a way to serve up data in a much more scalable way at the core of it. And so we were able to replatform our entire sass application. You know, the one that people would log on and run reports and get insights about their business. We moved a lot of that underlying that [00:10:00] code essentially to run on Snowflake. 

[00:10:01] Prabhath Nanisetty: And then we basically never had to look back at having to scale and handle lots of different types of data. So that's how I ended up at Snowflake. And now you know, my role at Snowflake is one of our industry leaders within our retail and consumer goods. And it's kind of an interesting role. 

[00:10:18] Prabhath Nanisetty: Basically every technology company, Usually starts with a product and they, you know, grow pretty rapidly, but at a certain point they need to actually verticalize their sales team. They need to be able to talk to that industry in a bit more specifics, right? They, they need to get into use cases . 

[00:10:36] Prabhath Nanisetty: And around that time, they usually hire people like myself that have just. Extensive experience in that industry, but then it can also bridge the gap between you know, a business person and the it or the data engineer that is going to be using a snowflake. So that's my, that's my current job. 

[00:10:54] Tom (2): So to get back to what Todd was asking you and to give you some foundation, my [00:11:00] experience with Snowflake, When I was at Quotient, we were using Snowflake. Again, I wasn't somebody who did the deal and didn't set it up. I understood LiveRamp a lot better because I understood identity. 

[00:11:10] Tom (2): I needed to make sure that we could find the consumers online. The database was in the background and I didn't really understand why it was important. But what I learned a little bit later on in my career there was that. To do a clean room or to collaborate between a brand and a retailer. 

[00:11:26] Tom (2): There was a need to share data, but the system that was set up at the time was called LiveRamp safe haven required copies of data. And it wasn't until much later that I've been learning about Habu and other others and Samuha and all the sort of. Crazy names that are related to data collaboration and very connected to the Snowflake ecosystem where I understood that you don't need to copy that data. 

[00:11:50] Tom (2): So could you explain that, where that came from and why Snowflake better positioned to figure that out? 

[00:11:56] Prabhath Nanisetty: Yeah, absolutely. And you hear about snowflake everywhere, especially in the [00:12:00] media side, but also in things like financial services and other industries, but. Yeah, just to take it kind of back to the beginning. So Snowflake is a data platform and data platform for everything that's analytics, A. I. M. L. 

[00:12:16] Prabhath Nanisetty: You name it. It's sort of a data platform. Do that. And one of the original taglines of Snowflake, I think, is very, very profound. Basically, it's to mobilize the world's data and mobilizing the world's data really quickly. How do you take this, this data swamp or this data swamp house and of, of data 

[00:12:36] Tom: Oh, so I'm starting to figure out where the lake came 

[00:12:38] Prabhath Nanisetty: yeah, the data lake data swamp, whatever you want to call it, but just this like gobs of information and actually turn it into first something useful. 

[00:12:47] Prabhath Nanisetty: But then part of that means being able to bring in things like data science and ML, but then also be able to collaborate with others if you're inside a CPG or retailer [00:13:00] or an advertising agency and then outside of outside of that. And so. Snowflake being a data platform, we kind of got our start in what's called a data warehouse, right? 

[00:13:09] Prabhath Nanisetty: Or you can call it a database in the cloud. And we sort of really invented or kind of made that a lot more popular because in the old days you know, if you wanted to do something with data, you'd have to get a server, like a physical machine that would run data. You'd have to load it up with data. 

[00:13:26] Prabhath Nanisetty: And then there's, you know, a bunch of CPUs that would. Crunch all the numbers. And that was great until maybe you start getting 1000 users or 2000 users or maybe 10, 000 users on like a Monday morning when everyone's running reports. And then everything crashes to a halt. 

[00:13:41] Prabhath Nanisetty: And so when the cloud started happening, everyone's like, Oh, we get all this infinite capability to run whatever we want. And what most database companies did was they just took that same model. And moved it to the cloud. So now when you run a bunch of data, there's , an [00:14:00] Amazon or Google or, Microsoft you're basically still saying, Hey, there's this machine, this virtual machine that I have, and I've got my data and I'm going to run it. 

[00:14:09] Prabhath Nanisetty: But now once I start getting thousands of users, I'm just going to copy that and over here. So you're kind of taking the same problem, but now you're just digitizing that same problem. And so in 2012, Snowflake came along and said, you know, that doesn't make any sense. It's not scalable. Why don't we actually just have only one copy of that data at any time? 

[00:14:31] Prabhath Nanisetty: So there's only one copy of data. And then let's bring the machines, all these compute engines and query engines or AI engines, and just bring it to the data and allow it to compute right there. Seems like a. Simple concept, but that was revolutionary in 2012 and Snowflake kind of really fielded that. What is also interesting is because you eliminate that, you know, need to copy data, you know, today, if you want to use something like a live ramp [00:15:00] port through their, through their portal, or maybe you want to collaborate with another company, what you have to do you know, with other data platforms is you have to like export that data. 

[00:15:10] Prabhath Nanisetty: Into maybe a CSV file or just some other, you know, file. And then you've got to like put it somewhere into some FTP site or like upload it into a portal. And what's scary is like, okay, what if that was PII data or something sensitive? You know, you've, you've just got files around. And so what Snowflake also did was said, well, we've got one copy of data. 

[00:15:33] Prabhath Nanisetty: You've got all these engines that can come and access it internally. But what if we said. You can then allow other people to bring their compute engine to your data and only let them see what they're allowed to see. And that's the beginning of what we call data sharing. It's a way to not ever copy that data. 

[00:15:52] Prabhath Nanisetty: It's actually bring the work. And You know, ML or AI, whatever you want to do to that data to be able to actually run that. So [00:16:00] that's sort of the foundation of what, what Snowflake is. And part of the reason why a lot of companies in the media industry and retail use us to do that is both of those industries are inherently very collaborative industry. 

[00:16:12] Prabhath Nanisetty: You can't really grow a business without being able to collaborate on data. 

[00:16:18] Todd: Yeah 

[00:16:18] Prabhath Nanisetty: So by underpinning that we've we've really grown there. 

[00:16:22] Todd: Now, has Snowflake built the actual clean room application functionality on top of sort of that data management layer? And so it may, I get it, it makes sense, right? If everybody is just, if the idea is not to share, not to make copies of data, but to share access to data, and a bunch of people are using Snowflake, You can basically build an application on top of that to say, Oh, who has access to, to match records or not match records? 

[00:16:47] Todd: And so therefore the question is, then is Snowflake actually built that sort of clean room matching functionality? So it's, it was easy because all we're really doing is building a match function. If it matches great, here's the [00:17:00] signal that says it's a match. 

[00:17:01] Prabhath Nanisetty: Yep, that's that's exactly right. So data sharing was kind of the foundation. Like, how do you actually collaborate on data without moving it? And then the next piece of that is how do I collaborate but not actually give you direct access to that data? How do I create some sort of a layer in between that allows me to control how you use it? 

[00:17:22] Prabhath Nanisetty: What kind of data you get back? And so There, there's many different flavors of it, but one of them is a data clean room where the data clean room is a number of things together. It's, you know, making sure that the two parties in this case or multiple parties none of them ever have possession of that data. 

[00:17:42] Prabhath Nanisetty: That's the first step of it. The second one is sort of defining what they're allowed to do. So it could be a simple matching right. How many of our like customer lists actually overlap. And so being able to do that where that actual overlap is done sort of in a [00:18:00] neutral, neutral place without actually any party having access to that data. 

[00:18:05] Prabhath Nanisetty: And then there's a whole bunch of like encryption and cryptography that. That surrounds it to to essentially make all of it basically secure and and privacy preserving. So all of that together makes a clean room. And so one of the things that Snowflake did a couple years ago is create essentially, you know, for lack of a better term, like an SDK, right? 

[00:18:26] Prabhath Nanisetty: A set of code that other developers can use to build a clean room on top. And so we provided all of the the capabilities so that somebody like a Disney or a NBC universals could bring their data. Look, link it up to this clean room and allow advertisers to be able to hook their own data sets about customers to that, you know, again, the data is not moving and to provide, you know more information about joint customers. 

[00:18:53] Prabhath Nanisetty: Maybe it's viewership data or better ways to activate. And so that was kind of the, the genesis of it. [00:19:00] And then more recently, snowflake also provides a user interface to make, make it easier because the, the biggest thing we heard from our customers is, Hey, I love being able to build this, but sometimes I just have an analyst and I just want to get started quickly with, with data. 

[00:19:15] Prabhath Nanisetty: But I wanna know it's secure, but I also want something easy to use. 

[00:19:19] Todd: Yeah. So I think what's interesting the nugget there is Snowflake's customers happen to be retailers and media companies. And they basically started raising their hands and saying, Hey, I need to use this data for these things. And if these other customers are on your platform too, can't we just figure out a way to make that all work seamlessly since we're all in the same place? 

[00:19:39] Todd: And so that makes a ton more sense, whereas I don't think a lot of us on the outside realized how many of these companies were using Snowflake for these large databases already. And so that makes a lot more sense versus the idea of Snowflake trying to sell Cleanroom on its own to people who have data potentially anywhere [00:20:00] versus hey, you're already using my platform, look what else you can do with it. 

[00:20:03] Prabhath Nanisetty: Yeah, that's exactly it. So it goes back to that mobilizing the world's data, Data clean room is a way to mobilize your data so that you can now collaborate in a secure way with others. And I think the technology is one piece of it. You know, Snowflake has been Done a lot of work to make this a secure thing. 

[00:20:21] Prabhath Nanisetty: And you'll see a lot of the large publishers in the world. Most recently, I think Netflix announced that they were basically live with a snowflake data clean room for for advertisers. But the other, there's a couple other reasons. One snowflakes also a neutral party. Right. We don't have our own media network. 

[00:20:39] Prabhath Nanisetty: We're not actually trying to compete with that where, whereas it's you know, it's maybe more difficult for a retailer that sees on one side there's Amazon and Amazon ads, and that's a, You know, pretty big threat to their retail media business. I mean, in some cases, you could say retail media really is a euphemism for for Amazon ads, because it's, I think, [00:21:00] 70 percent of the retail media spending. 

[00:21:03] Prabhath Nanisetty: And then, you know, on the other side, you also have companies like Google that offer cloud technology, but They also, you know, they're mostly an ads company, right? They're like 92 percent of their revenue is advertising. And so Snowflake being this neutral party where we don't have anything in there. 

[00:21:21] Prabhath Nanisetty: And I think the third piece and why many companies are choosing Snowflake is that we're also, Interoperable across technology stacks. So you can look across the retail industry, the media industry, and maybe some of the other industries, and you'll see a lot of choices on technology. Some people choose AWS, some people choose Microsoft, some people choose Google and because Snowflake is actually built on all three of them. 

[00:21:48] Prabhath Nanisetty: We are sort of that, that interoperable layer. So it makes it really easy for an agency or a publisher. To basically not have to build everything three times or four times, however [00:22:00] many times they can build something once, and then essentially that works regardless of what your underlying tech stack is. 

[00:22:08] Tom (2): And there's also the issue in retail media to your point about everything being Amazon. If you're trying to court Walmart or you're trying to court other retailers who don't want to be on Amazon's cloud because of competitive nature. I think that's, I'm starting to realize why it was like we've got all this work to do to shift everything from Amazon to Azure. 

[00:22:29] Tom (2): Now, at least Snowflake can help us with some of the issues we have with the rest of this tech 

[00:22:33] Todd: That neutrality is a big deal, and that, to the point of, Tom and I coming from a world where we were prior to Nextopedia, working in the SaaS space for e com tech, and there is absolutely a set of customers who will not touch Amazon's tech stack for competitive reasons, and I think, we, coming from the media space prior to that, I can, it's weird like what, huh? 

[00:22:57] Todd: You know as we move into retail media and obviously the [00:23:00] to me and I think people You know who don't have that background in retail don't realize those competitive aspects are critically Important to how the retail world like walmart's not going to touch amazon stack any way shape or form Like it's just not going to be the case netflix not likely to do it given the competitive threat from amazon on the prime tv side, so You It makes sense in terms of what you say there in the neutrality. 

[00:23:23] Todd: And I think people who aren't as familiar right from the media space, those sensitivities aren't as big, but in the retail space are a huge deal. 

[00:23:32] Prabhath Nanisetty: they are. And you know, there's actually a lot of retailers and companies in the space that are using Amazon. But again, it's a choice, and it's not necessarily something that other companies always, you know, One have to like understand, like, okay, to do business with this company, they're on Amazon. So now I've got to build something that links there versus this other company doesn't want to work with Amazon. 

[00:23:55] Prabhath Nanisetty: They have Google as a choice. And so that becomes difficult for [00:24:00] companies in the, in, in the space to be good at what they are doing, right. Be a great publisher, be a great agency. And then also have to like, Get into this like technology arena where it's like building. It's like the early days of web page development, right? 

[00:24:14] Prabhath Nanisetty: You had to build for like Internet Explorer, Chrome and Netscape. And like, you know, nobody wants to have to do that. And so it does help. And one of the interesting things I've Learn the other day is when you start to look at the kind of like the profitability of the retail industry, right? That's as a percentage, it's not huge. 

[00:24:35] Prabhath Nanisetty: And so there's there's operating income is kind of a big deal. Today. You know, you kind of think about it where retailers are looking at two areas, right? There's sort of operating income coming from their core business of like selling products and getting margin off there. And then there's operating income for basically all of the other alternative revenue sources, whether it's they're [00:25:00] monetizing their data, the retail media, all of that has come from a separate source. 

[00:25:05] Prabhath Nanisetty: Today, you know, it's about a quarter of. Operating income in the, in the industry is coming from this alternative revenue, like retail media, data, all of this other stuff. By 2028 I think retail media is supposed to be about 130 billion. As a, like from a revenue standpoint, that basically means a third now of operating income is going to come from stuff that's not. 

[00:25:27] Prabhath Nanisetty: Buying and selling goods. It's, it's about, you know, all the alternative sources. And what's interesting is if you look at it, right, the traditional side of the business, the biggest competition is Amazon both from. You know, just like buying and selling of goods, but then also other things that Amazon does, but in this other area, the biggest competitors actually Google they're the largest ads seller on the planet. 

[00:25:51] Prabhath Nanisetty: And so it's kind of an interesting dichotomy as as retail media continues to grow. I think we're going to see a lot of retailers start to say, well. [00:26:00] You know, there's there's stuff. Maybe I'm trying to compete with Amazon, but increasingly Google's a big competitor of ours as well. 

[00:26:08] Tom: It might be interesting to dig in a little bit on some of the use cases of how a retailer got into the place where Snowflake made sense of them. I think when we were talking initially, there was some sort of pandemic out of stock stuff that brought it to life for me, at least. 

[00:26:23] Tom: What are the use cases that a retailer would use it for? I obviously retail media is not the only one. 

[00:26:28] Prabhath Nanisetty: Yeah. It's funny when people talk about snowflake. I think sometimes people say, Oh, is that a, is that a CDP or is that a data clean room? And that's why we, we keep saying, yeah, data platform because a lot of retailers and consumer goods companies as well. They're putting kind A lot of other data in there too. 

[00:26:45] Prabhath Nanisetty: They're putting their supply chain data you know, like locations of trucks and inventory and distribution centers and and all of that. And so the idea here is. If you have sort of a, a single data [00:27:00] estate and like your data in a single platform, it starts to unlock use cases that, you know, traditionally have been very difficult. 

[00:27:06] Prabhath Nanisetty: So things like solving out of stocks during the pandemic, that was a huge area that, that Snowflake played a role. And it's because you know, you might have a system. On one side driving maybe your e commerce business, but that data is in a different format and a different capability than maybe the system that's being used for your supply chain organization and your financial planning. 

[00:27:31] Prabhath Nanisetty: And so all of these like silos of data are sitting around. Whereas if somebody had access to that data, they could create some models that say, yeah, I see inventories going down, so I should maybe turn off. Some advertising that I'm like some demand generation that I'm doing, and I need to automate those systems so that they're more in sync with each other. And then during the pandemic, obviously, there was gosh, I hate to even, like, think about all this [00:28:00] stuff that we had to, we had to go through, but like hospital, like, All the ICU data and all of the the symptom trackers and case trackers and all of that stuff, those all helped brands to be able to bring that data in and quickly link it up with all the other systems so that they could start to really manage things like out of stocks and allocation. 

[00:28:20] Tom: Yeah. What's interesting to me is that when you talk about pulling in like supply chain data, what Todd is really good at within our business, Nextpedia is taking garbage data from all sorts of parts of organizations and turning it into financial forecasts and things. And I think when we chatted initially, is, data visualization a big part of what Snowflake does. 

[00:28:38] Tom: Cause I was like thinking like, should I be using this instead of Excel?  

[00:28:41] Prabhath Nanisetty: Well, what's interesting is Again, snowflake's more on the, the data platform side. So we do have forms of visualization, but some of our biggest ways that people use it is through a BI tool. So when people have like a Tableau or a Power BI or a [00:29:00] Sigma running you know, oftentimes that might be connected to Snowflake. 

[00:29:03] Prabhath Nanisetty: So, you know, snowflake really kind of the core user is typically that data engineer, that data analyst, as well as some of the. The software engineers that are trying to use data to, to go create applications. The end user will typically use Snowflake indirectly through a BI tool or maybe an application that they're running, like, you know, my former company, the Numerator app that's powered by Snowflake, but, you know, it's got other things other things happening, and so you're not, like, that end business user is not using Snowflake directly. 

[00:29:34] Prabhath Nanisetty: Slowly, that's changing. I think AI is. 

[00:29:36] Tom: I was going to say is there, is all of that going to go away because maybe somebody like me can talk to the data 

[00:29:43] Prabhath Nanisetty: Yeah, I think that's, that's where AI kind of gets pretty exciting is that it starts to reduce kind of the, the middleman really between the business leader and that data. So ironic that we're, we're on the middleman podcast, but that's kind of where,[00:30:00]  

[00:30:00] Tom: will always be a 

[00:30:01] Prabhath Nanisetty: yeah, exactly. There's always going to be a middle, but how do we Bring that power to the end user. 

[00:30:06] Prabhath Nanisetty: I'd say, you know, I think certain things are closer, like chatbots that can help you translate a question into SQL language that can then run and answer questions. And then I think we're, I mean, like everything, Any new technology, the initial use cases sometimes aren't actually what the later use cases really are. 

[00:30:27] Prabhath Nanisetty: And so we're starting to see some really exciting things in retail and in the media space where, you know, you could potentially build audiences. It's used to be kind of a painstaking task of saying, I want to define an audience based on like these logic statements, and you had to kind of know how to put them together and string them together. 

[00:30:47] Prabhath Nanisetty: And now we've got some partners that are actually using Snowflake and some AI tools to freeform. You can, you can talk to your data and actually start to shape an audience, and it really [00:31:00] simplifies the experience and makes it much more accessible for that, for that business user. 

[00:31:04] Tom: We had a podcast earlier where we were talking to somebody who's looking to automate CDPs and to me, one of the most important things there 

[00:31:12] Todd: We can say his 

[00:31:13] Tom: the brands, that's it. 

[00:31:14] Todd: Neuralift, Jonathan 

[00:31:16] Tom: Yeah. So Neuralift 

[00:31:17] Todd: company. 

[00:31:19] Tom: And so when we talked to them, what that really sparked for me, and it'd be interesting to hear your perspective on this is giving the brands back a little bit of the power. 

[00:31:28] Tom: Cause a lot of what retail media has all been about is just giving more and more power to the retailer and having them dictate what the brand should do with their money. And if you can create audiences on the fly like that, To me, that's something that's more in the interest of the brand than it is of the retailer. 

[00:31:45] Tom: Retailer is going to not really care, actually, which brand gets the sales. I just want the sales to come in. So I don't know, what are your thoughts there? 

[00:31:53] Prabhath Nanisetty: Yeah, it's definitely gonna kind of that that data transparency and democratization is going to be interesting [00:32:00] because you know, I don't know. I'm sure the retailers understand it, but, you know, they were in much more of a buyer position and now, you know, more and more they are the seller of goods and services rather than just being the buyer. 

[00:32:14] Prabhath Nanisetty: And so it kind of from a power structure standpoint, it's probably gonna be more balanced in terms of, I don't know. You know, I see a lot of J. B. P. S. Now that are not just one sided towards category growth. It's also, you know, how are you going to grow my brand with this level of investment and capabilities? 

[00:32:32] Prabhath Nanisetty: So it'll be interesting to see how that plays out. But the brands, I think, generally want more data transparency, right? They want access to the data that might be resulting from their campaigns. Some of the the the deeper insights that that they can use to better find those audiences and shape them and get better at targeting, but also on the measurement side. 

[00:32:54] Prabhath Nanisetty: And that's another big area where snowflake plays is that because it's not just [00:33:00] You know, basic like queries and analytics. You're starting to get into kind of A. I. And M. L. Techniques. That need to be able need to be able to be done, but also done. Like right where the data is rather than, you know, needing something else. 

[00:33:15] Prabhath Nanisetty: So I, I think the, the brands are going to continue to demand more as the level of investment starts to go up. And I think we're going to start to see, you know, I think Walmart is definitely far ahead of a lot of other retailers and kind of thinking through their strategy across. You know, they've got Luminate, which kind of modernizes a lot of their point of sale data tracking and also their their, you know, kind of their shopper data as well. 

[00:33:40] Prabhath Nanisetty: And then they've got Walmart connect and they're starting to get better at connecting those two together because everyone wants the insights to sorry, like the data to insights to action. And if you can. Put all those into a single platform to be able to do that more [00:34:00] seamlessly. You've got essentially what a lot of the advertisers want. 

[00:34:03] Prabhath Nanisetty: They want to be able to understand what's happening. They want to be able to create strategies and activate it, and then be able to measure that altogether to inform the next set of activities. 

[00:34:14] Todd: As a follow up to that point gets into , whether it's the retailer or the brands working with Snowflake. Are the brands using Snowflake directly? Are they using things that work with Snowflake that are built on top? 

[00:34:26] Todd: And I think this also gets to the, wait, you're using Snowflake for a clean room? You might not necessarily be using Snowflake directly. You might be using an app on top of Snowflake. And I think that sort of gets to the, Again, how people are hand waving and they're throwing out terms and names without really knowing what it means. 

[00:34:40] Todd: And versus the practical reality of, oh no, you might not be using Snowflake directly, you might be using something that's built on top of Snowflake without realizing that you're using that app instead. 

[00:34:51] Prabhath Nanisetty: That's right. Yeah. And you'll see different examples of this around the marketplace. There's, you know, companies like Nielsen IQ, [00:35:00] and they've got a platform called Discover, which is the kind of the core platform that a lot of the CPGs are using to understand market share and all that. Well, that's a an application that is powered by Snowflake as the 

[00:35:13] Todd: Yeah, I wouldn't have known that. 

[00:35:14] Prabhath Nanisetty: you want on that, right? 

[00:35:15] Prabhath Nanisetty: And they're they're a great partner of ours and have gone from kind of the Nielsen of old to a very modern Nielsen IQ, which was doing like more data as a service and other other capabilities. So it's really amazing to watch. And so, yeah, we, you know, for the end user, you may not always know that it's Snowflake. 

[00:35:35] Prabhath Nanisetty: And so part of how Snowflake goes to market is there are it's Capabilities that you can do directly in Snowflake, but we're also a very partner focused company. We have a great ecosystem of partners that have you know, they might be sharing data using Snowflake. They might be building their applications connected to Snowflake. 

[00:35:53] Prabhath Nanisetty: And you know, for a lot of our partners, I mean, even LiveRamp being one of them they're building applications in Snowflake, [00:36:00] almost like a, like a iOS, like the Apple app store. We're building applications that are built in Snowflake and that adds another level of security where even the data doesn't have to actually go anywhere. 

[00:36:12] Prabhath Nanisetty: The application comes and does your transcoding and crosswalk generation right where your data already is. And so identity. Management is a great kind of use case for it, but we're seeing lots of other other interesting things like data harmonization, right? Like you've got 10 different data sets. 

[00:36:32] Prabhath Nanisetty: Like every CPG struggles with this. You've got 10 different data sets, you know, maybe some coming from a Nielsen, some coming from retailers and you've got to like harmonize this data together. And it's just a painstaking operation to figure out, okay, this product description looks like this one. And so we've got partners like narrative that are. 

[00:36:51] Prabhath Nanisetty: Building like native applications to just do that work and start to create almost like, you know, I would usually say Rosetta Stone. That's the actual product [00:37:00] name that they have, which is brilliant. Kind of the Rosetta Stone so that You know, in a future world, when an advertiser is working with a retail media network, they can start, there's no onboarding time because they don't have to map data together to make sure it is useful. 

[00:37:17] Prabhath Nanisetty: They could just start and say, I want to target gluten free buyers, like buyers of gluten free items with my new product. You know, cereal and whatnot. And all of that translation work is done automatically. And I think that's the, that's really the future of kind of creating ways to collaborate, but also removing all the obstacles and the barriers to to brands being able to do that. 

[00:37:41] Tom: Great. Thank you, Prabath. It was really cool to finally get into the weeds here and understand things. Thank you for joining the middlemen. 

[00:37:49] Prabhath Nanisetty: Yeah, absolutely. 

[00:37:50]