FirstMile Podcast: Dom Burch Talks MetaRouter, Retail Media, and the Next-Gen Data Layer

‍

On Episode 3 of The First Mile Podcast, Dom Burch sits down with our very own Greg Brunk, MetaRouter’s co-founder and Head of Product. If you’re curious about what really goes on behind the scenes in data infrastructure, server-side tracking, identity reconciliation, first-party data, this episode is definitely one to queue up.

MetaRouter started as a privacy-safe, server-side data layer, but it’s quickly evolved into something much bigger: an application-aware platform that helps brands, retailers, and CPGs do more with their data as cookies disappear and AI-driven commerce accelerates.

From Analytics to Infrastructure

Greg walks through how he moved from analytics and fintech into building software that solves real, day-to-day data problems. His background spans big data engineering, product leadership, and everything in between. The thread tying it all together? Making customer data usable, accurate, and secure.

Our mission from day one has been straightforward: help companies fully own and control their first-party data, reduce risk, and make sure every signal actually counts.

Tackling the Big Data Problems

Greg also breaks down the three major challenges we focused on early in MetaRouter’s journey:

• Accuracy: Traditional client-side tracking often misses 30–40% of data. With our server-side integrations, that gap drops to nearly zero.
• Performance: When you remove noisy client-side tags, pages load faster, SEO improves, and users get a smoother experience.
• Risk: Third-party tags can break pages or leak data. Our single-tenant, server-side architecture keeps data safe, controlled, and reliable.

In short: less chaos, more control.

Leveling Up With Application-Aware Infrastructure

MetaRouter isn’t just routing data anymore, we’re powering real-time application logic, reconciling identities across platforms, and connecting logged-in and anonymous users.

Our Sync Injector makes it possible to:

• Consolidate identities inside a fully first-party context
• Work seamlessly with both known and unknown users
• Deliver clean, ready-to-use data to multiple marketing and ad platforms
• Improve downstream results, often outperforming traditional client-side tracking

This is what makes MetaRouter the “first-mile” layer for next-generation applications, the bridge between infrastructure and activation.

Retail Media, for Real

Retail media has moved beyond reporting. It’s becoming a full operating system for ads, measurement, and activation.

With MetaRouter, brands and retailers can:

• Share only the right, consented data with partners
• Activate audiences in real time
• Reduce the friction for partners who want to work with them
• Boost ROAS and performance with instant audience updates

In other words: we make it faster, safer, and more efficient to act on your own data.

AI, Programmatic Shopping & the Visibility Challenge

Platforms like TikTok Shop and emerging AI shopping agents are reshaping how people discover and buy. The lines between content, identity, and purchase are blurring—and brands risk losing visibility into key behavioral signals if they rely solely on third-party systems.

That’s why owning your infrastructure is becoming essential.

Greg highlights three must-haves for companies preparing for this shift:

• First-party data control
• Real-time identity resolution
• Privacy-first architecture

The Takeaways

• Track once, track right. Accurate data is the foundation of everything.
• Server-side reduces risk. Better performance, reliability, and compliance come standard.
• Identity is king. Real-time identity powers real applications.
• Retail media is accelerating. Real-time activation and lower friction equal better outcomes.
• AI shopping is here. Protect your signals and your customer relationships.

The bottom line: infrastructure isn’t just plumbing anymore. It’s the engine powering modern marketing, retail media, and AI-driven commerce and we’re building it.

Read the full transcript :

‍
Dom Burch: Welcome back to the FirstMile podcast with me, Don Birch. Today I'm joined by someone who sits right at the fault line where infrastructure, identity, and the next generation of media collide. Greg Brunt is the co-founder and head of product at MetaRouter, a company that started life as the cleanest, most privacy safe data infrastructure layer on the market and has quietly evolved into something bigger, infrastructure that's becoming aware of applications. Now, the timing couldn't be more relevant. If you've been reading the news, TikTok shop is rewriting what commerce data looks like. Amazon is making a full-funnel AI-driven advertising push that blurs the lines between media, identity, and retail. And every publisher, retailer, CPG and B2B platform is waking up to the fact that old batch-based cookie-driven world, if it's not already gone, it's going. So today we're going to set the scene and discuss where infrastructure ends, where application level value begins, and why MetaRouter might be one of the very few companies positioned to bridge that gap. We'll talk about retail media, publisher, CPG identity, real-time prediction, and the emerging application aware layer that sits between data engineering and activation. Right, let's get into it. Greg, thanks for joining me.

Greg Brunk: Yeah, thanks for having me, man. That was quite the intro.

Dom Burch: Well, listen, let's start with your path. What did early building, your work as an analyst, fintech, startup engineering, what did all of that teach you that shape the way that you think about infrastructure today?

Greg Brunk: Yeah, it's a good question. I mean, I think, you know, obviously the underlying underpinning, and this doesn't sound particularly novel, is data. You know, it doesn't really matter what you're attempting to do, what you're attempting to build. Data has always been, in my opinion, the most powerful asset and has been the most sort of critical signal for providing value. And so, in maybe less generic terms, I think uh, you know, my path really started with uh a heavy emphasis on on years and years and years of spreadsheeting and trying to accomplish things for for various different groups in kind of financial spaces and and contract spaces and media spaces.

I had some cool opportunities to get some exposure to a lot of kind of the inner workings of uh media upfronts and and television buying and a lot of sort of advertising mechanics and CPMs that happened in that space when I was working in family-friendly film advocacy and and negotiation. Um, and then that parlayed really nicely into software. I had always kind of had a ton of interest in software, and I actually went to a boot camp very early in my career to sort of learn the basics of full stack engineering, sort of made that transition into customer data-driven data infrastructure startup worlds, and um, you know, got to to spend a lot of time on the engineering side, kind of understanding the mechanics of at the time what was very much sort of the big data world and how newer technologies like containerization and Docker and Kubernetes and languages that were emerging uh as kind of in the forefront of data processing Python and Go, and you know, just got some opportunity to work kind of in the technical underpinnings of very highly scaled data infrastructure and eventually kind of got pulled in through some sort of leadership orientation that's based more into product leadership, less of the IC work and more into the sort of general strategy of how to solve problems in the customer data space, which then parlayed nicely from our previous business into this one, where we uh were kind of founded under the thesis that a very, very large enterprise wanted to take significantly more control over the data that they were sharing with the various different media and advertising and marketing tools that were consuming it.

And they wanted to have an in-house, wholly owned and operated, very private, very secure, very highly scaled infrastructure for sharing all of that data out of their first party context, under their control, and more of a server-side model. And that was kind of the founding thesis of MetaRouter was how do you track everything exactly once and then give it to the 37 people that need it in ways that you can uh much more highly regulate than the sort of older model of cookies and tags and pixels. Yeah, that's kind of how we arrived here.

Dom Burch: I love that phrase. How to track everything exactly once. Like that's critical, right? So so that I guess that's what was broken then in that sort of customer data stack that you wanted to fix for that customer number one, right? What were some of the implications of the way that they were doing it previously? And we're talking nearly seven years ago, right?

Greg Brunk: Yeah, and server-side was less of a thing back then in terms of you know traditional advertising technologies, like think Google and Facebook and Trade Desk and Um Xander and Yahoo and these guys. Like there weren't many cappies back in 2017, 2018. You know, there were some early ideas, but mostly all the data that was collected for those platforms to operate, be able to do things like measure and attribute and build segments and build audiences and to run targeting and retargeting. The way all that was powered was primarily that you supplied your data, let's say as a retailer, you supplied your data via dropping their tag on the page. Um, and that allowed them to collect everything that they were doing within their platform and everything you were doing within your platform, and then measure the whole thing end to end, build audiences end to end, understand behavior end-to-end, and offer ultimately a better advertising or marketing platform accordingly. Uh, there were probably two major, three major problems that we tried to solve early on.

‍

The first was the exactly once element, which was that there's a ton of disparity that we were noticing between these platforms. So, like when we got into some of the early testing of a fully server-side integration with what I would consider like the big five advertising platforms is where we started. You had as much as like 30, 40% discrepancy, and it's something as that should be as exact as transactions. Like you had Facebook reporting, you know, 150,000 transactions, Google reporting 220,000 transactions, and Trade Desk reporting 90,000 transactions in the same time window. And they were all collecting data from the same source. There should be a relatively accurate source of truth for how many people bought a product that day. And there's a lot of reasons for that. There was a lot of noise, but a lot of that had to do with third-party tracking, you know, ad blockers, browser restrictions, inconsistency in those APIs, inconsistencies in those tags, what qualifies as a transaction. So we wanted to clean that up. And in those early tests, we took an average 30% discrepancy down to 0.05% discrepancy on that first test. And that was a big boon for us because as a CMO, just as a broad layer, having it reliability and accuracy as you're trying to analyze performance across these channels to have much more consistency in the basic conversion layer was really critical. The second thing was page performance. Walmart was actually a pioneer in this space way before we were doing our thing.

They were starting to analyze the impact of client-side tags and technology on sites directly related to revenue. So every hundred milliseconds, they released a report that said every hundred milliseconds of asynchronous latency on the page results in a 1% drop in online revenue. And so you had some of these tags, some of the bigger tags contributing to four or 500 milliseconds of asynchronous latency. And then you take 30, 40 of those tags, you have very slow sites, you have very bloated sites, you have very poor SEO performance, you have very poor uh load times, which means users bounce and they get irritated and they're gonna go somewhere else, right? So that was the second thing we wanted to solve, which is what if we got rid of those 26 tags, just left one in place, and did all the communication with those platforms from the server. So we saw meaningful SEO rankings and meaningful page performance and lighthouse indicators that the page was speeding way up. That was the second thing. And then the third thing was the risk. Um, actually, the founding thesis was that we had a very big customer in our previous business who came to us and said, Hey, we just had a third-party tag ship some malignant code to the site, took the site down for 24 hours, and they lost like $14 million in revenue overnight. And they were like, CIO, CTO level mandate, get rid of all third-party tags in 18 months. Can you help us do that?

And that's where we started to say, maybe instead of just being a traditional tag management solution that's just dealing all these third-party tags, what if we've just got rid of them altogether and figured out how to collect everything that was necessary from the page exactly once? Things like identities, things like behavioral data, things like consent preferences, everything that's necessary to help these platforms operate at their best, but to communicate to those 26 different vendors from the server instead. And obviously there's a 26x reduction in risk because there's 26 fewer libraries and et cetera, et cetera. And so that's kind of those are the three major areas we tried to solve.

Dom Burch: It's fair to say you've always been pretty purist about infrastructure. So it should be clean, deterministic, privacy first. What was it that you wanted to avoid becoming?

Greg Brunk: It's a good question. I think a couple of tenants we held very true were that we we always wanted single tenancy wherever possible. And we've actually always done that up to this point. There are certainly multi-tenant use cases that are very valid. You know, in a retail media context, for instance, you have a retailer and a brand, they want to share their data. Obviously, CPGs are generally purchasing quite a bit of targeting data. Retailers are generating quite a bit of targeting data. Those two have a shared value proposition in that they want to sell more crest toothpaste within Albertons, whatever it is.

So the there are certainly multi-tenant sort of shared data ecosystems that make a ton of sense. This is primarily clean rooms, but there's there's other applications. But one of the things we really wanted to focus on was kind of being that first mile for these businesses that says, at some point, if a multi-tenant ecosystem is one of your goals and you have reasons to do it, great. But before that moment, as you as a business are looking across all of your own media and even your paid media and even your in-store, all of that data should flow first into a completely controlled first-party private cloud single tenant environment. And that way, in that first mile, you can do all kinds of things to not only protect yourself, but protect your users. You can apply consent preferences there and guarantee that they're applied rather than going to 26 different ad tech platforms and be like, are you sure your tags are honoring CCPA and GDPR? Like you, you collect everything first, enforce your controls, own that, take on that liability and that responsibility because they're your users.

They're trusting you with their business and their behavioral data. They're coming to you. You want to retain them as a long-term customer. So you should be demonstrating a lot of maturity in the way that you process their data, right? And so that single tenancy was a big piece that we did not see in a lot of our peers. There were a lot of big multi-tenant ecosystems, and primarily those were built for scale, right? It's much more expensive to have a tenant for every single customer that you onboard. It's much more SaaS-like to kind of merge everything into a giant centrifuge. You get a lot of optimizations and potential margin improvements in that process. But we were really focused on that Fortune 500 space where that data ownership was really going to resonate with leadership. And they it was very expensive and difficult for them to figure out how to do that. And the SaaS alternatives were not providing that. So that single tenancy was really important. The other thing that I think we focused on was very much the ability to operate with a high degree of control in a way that was not limiting by making everything real time.

So essentially, instead of saying, look, we'll load everything into a warehouse and then we might go through and filter out sensitive parameters or eliminate certain categories that really shouldn't be used in retargeting and like take the whole children's toys section out because you can't retarget to kids. And like rather than sort of that, we said like you have to be able to enforce all these rules in real time because even though every use case doesn't require real time, you don't want to then bypass this infrastructure for real time and then have a bunch of risk in that bypassed infrastructure. Like this infrastructure needs to be able to scale and enforce all of your rules, all of your privacy, all of your single tenancy, everything you want to control as a business around your customers and their data in real time so that the maximum number of use cases are applicable to that methodology. And you don't end up with the scattered, bifurcated, you got this pipeline going to this vendor, you got this pipeline, most of it goes into this warehouse. This is like, you know, it was definitely still, but was very much back in 2018, much more of a scattered environment. So we focused really heavily on making sure that every value proposition we brought to the table from an infrastructure standpoint could operate at pure real-time SLAs, regardless of the scale, which was basically a lot of complex cloud engineering stuff and figuring out load balancers and pod scaling and um, you know, what was necessary in Kubernetes configuration to make sure that the whole infrastructure was capable of doing everything it needed to do at 50,000 events a second if necessary. Those are probably the two major infrastructure investments.

Dom Burch: You weren't just sort of routing data, right? You were then dealing with retail media, publisher identity. You mentioned CPG clean rooms, you've got this kind of cross-domain identity, logged out users. So when did you like realize you were drifting into this? I mean, I want to call it like an application-aware territory, right?

Greg Brunk: Yeah, I think right away. I think uh I think what we realized very quickly, and we we I'm glad we realized this very quickly because it allowed us to have some first mover advantages in this space. I think identity and therefore how identity was relevant to applications was something that we realized was going to be an immediate hurdle to this entire server-side value proposition, which was cookies and IDs live on the client. If you want to communicate from the server, there has to be some way to hoist that identity into the data feed so you can leverage it from the server. Because there's no point in sending conversion data to Trade Desk if you don't have Trade Desk IDs. The answer that everyone else was saying was just use PII from the server and use cookies and client-side tags for Anonymous. And that that works.

But if you want to have kind of these sacred owned first-party ecosystems, which I think a lot of bigger enterprises do, you have to have a complete answer for the full spectrum of your users available in the server side. And so when we started to look at how applications actually worked, like how click-through advertising worked versus view-through versus on-site versus in-store versus marketing and remarketing and how email was used and how various different onboarders would obfuscate that email for protection, but then reactualize that email and how hashing and all that stuff worked. You start to look at how each one of these vendor ecosystems, at the end of the day, they all offer a slightly different value proposition, but there's this core identity framework that they depend on. Otherwise, they can't do anything. What we realized was that what bigger enterprises really needed was a very modular, uh, very use case agnostic and highly customizable framework to collect identities that were relevant to their use cases. Not every identity ever. That would be way too cost prohibitive. And, but something very specific. I work with these two onboarders, these five advertising ecosystems, these two marketing ecosystems, this CDP, this warehouse, this on-site personalization engine, and these two ad servers. That's my group.

How do I build an identity reconciliation framework that gathers and centralizes all that identity into a first-party context from every single one of those, whatever that is, nine vendors, ten vendors, and build me a bespoke user-by-user understanding of that user that's fully anonymous to known? PII when available, great. PII when not available, what are all the IDs that these tags would otherwise be using? What are these IDs that these onboarders are using? How else are they identifying users when they're running their own third-party ecosystem? How do you move all that information into the first party? And that's where we built the sync injector and several other sort of supporting technologies around the sync injector, and then we patented it. And it's given us a really kind of cool differentiated position where most other solutions that look like customer data infrastructure are mostly in that dichotomy of PII from the server, anonymous on the client side. And we sort of built something that can sync with all these different ecosystems, pull that identity into a first-party context, store it as secure first-party cookies or index DB, and then hydrate it into the data feed at scale so that when you're going and then routing and picking apart a conversion event that needs to go to Trade Desk, you give trade desk IDs, Pinterest, you give Pinterest IDs, Adobe, you give Adobe IDs. And that's kind of how we made sure that we could bolster and minimize any performance loss. And in fact, like most of the time, we actually see a pretty significant gain in the performance of those downstream applications because we are operating in a true first-party context end-to-end within the walls of the business. So there's a lot less restriction from things like ad blockers and and and browsers and other things, because you know, again, this business is taking full autonomy and ownership over its own data and you know, collecting that in a first-party context. And that's much more compatible with the compliance frameworks that live within browsers and within the public sector.

Dom Burch: Let's flip into retail media, right? Because now, you know, that has become well, started as a sort of reporting category, right? Now it's activation, audience building, measurement, identity, like it's almost a mini operating system. How is that shaping or reshaping what MetaRouter needs to be?

Greg Brunk: Yeah, I think there's I think it's been really interesting to see. I think there's the older models that have existed for a very long time, even though retail media is kind of the new hotness, uh it's it's also like the idea of data collaboration between retailers and brands, or really any commerce platform and any brand that wants to sell a product, which definitely we've seen extend way beyond retail, right? Like there's there's travel and hospitality media networks, there's flight and airline media networks, there's rideshare media networks, there's tons now, right? It's not just retail. But um, as we started to see what was happening, kind of that old model was very much based on uh warehouses and clean rooms, which still very much have a place. But it was this idea that like, let's house everything and then let's pay somebody a bunch of money to do a bunch of uh, you know, mathematical obfuscation and encryption of the data in ways that allow us to not actually share our insights with the other party, but collaborate on the insights to get what we actually need, which is like an audience or uh a measurement look at performance or something like that.

And there's just a ton of layers in that share of requirements. There's like how do you get the data in, then how do you structure the data, then how do you structure the other side of the data to make sure that those schemas are compatible, and then how do you put in all the encryption, and then how do you share that into a shared clean room, and then how do you query that clean room and how do you pull back insights and how do you have auditing trails? And I think what we've started to see is that, like, there, especially on the CPG side of things and the supplier side, there's so many commerce media networks now that the tedium and overhead of collaborating within these networks at some point is gonna become prohibitive and they're gonna start consolidating like crazy. It's just way too much work to have 14 places to go to check for measurement. And if you're a PNG scale, you probably have 500 retail media networks you're working with, right? Maybe more. So, like, they can't have 500 measurement ecosystems.

There's nobody can keep track of that. Like, they can't have 500 different FTP ways to go pull an audience. They can't, it need you need to start to, if you want to be competitive on the commerce media side as a retailer, as a hotel, or as an airline, you have to start lowering the barriers to entry to participate in your retail media network and to start getting a share of those dollars while also differentiating yourself on value. You have to actually outperform the other retail media networks. The the hidden secret is that they always carry the big stick, right? They could just say, Well, listen, if you don't spend X amount of money in my retail media network, I'll just move your product from the top shelf to the bottom shelf. That will only go so far because now they have so many options where they'll say, Great, fine. Like outside of the top five retailers in the world, they're not gonna just bend over and play ball, right? Like they're at some point, they're gonna say, fine, I'll just reallocate because I'm getting better performance out of Amazon or whatever it is. And they'll just reallocate dollars and they'll just accept the loss because it's just too tedious and too expensive to just do something just because you have to.

That model doesn't scale. So, where we start to think there's a really cool opportunity is if you already have on the commerce side, if you've already made these investments in these true owned and operated, very flexible, first-party controlled environments, you can start to do a lot of the things that normally would require several vendors and several data stores and several obfuscation layers in real time to start to lower the barriers to entry and increase the compatibility of the data sets that you have that are valuable to those brands and make it easier and faster for them, not just from like a self-service standpoint where they have a portal they can go into, but literally the data arrives where it needs to be for them so they can just fold it into their master measurement mix or their existing advertising accounts. And as retailers and others start to reconcile what it means to share their data, they can't just pipe an unlimited feed to a CPG of everything that happens on their site. It doesn't make sense. But what they could do is if they can filter down the 10,000 events a second of what people are doing on their sites down to the 120 that are relevant to their brand, that are things like what categories those users are shopping, whether they're adding that brand to cart, whether they're purchasing that brand. So you can optimize look-alike models.

As you start to give them the optionality to filter and control and route with a high degree of precision in that first-party environment, it actually makes it a lot easier to open up this little tap that can go straight into PG's measurement framework or can go straight into PG's owned and operated advertising accounts. So now PG says, yes, now that you've made this compatible with my ecosystem, I'm happy to invest really heavily in your retail media network and put some dollars into getting this data because you can just fold it straight into what I'm already doing rather than require me to build a whole business and team around managing just this one retail media network and its weird infrastructure. And what we're actually seeing is that the performance is generally way better, right?

Like when you take an audience that gets shipped through a clean room once every three days, and then you have to kind of reconcile that and load that in, it's gonna perform worse than every single time a user adds your product to cart and purchases it on a retailer, goes straight into the advertising account within that owned and operated CPG advertising account. The optimization models are gonna happen in Facebook and Google and these other advertising platforms. It's gonna be faster and better. And you're gonna start to see ROAS start to climb. We've already seen literal quantitative results that are staggering around ROAS against real-time optimization in a retail media context versus traditional ETL-based audiencing. And it's a really cool opportunity to at the same time lower the barriers to entry and improve the performance of your offering, especially when it comes to off-site, on-site, digital. Obviously, the whole in-store retail media network thing is different. Email-based campaigns are slightly different. I'm mostly talking about like digital experiences here. But, you know, yeah, that's kind of the opportunity that we're seeing right now. We feel like we're pretty well positioned to kind of support those efforts.

Dom Burch: So talking about barriers to entry, right? Let's pivot to some of the news that's been coming out over the last couple of weeks. So we've got TikTok shop is now reportedly bigger than eBay, right? Which is, you know, you have to, every now and then in our tech, you just have to kind of put your foot on the ball and go, okay, that's quite amazing. Like, you know, eBay's been going 20 years, TikTok five or whatever, right? And it's just scaling it. And it's not just got to that point, it's growing at like, well, some reports saying like 125% or something. Yeah. Um now TikTok is interesting, right? Because they've sort of collapsing content, they've got comment identity and fulfillment all in their one loop. Does that kind of ecosystem change like what does that mean in terms of data infrastructure layers? Like what how do you read that? Like you go, okay, that's interesting, right? Um, Amazon talking like last week, like they're almost positioning themselves as being like a holding code for the whole of advertising, retail, marketing, like you can create ads, you know, like big shifts, right? And and and in those kind of walled gardens where if you're the brand, I need that audience, I want that sale, but what am I giving up, right? You know, I mean, just I just jumped off a call with one of your colleagues, and we're talking about, you know, maybe we should have paid for Facebook 15 years ago, but we didn't, right? But you know, the fact that we didn't means we've got the Facebook we've got today, right?

Greg Brunk: But Yeah, right.

Dom Burch: When you when you see something like that TikTok announcement, what what what goes through your head?

Greg Brunk: It's a good question. I think uh to get sort of myopic, especially when you're talking about like bigger dynamics like e-commerce and retail, to start to pattern match against Walmart and Amazon. And what you start, what you're I think we're starting to realize with like Instagram, Facebook, and Meta Shopping and Shopify and TikTok and some others is that the collective scale of a lot of smaller organizations, but operating on a common model and operating through a common infrastructure is actually like very competitive from a from a revenue reach, distribution, value and opportunity standpoint to some of those bigger retailers. And so, like the sort of unity of thought that you get from a Walmart that it applies across all of its digital touch points, efforts, products, customers, and sales is not going to be there in the software, more the software-driven side of some of the programmatic, like, you know, small shop uh aggregated, full marketing suite of services, full, very simple to go in and say, this is my shop. I sell these products, I want to target these users on Instagram and the flow just sort of takes care of itself. You know, it's probably a fault of mine, honestly, that I tend to myopically model off of more of an of a traditional e-commerce platform. I think part of the reason for that is because those are very, those ecosystems are very black box, right? Like they they want to own the share of requirements end to end. And so if you want to target the value that they bring to the table is the population of users and the eyeballs. And so the reason you would build a whole shop around a Facebook marketplace or a whole shop around a uh TikTok shop is because the eyeballs are built in, right? And so because they have that value and that's the fastest start, especially as like a smaller brand, they get to control almost everything.

And I think that that's like a necessary trade-off as a smaller brand trying to get off the ground that doesn't have the autonomy and power and negotiating weight that like a Walmart does. And so, in some ways, I see that and I go, it's a great opportunity for smaller brands, but there's obviously a lot of risk because essentially you're trusting TikTok and Meta to act responsibly with the hundreds of thousands, millions of shops, and then the hundreds of millions of users that are going to those shops, aggregating and centrifuging all of that data into their back end, and then basically providing at the end of the day performance with very little visibility. You don't have a ton of visibility into exactly who's being targeted and why they're being targeted, whether their consent preferences are being honored. And so I think, you know, at the end of the day, it's a new model, but it's a model that's very big tech driven as opposed to traditional big box driven. And I don't know necessarily what that means for people, but I think probably not great things because at the end of the day, there's this one standard deviation removed now, which is Walmart and these bigger retailers know that in order to maintain a long-term relationship with you, there has to be trust. And so as they're processing your data, they're leveraging your data, they're doing advertising, at some point, a gross violation of privacy is bad.

Doesn't mean it doesn't happen, it happens all the time. But at some point, there's at least a natural pre-programmed desire to respect you as a consumer because they want something out of you. They want you to come back and buy more stuff. And so there's some amount of healthy pressure there. When the people processing all the data and making all the decisions around the data and doing all the advertising and doing all the targeting and doing all the promotion actually are one standard deviation away. There's a standard deviation into this big tech monolith behind the scenes that's taking your data along with your data across hundreds of thousands of other sites and getting great performance. I mean, like there are kings that are being made every day in these ecosystems and brands that are growing tremendously. But there's they don't like Facebook doesn't have to worry about your relationship to that brand. There's no incentive for them to operate super responsibly with that data. It doesn't mean that. They're not going to. They obviously have government oversight and other things.

But like at the end of the day, that to me is the concerning part of when you sign away the control of all the targeting and marketing and advertising models to kind of a closed loop black box ecosystem that is one step away from the actual user trust. You know, I think that's where you see a lot of uh weird stuff. And I think if you've followed along over the last 10 years and litigation and antitrust and privacy violations and weird data sharing agreements between parties that shouldn't have the data and sort of like operating out in the open doing really sketchy stuff, it usually comes more from that big tech side of things than it does from the from the businesses themselves that are actually selling the products. And I think it's because that separation. So like I don't have any answer necessarily, but obviously that's a concerning trend that I'm not 100% sure that Congresses have the technical savvy to keep up with and would even really know how to regulate properly. So they kind of operate in broad platitudes and uh it's usually not sufficient. Stuff sleeps through the cracks all the time.

Dom Burch: Half hours just whizzed by, right? But it's been which just basically means we have to do more of these because we're only scratching the surface. But what's the capability you think everyone's gonna need 18 months from now that most leaders out there haven't realized yet?

Greg Brunk: It depends on who you're talking about, but I would say if you're a retailer of any presence or a major commerce platform of any presence, you're gonna have to have a dedicated answer with which to deal with the perplexity, Chat GPT, open AI type of uh off-site shopping behavior. We're already seeing what those businesses are doing in terms of crawling sites and gathering information, deploying their own browsers, kind of having unlimited ability. What we're just we're gonna start to see is a shift away from again, a shift away where if you are using perplexity and and they're saying, hey, this product's available at at Kroger for this price or whatever, and a lot of that funnel of getting exposed to those products and even adding those products to cart is starting to be handled off-site by, again, a party who isn't necessarily in a direct commercial relationship with you. The the bigger organizations, the retailers and such, are going to need to develop a methodology with which to control that process. Because A, if they lose visibility into most of their funnel and they turn at the end of the day just to a cart and checkout experience, they lose a meaningful ability to add, continue to add value to their customers, to continue to grow value of their business. So there's like a financial and positioning and control risk.

But I think there's also you're shifting more into those dynamics of it's not going to be great for users either, because there's nothing when you're able to, as a programmatic shopping engine, like eventually ChatGPT will have, and Perplexity already has them. There's a ton of these now. If they're able to go collect the data with bots about everything people are doing everywhere on the internet, they're not beholden to anyone with what they do with that data. They collected it organically. They didn't have to sh sign a data sharing agreement and a bunch of legal documentation to get access to that data. And therefore, there's rules in place for how they can operate with it. They can do whatever they want with it. And the thing that they'll naturally do in a capitalist world is try to offer the most value with the data. And that usually looks like, well, great, what can I learn across these sites? How can I commingle this stuff? And how can I build even better targeting? And before you know it, your data as an individual is starting to get sent around in a largely unregulated space to tons and tons and tons of targeting models and exposure, how the PII is handled, how the consent preferences are handled, none of that stuff has been figured out anywhere. And if you go to some of these bigger retailer sites, all that stuff has been figured out.

They have to follow CCPA guidelines and all these different things. But like if all the data is being scraped and collected organically and you're not in control of the situation, it's a risk for users and it's obviously a risk for the position of these bigger businesses. So what I would say people need is a formal plan and a formal set of technologies that allow them to still control the data. That's the most important thing. You need to be figuring out ways to identify and prevent programmatic shopping from accessing an unlimited feed of everything that's happening on your site. That's really important. The second thing is you need to have a really explicit direction from an executive layer on the negotiating position of when and how you share data, how it gets used, and what you expect in return. Right. And so if you're head of online at uh some of these bigger uh, you know, commerce or retail or travel or whatever it is, platforms, one of the things you want to do is actually go push to have a negotiating position with these guys who are gonna be showing your products on their chat widget and say, I expect to get full exposure to all that data. I'm not gonna lose access to my full funnel. And I'm also gonna block all of your bots from coming to my site. And you need to figure out technically how to do that, how to identify this type of crawling, block them from scanning your entire inventory and uploading it to their platform for free and all these different things, and say, until I get full exposure in my own control for everything that's happening with my products on your ecosystem, I'm not going to share any of my data around conversion, which was the only way that these places are gonna, or these these programmatic uh AI agents are gonna be able to optimize their models.

They're not gonna get better at showing the right products until they know if a conversion happens. So they're gonna start coming to everybody and say, hey, I need all your conversion data. And you need to say, no, I'm not gonna give it to you until you give me all the top-of-funnel data. But on top of that, you need to advocate for consent and privacy. You need to understand the frameworks that those platforms are using, you need to understand your competitive risk. Uh, put in the DPAs to make sure that when you share that data, that doesn't go into a retargeting pool that allows your competitors to undercut your prices within these things. Like, you need to advocate for competitive obfuscation, you need to advocate for privacy obfuscation, how is PII being used in these models? What's being stored? How is it being stored, et cetera? All this stuff, like if you can't gatekeep the information technically as step one, then you have no negotiating power. But if you can, then you have the negotiating power to say, I'm actually going to embrace this programmatic Chat GPT-based shopping or perplexity-based shopping or whatever it is, but I'm going to do it with my rules and my controls. And that goes back to something like a meta router where if you don't have an owned and operated platform that can allow you to filter this stuff down to only what you want to share and to have full first party control of all of your data before you then integrate it with perplexity, then they're going to come to you with their tag.

And your only option to not lose out to your competitors who are making a killing on these platforms is to throw the tag on the page. And at that point, you've given up all control. You've just allowed them to have unlimited, unferreted access to everything that's happening on your site, and you have no negotiating position. So you have to have the data control and the protection of your ecosystem as a first step. And then you can go in really strongly and say, I do have everything you need and I can make it available to you tomorrow, but these are my lists of demands. And that's, I think, what most of these commerce platforms and anybody that basically ultimately sells anything online, whether it's travel or trips or ride share or whatever, you need to be able to own the data and protect the data so that then you can negotiate with this changing new modality of shopping and exposure to optimize it to the best end of both your users and your business. And if you're not already thinking about that or actively thinking about that, then you're going to be behind because this. I know people overstate how many people are shopping via Perplexity right now, but it will happen. It will change, it will become more convenient. Those models will get better. And eventually your behavior data on your own media is going to start dropping. And you need to already be in a strong position to just shift your value into those ecosystems and be one of the best suppliers on those ecosystems. That's a really important investment.

Dom Burch: Certainly is. Brilliant. Greg, been an absolute pleasure. Thank you so much for coming on to the first mile.

Greg Brunk: Of course. Thanks for having me, man. Appreciate it.

Dom Burch: So there we have it. We started out with the fundamentals, the mechanics of data, tracking events exactly once, and why accuracy isn't a nice to have, but the foundation every retailer, publisher, and platform is standing on. We talked about the shift away from bloated, fragile client-side ecosystems towards real-time, server-side, single tenant infrastructure. The kind of work that quietly reduces risk by 20 plus times and gives businesses page speed and measurement performance that they can actually trust. And we explored where things are heading. Identity reconciliation, application-aware infrastructure, logged out user journeys, and the new demands created by players like TikTok Shop and Amazon's full funnel advertising push. What's clear is that data infrastructure isn't just plumbing anymore, it's becoming the operating layer that powers activation, measurement, and everything that comes next. So thank you for joining us. We'll be picking up some of the themes from this podcast again soon and going even deeper because we're only just scratching the surface of the first mile.

‍

‍