← AdExchanger
AdExchanger
The Force Of Data Gravity, With Snowflake
00:50:04
Why bring data to applications when you can bring the applications to your data? That’s what modern data platforms do, says Erin Foxworthy, global industry go-to-market lead for marketers and advertisers at Snowflake. Cloud-based platforms have...
Loading summary
Transcript78 lines
- [00:00]
Aaron Foxworthy
Foreign.
- [00:13]
Allison Schiff
Welcome to Ad Exchanger Talks, the podcast devoted to examining the issues and trends in advertising and marketing technology that matter most to you.
- [00:24]
Sarah Sluice
This episode is brought to you by Mutinex, paper pioneers in AI powered market mix modeling. Get fast answers to hard questions with Mutinex, your growth co pilot. Ask for a demo at Mutinex Co. That's M U T I N E X Co.
- [00:52]
Allison Schiff
I'm Allison Schiff. You're listening to Ad Exchanger Talks and I'm pleased as punch about that. Thanks for lending an ear. My guest this week is Aaron Foxworthy, global go to market lead for marketers and advertisers at Snowflake. What does that entail? We'll talk about it and we'll get into the concept of data gravity, which if you're not familiar with the term, is a reference to the trend of SaaS applications, quote unquote coming to you to help you with downstream marketing activities rather than you bringing your data to the application anyway. Erin can explain it better than I can. We'll also talk about AI and get advice for marketers that are implementing AI into their tech stack. Step 1 Define the use case, then understand what data is available and choose the right model for the job. If you do that, you'll be standing yourself in good stead. But first, there's still time to snag your ticket to programmatic IO Innovate, taking place in Las Vegas from May 19th through the 21st. If you want to know how to do your job better, there's no better place. You could spend a few days learning, networking and soaking up some rays. We'll have highly practical sessions on ctv, AI, Privacy Measurement, Retail media, and much more. And we're also going to record a live episode of the Big Story podcast, which is always a good time. Visit our website to learn more and reserve your spot. Podcast listeners get 20% off as a thank you for listening. Use code POD20P POD20 to get your discount. It's a steal. See you there. Hey Erin, welcome to the podcast.
- [02:24]
Aaron Foxworthy
Hi Allison.
- [02:26]
Allison Schiff
All right, what is one thing about you? And it could be anything that not a lot of other people already know.
- [02:33]
Aaron Foxworthy
I am a lover of St Bernard's so on top of three children I actually have two very large 120 pound St Bernards and I love the best dog breed ever.
- [02:44]
Allison Schiff
What are their names? I need to know.
- [02:46]
Aaron Foxworthy
Winona and Darby.
- [02:48]
Allison Schiff
Winona and Darby.
- [02:50]
Aaron Foxworthy
My kids name them.
- [02:51]
Allison Schiff
So those are really good names because I remember our our former Executive editor Zach Rogers when his kids were young and they got a dog. He delegated his daughter, I think, to come up with a name, and she wanted to name it Sarah Sink. Like the sink in a bathroom or a kitchen, if I recall. This is such a long time ago. She's in college now, and I think it's because she was just looking around the room and she's like, sink. Let's call it Sink. So Winona and Darby are pretty solid names.
- [03:26]
Aaron Foxworthy
I was thinking of, like, in sync. Right? Like, that's the fun.
- [03:29]
Allison Schiff
Oh, no, I think, like, S I N K. Well, it's.
- [03:34]
Aaron Foxworthy
People ask about Winona, like, oh, Winona Judd or Winona Ryder. And I'm like, no, it's actually a character in My Little Pony.
- [03:40]
Allison Schiff
Oh, see? Okay. Right. No one should just make assumptions.
- [03:44]
Sarah Sluice
That's right.
- [03:46]
Allison Schiff
So your title is Global Industry gtm. Go to Market Lead for marketers and advertisers at Snowflake. It's kind of a. It's kind of a mouthful. I have a facetious question, a somewhat facetious question. Do your parents, other family, dear friends, do they understand what your job is? And then how do you explain what you do to people who aren't in the industry?
- [04:14]
Aaron Foxworthy
They do not. And especially because my background for a long time was advertising or agencies or a marketer, and that was a little bit easier. But Snowflake, I think the translation that I've said is that I help customers understand how to ask questions of their data. And usually I'm talking to advertisers or marketers. That's kind of as simple as I make it. And I think that helps them understand, like, okay, like, there's a lot of data you can ask about consumers and the data you have as an organization. And it's kind of as simple as I make it.
- [04:42]
Allison Schiff
That's a. That's a good one. When I start to explain that I write about the ad tech industry, I usually end up falling back on the most obvious example because I get blank stares. And so I end up saying, you know those ads that follow you around the Internet? And then the stare. The blank stare turns to, like, anger, and they're like, are you responsible for that? I'm like, no, I write about a lot of the companies that facilitate that, but there's a lot more going on than that. But it's so hard to explain.
- [05:13]
Aaron Foxworthy
It's just really hard to explain. Agreed.
- [05:16]
Allison Schiff
Well, what does the job entail, though, right? You make your cup of tea or coffee. I don't know what you drink in the morning. And then what do you do as a GTM lead for marketers and advertisers at a cloud based data warehouse.
- [05:32]
Aaron Foxworthy
Yeah, so this is, it's a really fun job because this rule sits in between product, but also really very much kind of how we explain to our customers. Right. How Snowflake can solve their challenges. And I sit specifically in a role that translate that from the technical into the business. And as a cloud data platform like Snowflake can really answer so many challenges that the modern marketer or ad agency is facing. And so it's exciting because I get to define the use cases and work with product and think about.
- [06:07]
Henry Innes
Right.
- [06:07]
Aaron Foxworthy
Ways that we can talk about that to the industry and evangelize some of those ways we are solving their challenges. So that's really my role, my team, you can think about them specifically calling on the buy side of the industry. So I spend a lot of my time obviously talking to various departments of marketing organizations and also within ad agencies. And then we have counterparts obviously that call into more of the sell side. So they're calling on SSPs and DSPs and kind of martech companies. But my remit specifically is the buy side.
- [06:37]
Allison Schiff
I feel like Snowflake is often misunderstood. Like people are a little reductive when they describe it like I just did. Right. They'll call it a data warehouse or a cloud based data warehouse and then just leave it at that. But I know there's a lot of nuance there. So what, without jargon, what exactly does Snowflake do for advertisers and, and marketers? And I want you to pretend that you're at like a cocktail party and the person you're explaining this to is two vodka tonics in and maybe they're holding a glass of wine. And so. And they want to know what Snowflake is. They asked. It's, you know, you're not just talking their ear off, but what, what people you really use it for and who your customers are and. Yeah, just how it, how it works.
- [07:30]
Aaron Foxworthy
Yeah, I'll take it from. I think the easiest answer take it from is let's talk about it like, like I'm talking to you as you're a marketer. So that's the kind of an easy way to start because we talk about this a lot.
- [07:39]
Allison Schiff
A drunk marketer.
- [07:40]
Aaron Foxworthy
Yes, drunk marketer, 100%. The first thing we talk about is, as an organization is. And this is every marketer's right challenge right now is that we have so much data around our consumers and we talk about the forms of that data Right. That could be on a mobile app, that could be on a website, that could be in a PDF document, right? That type of data, you'll hear words like structured and unstructured and all those different formats that they're in. The hardest part is that how do you bring that into a foundational view that you can start to actually create really amazing opportunities to use that data within your marketing suite and across all your operations of marketing. And so what you need to think about in those decisions is a platform that can handle that scale and all those different types of formats and the speed that they come in. And when you start to think about where that lands, that requires a platform that is agnostic to the ecosystem, that works with various different clouds, and that can handle kind of what we call compute. This is just the speed, right? The engine it takes. Ask questions of that big data that's coming in. And that's really where Snowflake has excelled. Now, the exciting part is that once that data is landed and kind of cleaned and organized, right? Which is where a lot of that data engineering and your IT teams are kind of coming in and doing that organization. That's hard, right? We talk. I say that very easily. In a simple terms, it's very hard to do. But once it's done, what you've built is like a foundation of a view of your customer. And now what you got to start to do is think about things like how do I bring my consent to that data, how do I make sure I govern that data, how do I think about the granular privacy policies? As an example of what data can I use, right. Is that someone under 13 are these segments that really that my privacy team would say should not be activating to downstream tools. So now I'm starting to kind of build out on top of that foundation things like how do I build trust with my consumer? You're doing that by looking at your governance policy, by bringing that very quickly to your consent data. Then what starts to happen once that data is set and governed and under a really good, clean view? The interesting first snowflake, and you'll hear this term, data gravity, is that applications now, which you can think about a SaaS application, are actually starting to move to that data, because in a lot of organizations, once they've done that, specifically, if you think about a regulated industry, the last thing you want to do now is start to make copies out and kind of lose that governance and control and view of your data. So the industry is using this term called data gravity, which just means that these applications are coming to you to do jobs that'll help you in downstream marketing activities. So a great example is called an application or something we call native application, which is a framework that we created inside of Snowflake. And these applications can bring things like identity resolution. So that identity resolution can now come to your data. Look at your first party data and resolve it. Bring household information, maybe they clean it with hygiene. And those applications make it really easy without moving data to really bring that resolution to the data. Another example, a public example, Allison, is some of the walled gardens have built applications to allow you very quickly send that data to a CAPI integration, so conversion API. So instead of you trying to figure out a data model, how do I send that really complicated data set down, really streamline and quickly to an endpoint so I can optimize? These applications are now coming to the data sets to do that work there for you. What you're seeing is this convergence. MarTech companies CDPs are building to that data and ad tech companies are building to that data. And so the marketer is sitting in this wonderful place where they can say I can have all this beautiful view of my data and really start to orchestrate how that data flows between my own channels, like how I do email and push notifications, but also how do I do that to paid, how do I send that down to my paid channels and then how do I bring that data back and do measurement? And so it's really that first piece we talked about of getting all that different type of data into your platform is hard. And it definitely is something that is an investment for a modern marketing organization who wants to transform their business. But it's opening up these amazing opportunities downstream to collaborate across the ecosystem.
- [12:05]
Allison Schiff
I love the term data gravity and I have not heard it before. This is new to me. The concept makes a lot of sense, but I've never heard anyone say data gravity before. I really like it.
- [12:16]
Aaron Foxworthy
We talk about it a lot.
- [12:19]
Allison Schiff
I had miles younger on this podcast. It was a couple of years ago now and I don't know if you know him, he's the Chief Growth officer at U of Digital, which does digital marketing education. Yeah, he's a great guy in training and we were talking two years ago about the trend of ad tech converging with cloud technology, that we're seeing more and more cloud based platforms like AWS and Google Cloud platform and Azure and Snowflake obviously become the foundation or like the facilitator for technologies that are really central to how digital advertising Functions like clean rooms. And that puts me in mind of data gravity or real time analytics tools or a data marketplace or any number of things. So what's driving that convergence. But I think of it more as like a symbiosis, you know, because they really do. There's a really good interplay here. And where does like Snowflake enter the scene there?
- [13:16]
Aaron Foxworthy
Yeah, this is a great question if you think about. And it's. I think a lot about kind of my journey to Snowflake when I answer this question. Like we were. The word is that when I worked at my last ad agency, we were kind of in house in some, on some marketing organizations and we were challenged with this concept of we have no data silos. We have this. Before we stand up our MarTech team and our ad tech team and our paid ads team, we want to think about how we just build the best view of our data before we decide Downstream is really modern thought. This was almost seven years ago. And so we're like, wow, how do we do that? Right? So we're going to think about, maybe we do have. We actually have to build our first party data set, right? We have to think about what that means. We probably need identity. So the first thing we kind of went to back was kind of more of this DMP world, right? How do we tag data and build these segments, you know, inside of a SaaS platform and then, okay, now do we need to roll that to a cdp? But what we found is that within a lot of those kind of applications, we felt very rigid and kind of locked to the view of our data. We couldn't ask the questions or kind of write the questions to the data that we really wanted to see. Sizing an audience by making decisions. And the reason why that was happening is that what we'd found, especially because we're coming from a paid ads background, was that our first party data was always our best performing data set. Whether we were using it as a seed and then turning into a lookalike model or we're doing the look like models or propensities ourselves and building those models, that was always the best performing data set in any channel, whether that was social or linear, etc. And so when you start to think about cookie deprecation and you realize one of those important things you're going to have in the paid ads environment is first party data, you naturally start to think about, okay, well I need to be able to have really strong flexibility in how I think about and enrich my first party data. So it's as accurate as possible. And then if you think about your CRM or your typical kind of loyalty growth teams, right, who are using more of that Martech features, deploying emails, they're doing the same thing, right? They're using that same data set of understanding someone's propensity to churn or what pages did they visit or are they part of my loyalty program and they're thinking the same thing. That same data set, right, is really what fuels kind of the owned ecosystem, right, the Martech ecosystem. But it's the same data, it's the same person. And so it starts, you think about going back to that first conversation we had is if I have a clean view, I have that data coming in, I understand it's consented. I built kind of the trust that I need to do to protect my consumer data. Now I want to be relevant across both those channels. And so, you know, in a large marketing organization, there's hundreds of marketers and a lot of them are in a very specific type of channel. But if you think about kind of moving up and you have that data and you have the understanding of where your consumers are kind of touching, right? Your, your own channels and understanding paid, it naturally makes you start to think about how do I start converging if I have an email deployment and it wasn't opened, or I push notification and they're not responding. Or maybe they told me in my consent practices that I can't email them after a certain time or. Right. So I can send them an ad. And I think that what that's what the exciting part is, is that once you have that view kind of upper right of your, of your SaaS applications into a platform that allows you to make the decisions, that's where we start to see the convergence. And it's, it's new, right? So we start to see there's a customer we're working with that, you know, is doing a push notification into like a CTV channel and then they want to follow up with an ad, right, to make sure they're hitting some decent frequency. That's really very, very, very challenging, right, for a modern organization to do. If you don't have something like Snowflake sitting there to unify those two different channels.
- [16:56]
Allison Schiff
Ad Exchanger published this story back in September 2022. So it's like three years ago. It's roughly around the time that you joined, I think a little bit after or before. When did you join exactly.
- [17:09]
Aaron Foxworthy
Snowflake March, three years. Yes, I just hit my three year.
- [17:13]
Allison Schiff
So this story, it had the headline snowflake is aggressively pushing into martech and advertising, written by James Hercher. And it was specifically about Snowflake publishing its first and what would become an annual report called the Modern Marketing Data Stack Report, which aggregates how your customers use cloud and Martech vendors. And it's this analysis of thousands of media data cloud clients. And the media data cloud, for listeners who may not know, is Snowflake's suite of data products and integrations for advertising related services. So like measurements and attribution, identity resolution, profile enrichment, activation, all of that stuff. But my question is why is Snowflake investing so heavily in adtech and martech?
- [18:02]
Aaron Foxworthy
So from the position I sit in where, right, we're talking to the marketer, if you think about again, I'll go back to this concept. A lot of data gravity a lot of times for the Martech and edtech community to collaborate with our customers who are marketers, they are coming to Snowflake because that's where the data lives. And so it's natural, right, for us to start to see this ecosystem of data providers through our marketplace or you know, ad tech providers offering ad logs or native applications, right. To do things that we talked about earlier, the CAPI integrations, that ecosystem needs to come to the first party data because that's the fuel, right. Of the industry. And so I think what you're naturally seeing from a marketer perspective is the pull of the gravity to that first party data as ad tech and Martech are customers. But on the flip side, what's really interesting too is that our platform is really powerful for those as well, right. So if you think about, if you're standing up a retail media business and you want to consolidate, right, all your sales data, you know, between your on site and your off site. Snowflake's also a solution on the sales side as well. And so I think whether it is building applications between AdTech and MarTech to come to the marketers data set or to use Snowflake for your own internal tools, right. For the industry, we're answering both sides. And so I think that's why you see this huge growth, right, of our kind of marketing data cloud and marketing stack is that everything is coming to where the data lives. And if the data lives right in Snowflake, that's that gravity that we're talking about. So it's just, it's amazing to see it in three years to watch the growth. I mean, you'll see our new report we're working on it right now it's the amount of growth of this industry kind of moving to this direction. It's exciting. It's really fun to be a part of it.
- [19:50]
Allison Schiff
It's all gravity, I guess. Do you want to give us any little teasers from the new report?
- [19:59]
Aaron Foxworthy
Let's see. Last year we talked a new entrant. Last year was definitely more around privacy. We're seeing consent and privacy move a lot closer into our ecosystem. I think this year in the report. Obviously, you know, we talked a little bit about this year and we're going to have a lot more detail around how AI is being leveraged in the platform and some use cases there and some partners you're going to see growing. I would say that's probably going to be definitely a key area for the report this year.
- [20:26]
Allison Schiff
So I'm glad you brought up AI because the entire second half of this episode will be about AI But I want to ask you a question before we hit our break. Obviously first party data is the best, et cetera, et cetera. But I want to, I want to get your opinion on this whole third party cookies not going away thing. So are you happy? Sad, Indifferent, Baffled, Angry, Rageful, Apathetic, Annoyed. Like what was your reaction to the fact that Google isn't definitely not deprecating third party cookies at all? Ain't happening. What emotion did you feel or were you just totally benumbed by this point?
- [21:04]
Aaron Foxworthy
I don't want to say it. I do think I was slightly benumbed. It's interesting. Like it depends on what hat like in my further career. Right. If I was back on the buy side as a marketer I would probably be furious because I spent years trying to understand what my future state was going to be and making actually very specific investments probably to my infrastructure. I think it's Snowflake. You know, it's a little bit, I'm a little bit more numb only in the fact that right. Cookie based data, ad log data, aggregate data, like all of that data lands right in Snowflake. First party, all of it. And so I think I take less of a stance but I definitely feel for so many companies that made pretty significant investments in preparing for that. I um. So I'm sure there's quite a bit of enraged people out there.
- [21:53]
Allison Schiff
The reactions on LinkedIn were choice. Some people were pretty mad. I spent I don't years writing about this. I mean the amount of digital ink we spilled. My goodness gracious. I mean literally hundreds of stories.
- [22:08]
Aaron Foxworthy
Yes, I agree. And then it's funny Because a lot of people will say, like, oh, I knew that was happening. I knew they were going to back out. Like, did you? Did you really? Yeah. It's interesting.
- [22:17]
Allison Schiff
There are a couple of people that I would give props to. Jeff Green from the trade desk. He was skeptical from the beginning, of course, big horse in the race. But there's also another guy from adstra who wrote a few columns for us and he predicted it wasn't going to happen, and he was right. So it's, it's a little bit tricky when you, you know, draw a line in the sand and say, I think this or I think that, and then you publish it in a data driven thinking column on Ad Exchanger because people can look back and say, ooh, bad take. But it's pretty gratifying when you call it.
- [22:58]
Aaron Foxworthy
Yeah. I won't say the name, but I remember talking to a CTO of a global holding company and they said it, they're like, they're never going to do it. And I was like, I remember in my mind being, what does he know that I don't know? What does he know that I don't know? And then here we are. So it's interesting.
- [23:13]
Allison Schiff
Well, go ask him for some stock tips.
- [23:15]
Aaron Foxworthy
I know. I want to. I want to.
- [23:18]
Allison Schiff
All right, we're going to take a quick break and like I mentioned, it's going to be artificial intelligence for the whole second half, so stick with us.
- [23:33]
Aaron Foxworthy
Foreign.
- [23:39]
Sarah Sluice
Hello, I'm Sarah Sluice, executive editor of Ad Exchanger, and I'm here with Henry Innes, the co founder and global CEO of Mutinex. With a focus on driving product development, Henry leads the Mutinex product vision. He's obsessed with what's next for customers. With a background in software engineering, marketing strategy, works with a lot of the world's largest brands. Welcome, Henry. Let's dive right in. Every organization talks about being very data driven in their marketing, but a lot of them struggle to get there. What's holding them back, in your opinion?
- [24:16]
Henry Innes
I think the number one problem and issue holding any organization back is the cost of asking a question. So if you think about the cost of an organization asking a question, when a CMO goes to ask that question, what tends to happen is a cascade of things happen. People look at various platforms, they look at various sources of truth, they wrangle that together in a presentation and it takes weeks and weeks and weeks and arguably costs millions and millions of dollars. So the number one way to solve that is to build systems and processes that go beyond just aggregating and building models around data analytics, but are able to deliver answers to complex questions quickly, immediately and without cost. And if you can start to solve that problem, that allows you to answer more questions. When you can answer more questions, you can become more data driven.
- [25:05]
Sarah Sluice
So there's a lot of analytics tools out there now. Why isn't it that the tools are solving the problem?
- [25:13]
Henry Innes
Well, I think if you have a bunch of data scientists building tools for other data scientists, it becomes ridiculously hard to build something workable for actual organizations and users. So I'd say that's probably the first thing. I think ultimately it's because we're all focused on the wrong thing. You know, we're focused on more data, not, not good answers. And you know, and I think ultimately every data product focuses on generating more and more data for organizations, which adds more and more cost to organizations. When you have more and more cost to organizations being added from more data, what ultimately happens is they get overwhelmed. How many times have we seen analytics platforms being added in only for new teams to spring up around them, new owners to spring up around them, and them not to be integrated into the business? So we focus very much on ignoring the data problem. And in fact, you know, we know organizations don't want more data, they want less data crafted to better answers. And so ultimately that's, that's what we shoot for, right?
- [26:13]
Sarah Sluice
Less data crafted to better answers. I really like that as a North Star. How do, how do you fix this answers problem?
- [26:23]
Henry Innes
Well, the first thing is to acknowledge that no one wakes up in the morning wanting to buy a market mix model. Everyone wakes up kind of, you know, going, I have a problem. What's the data I need to understand to answer that question. And so that is the product that you're ultimately trying to build. You're trying to build answers platforms, you're not trying to build market mix modeling platforms. The entire framing of the MMM industry around models betrays the fact that it's not a customer centric or focused industry. And so I think if you start to change that fundamentally and start to go, okay, how do we get really good answers? Where we have very fast models that we know are very accurate and have extreme granularity so that we can actually select the right components that matter to analyze a problem, rather than providing very high level generic answers with lots of strategic waffle to justify those answers, then we start to get to a much better place.
- [27:20]
Sarah Sluice
So I hear you alluding to what you're building at Mutinex. How does your team Tackle this challenge, give us a little more detail.
- [27:29]
Henry Innes
So we tackle the challenge firstly by, you know, making data ingestion very easy. You know, if you, if you can't take in unstructured data, if you have to spend weeks and weeks in data, it's impossible to get to a good answer quickly. We then focus on having models that run really fast in a very generalized fashion. The reason why is that means those models are not biased by analysts, and they also stand up over time and they're more robustly tested across different environments and scenarios. And that means they're more reliable for forecasting, they're better in parameter recovery, they're better in holdout testing, which are objective tests that you can run. And for. Finally, we then build usable interfaces that sit over those generalized models that select the right answers with the right level of granularity to present the information that a user needs, rather than forcing the user to consume lots of data that they don't need.
- [28:25]
Sarah Sluice
Well, I like this approach. It's very thoughtful of what people need versus just trying to maybe push another tool for the sake of a tool. As we were talking about early in the conversation, thanks for coming on the podcast, Henry, and thank you to mutinext for supporting our podcast.
- [28:44]
Henry Innes
Thanks.
- [29:01]
Allison Schiff
All right, we're back. But before we get back into it, I want to talk for a minute about the. The path that brought you here. A little bit about your background, because like you mentioned, before joining Snowflake, you spent seven years at Horizon Media, so on the buy side, and you had a couple of different roles. SVP and Managing Director and then EVP and Managing Director of Innovation and Partnerships. So you know how ad buyers and advertisers think. So what kind of questions are you fielding from marketers, like, in your role now? And is it helpful that you can speak their language, you know, in a way that maybe other people at an organization that sells cloud based tech might be less adept at doing. Just naturally, you know, people like to be sold to, if they have to be sold to by someone who at least understands their needs.
- [29:54]
Aaron Foxworthy
Yeah, that's. That's the exact point of the team. Right. And the team that we're building is that we come from walking in the footsteps, right. Of what it means to be, you know, working at an agency or kind of working in a marketing organization. And it does help because I think that sharing your journey and kind of explaining kind of the pitfalls of how you got to where you are and why Snowflake was your choice and having that firsthand experience. Exactly it sets a foundation of trust. Right. And collaboration that I think is very different than if you were just coming in and trying to sell, right. A platform to somebody. So it does make a big difference. My, my journey to Snowflake I kind of touched on a little bit was really interesting and, and I joined Horizon actually I don't actually have it as a quick stint at Rubicon project before, before it was Magnite and then quite a while at Microsoft Advertising. And that was an interesting time because it was when they had made the Equanib acquisition and there's a lot of yes. So I was a big part of. So Jeff Green's ADCN was there and we had an early ad server and Scott Howe was running the business unit. Right. Who was obviously at Liveramp. So it was a really interesting time to be there. And then obviously with the rise of Bing and so they kind of migrated kind of slowly out of the ad business. At the time I was there and I really didn't want to go back to agency side because I'd been there early in my career. But I got this really unique opportunity to come back to Horizon Media which as an independent had a little bit more leverage on how they structured teams. And they really brought me in on that title to say to sit back and think about really what does it mean to think about innovation and data inside of an organization. And I was able to build this amazing team of people, very forward leaning, curious team that really got to sit back and think about for very forward leaning clients. Kind of what I talked about earlier, which is how do we become more data driven? Right. There's so much waste in how we buy media and we want to be more relevant to our consumers and then we want to understand and measure what's really working and we want to do it the right way because there's so much legacy silos of data and applications that people purchased. And so we kind of sat as this team and we got to walk in and really start to kind of test and learn what was working. And that was that journey that kind of brought us to Snowflake, which was if you really wanted to get more granular and have more control and transparency over, right, your ad buying and your marketing, you needed a platform that allowed for that because. Right. Even just the conversation of bringing back ad logs right from an ad server into a platform so you can understand, you know, win loss rates or you know, even just standing at PMPs and understanding they're delivering quickly, like those basic things that, you know, that that happening some is in platform but you want more control of. You needed a place for that data to come. And so that was kind of my natural progression to Snowflake was seeing that I had all this amazing data and I wanted the flexibility and the transparency right. That I desired to be a better marketer. And that's how I got there. And so we did that. It was the journey though, right? It was definitely. That's, that's seven years. You know, we learned a lot over what it means to have durable identity and security and better relevance of our data set. And that was my journey to Snowflake. And so the other thing that I thought was interesting kind of the really the pushover for me which we mentioned earlier was we were also testing really early as data hub, faa, Amazon, marketing clouds, clean room. And so then you start to see, well, Snowflake rolling out, right? A collaboration and a data clean room. And then I was like, okay, wait, like my entire data set and all the view of my customer data is here and now I have these powerful tools to protect my data with collaboration to the ecosystem. And that's when I was just that I got to make the jump because that's where it's going.
- [33:50]
Allison Schiff
So that was your journey and marketers in general are on kind of a journey now with AI things. That's my, my little segue. I mean the amount of noise and the sheer volume of BS that I think marketers have to deal with when it comes to AI solutions. Not that there's a lot of value and a lot of valid solutions and use cases too, but I have pity and like some measure of sympathy because if I don't get pitched on at least five companies with supposedly game changing AI solutions that are like insanely thirsty for coverage before lunch, it doesn't even feel like a day anymore. So do you have any advice for marketers? I imagine they're being inundated with tons of new SaaS tools with supposedly magical AI solutions for just for all the things.
- [34:42]
Aaron Foxworthy
Yes. Couple. There's a couple questions I would recommend they ask and I'll stay is kind of how I think about them when it comes to a marketer or an agency. I think the first one is to understand is to ask the question, which we do a lot is what's the right model for the job? Because there's so many right models that are out there, you know, open source models and others and understanding what's the right model for the right task. Right is important and a lot of that goes back to defining just the use case. Are you trying to ask questions of a large data set? Are you wanting to actually create a natural interface so I can chat to my data? Are you just trying to search across and understand maybe the value of data that's sitting inside of the database? And so really what's the use case? And then the question becomes is what's the right model? But I think even before you get into which model and what's the right solution of the model, of the task, I think the big question I always have as a marketer is, well, before I get to that, how is my data being protected? So I have this, we go back to this conversation of data gravity in Snowflake. What's great is that the models actually come to you in your data, in your environment. Because I think that what a lot of marketers are doing now is they're sending out this really secure, amazing IP out to an API endpoint to maybe build a chatbot or think about how do they create an image based on that first party data. But I would ask the question of when I do that, when I send that data out from my governed control and what am I feeding? Like what am I teaching? Am I building a competitive Solution? Is this SaaS application, right, taking my data and building out models that they're applying to other businesses. So I think there's a lot of early conversations of kind of we talk about, right, the security of our first party. But also now I think it's important to think about, I want a place that I can bring that those same models, right? All the best models of the industry to the data set that I've secured. And then what I get to do. Because the next question you have is like, okay, well what's accurate? Like I don't want hallucinations, I don't want anything happening, right, that's going to hurt my brand as a marketer. And so the great part about thinking about AI running in your foundational, right, data platform like Snowflake is that there's really, really fine grained tools to allow you to understand how that data works with the model, right? So which data, how do I want to parse it? How do I want to fine tune what's the data? I want to let the model read and that that learning stays in my environment. And so that's what I think a lot of the industry is going to start to think about because rip of our brand, our tone, right? The language, our customer data, that is the value of our organization, right? Especially when you start to think a lot of these businesses like Allison are starting to become retail media companies and then they really understand the value of their data. So I think that there needs to be a couple more, you know, questions marketers really need to think about, about how do they govern and keep their IP of their customers protected while they get to use these amazing new tools and features to answer questions and needs. And so we coach a lot about the some of those questions to our marketers and, and kind of explain the difference on on not only what model, but where should that model run? That's important.
- [38:17]
Allison Schiff
Well, I'm glad your answer was a little more like upbeat than the tone of my question because I don't want to be too cynical. AI is obviously transforming so many things and it's not really a question of whether to embrace AI and integrate it into your process or business. It's more like how to do it and how to do it right. And at the risk of repeating myself because I think I've said this on this podcast before, but I've been personally surprised by how quickly perplexity and I use perplexity for some reason. It's just the one that I was exposed to first rather than ChatGPT or Claude or any of the others. But I'm amazed by how quickly it's supplanted search for me. I have to be careful because it's not always correct. I have noticed that. But the summaries of information that it's able to pull together for me when I'm on a deadline, I will admit it's been very helpful. Like I have found myself and asking fully formed full sentence questions of perplexity, including follow ups. And as recently as two or three months ago I'd have just done a bunch of keyword searches using Google. For the record, Gemini does not do it for me. I have not found value there. But anyway, I do have a question. So I promise I have very strict parameters for myself in terms of how I'm willing to use AI to help me with my work. Like it's a research tool and that's it. And I'm really only using it to research things that I already know about so that I know if there's something wrong in there. I'm just using it to help me really quickly aggregate. So it's a form of research, but to bring it back to marketers like what can they do to make sure they're using the right tools for them and that the tools they choose to use are as accurate as possible as secure as possible and actually make their lives better instead of them just trying to bow to the machines, you know, because their boss told them to.
- [40:16]
Aaron Foxworthy
That's right. I think that's exactly the question. And I think that again it's. While Snowflake is a really, you know, obviously technical platform, right. It's made for technical builders. I think that marketers need to understand what's happening under the hood, right? So we go back to, you know, defining the use case. So, you know, I want to build, you know, a chatbot to understand how my consumers feel about a new product launch, right. And I know what they are because they're part of my loyalty program. You know, when I want to ask, you know, I want to be able to have them ask questions and answer questions to understand that. And I think that what you're going to say to yourself is, okay, well how do I make sure that wherever I build that at is making sure that, that that data is. I'm allowed to use that data. And then again, the model is only learning about the data that I'm allowing it to learn. And so that again that's. And then the cool thing about Snowflake is that there is the ability also. And we have this in something called Cortex Studio. Cortex is the name of our foundational AI services. But you can say I want to ask this prompt and this question and I want to put two different LLMs next to each other and I want to see which one actually outperforms the other, which one's more accurate. And you can actually see that in Snowflake. It allows you to test and see accuracy of various models. And that's what's nice about again, thinking about the platform you're running the models on, not just the model itself, right? Because you want that optionality, you want to understand the accuracy and also cost. So one thing that marketers also have to understand is what's the right model for the right job to make sure that I'm paying for the right model for the right job. And so that's what our platform really allows. So I think that to again, the sensitive data you're going to use around your consumers, you know, even to build Gen AI, right. I want to use first party data to think about building an image. Well, you really want to protect, right, that that model is protecting that IP and it's again, not training outside of that. So again it kind of comes back to asking the question of not just the model, but where is your model being run and what are the actual technical and then also probably legal ways that you're protecting that data.
- [42:25]
Allison Schiff
And Snowflake has its own LLM. Right. I was reading about Snowflake Arctic. I love all the winter imagery. But I only found out about it honestly because I was doing some reading in preparation for our chat. I wasn't familiar with it, but it's an open source LLM designed specifically for large businesses. That's what I got from reading online. But tell me more about it, like what makes it different? And I'm assuming you guys just develop that yourselves. Completely.
- [42:55]
Aaron Foxworthy
Yeah. So a lot of obviously the amazing IP that we have around AI comes from obviously Shamaswamy and the acquisition of Neva into Snowflake and kind of bringing those amazing minds that, that have come from Google and other places into Snowflake. And so what you can imagine is, is the amount of SQL that we understand. Right. So us obviously building our own model around what's core to our business and being able to allow that to be, you know, an amazing, you know, text to SQL based LLM. Because we know that so well. Right. And allowing our customers to leverage that inside of our platform was very natural to us and natural to kind of, you know, some of the talent we brought over from Niva. And I think that that's what's great about our platform is that, you know, we'll be able to say this is the models we have for you that we've created because, you know, I would say we probably know SQL better than most people in the world, but there's also different types of models for different jobs and so we option that as well. Yeah.
- [43:55]
Allison Schiff
Out of curiosity, by the way, what AI tools do you use, like for your own work or maybe in your. In your daily life? And why?
- [44:04]
Aaron Foxworthy
It's a good question from a consumer side. I've tested a lot of them just because I'm curious like you. Right. I've tested some sort of perplexity. I want to see Gemini. I tried Claude, I've even tried Grok. I use Chat GPT. I don't feel like I've been loyal to one yet. Maybe I think that's just because I. Maybe it's just my curiosity. I haven't had one that's really kind of answered all my questions in a way that I feel like I'm really tied to it. I would say if I had to pick one, I would say maybe Chat GPT I use quite a bit. I feel like a lot of the fun stuff that I've been watching is a lot of like the typing in a prompt to getting an image back right from a consumer piston. I can see my kids kind of playing around with that and we'll kind of test the different models and kind of play around with what we can generate. And I think that that's really fun from a consumer perspective, but I don't feel like I've kind of chosen my specific companion just yet.
- [45:03]
Allison Schiff
You're familiar.
- [45:04]
Aaron Foxworthy
Yeah.
- [45:06]
Allison Schiff
So I'm a marketer and I want to implement AI writ large into my tech stack and into my day to day operations. It's a journey, like we were saying. But what would you say are the first? I don't know, three concrete steps to take.
- [45:25]
Aaron Foxworthy
So define your use case. That's obviously number one and sometimes that's just hard for an organization because you don't know necessarily like all the options that you can do. So it's really like where is a challenge where I can find either business challenge that I have where maybe I'm locked inside, I can't understand a data set because it's, you know, it's hard to understand it or there's an efficiency that I know I can gain. Right. That I'm, that I'm trying to unlock. And so that's your first piece. The second piece is really understanding, well, do you have the data to unlock that? Right. Because we talk a lot about there's no AI strategy without a data strategy because if you don't have the data for the model to read, then you obviously are not going to be able to. Right. Leverage that use case. So really defining the use case, then working on what is the data that I need to do this and then kind of spending the time, you know, working with your technical counterparts to again which we talked about earlier is what's the right model, what's the right job, right for the task and what platform gives me the optionality to not have to tie into one model and allows me to have kind of variants because this is such a fast changing industry. So let me think about where I want to actually create that, you know, AI use case to give me enough optionality and to let me kind of test and learn as I go. So those are the three things we always walk in. So you know, I'll give you an example. We were Talking to a B2B marketer the other day and they were really, really struggling with some of their web based data that was coming off of some of their analytics tools and they're like, you know, it's getting stuck in a dashboard and it's not updating really quickly. Like can I just build like a tool that allow me ask natural language over, you know, my website visitation. Like absolutely. Does that data, where's that data live? Oh well I'm bringing that through into Connector into Snowflake. I'm like, well here you go. Right. So that's it, right. Just finding that business use case and like how you get the data, is the data there and then you start to get. I mean that's what's wonderful also about Snowflake is just how easy it is, right? So to a technical, you know, someone that's in Snowflake, I mean just a couple lines of SQL and you're, you're leveraging AI in Snowflake. So that was the, that's, that's always been core to our platform is that we just make it easy. So sometimes it's not even the actual AI part, it's more about do you have the data and what's the use case that needs to get to find. So those are the three that we kind of walk into a room is, you know, it's, it's up to you, right to really think about where can I see the business unlock and then from there, right. Usually if the data is in a place that it can be leveraged then, then you're off and running.
- [47:54]
Allison Schiff
So we're, we're nearly out of time. My last question before we hop off, which I'm sure is something you'll be able to easily answer on this call. Snowflake has made a few very interesting acquisitions recently. You mentioned Neva, the AI powered search engine and one time sort of Google competitor. And then there was the data clean room startups in Muha, which is really interesting. Kamakshi Sivarama Krishnan, who's very brilliant, so hiring a lot of very smart people. Also Sridhar, what else is Snowflake gonna buy?
- [48:30]
Aaron Foxworthy
Well, I can't speculate on. I think Summit is a really good time to come. So I'm going to plug that because we will make some announcements there and it's right around the corner, it's just in June. You watch closely because I think Snowflake Summit is always where kind of our product teams will come and announce what we're doing next. I actually wish I knew because they keep that under a pretty good lock and key. So I can't even give you some good highlights. But Summit is the time so just in a few weeks you will probably see. I'm sure our PR team is working very quickly right now. To kind of prepare some of the announcements we're going to see. So watch for Snowflake Summit and then you'll probably see some new potential, either product releases or other announcements.
- [49:15]
Allison Schiff
Well, this has been a pleasure and please email me some stock tips from that person who told you that third party cookies weren't going away. Because I'd like to retire early. Thank you very much.
- [49:27]
Aaron Foxworthy
We'll call him Taylor.
- [49:37]
Sarah Sluice
This episode was brought to you by Mfinex, pioneers in AI powered market mix modeling. Get fast answers to hard questions with Mutinex. You can ask for a demo@mutinex.co. that's m u t I n e x co Sam.