Loading summary
A
This podcast is brought to you by Rise in Programmatic. Most bid requests don't tell the full story. Around 60% have no identity signal. Another 30% are mislabeled or too generic. Buyers end up bidding in the dark. Rise changes that by introducing an agentic layer built directly inside the bidstream. Analyzing every bid request in real time. Rise reasons across context, environment and behavior to generate clear intent signals, even in anonymous and cookie less environments. The result? Better targeting, less waste and up to 3x higher CPMs on enriched inventory. Not just more data. Better decisions in milliseconds at scale. Find out more about rise@rise codes.com. that's rise codes.com. Hi, this is Ari. We're happy to bring you another recording from our Architecture Live event. This is with friend of the pod, Keith Petrie, who is the SVP of Data Identity and apply for Vayant. And he's got a punchy title. It's called Deterministic. Prove it. Hope you enjoy this.
B
Thank you everyone for being here. Thank you for being at marketecture Live. My name is Sam Corey.
C
I am the Chief Strategy Officer of Market Share. With me today is Keith Petrie from Vyant and we are going to talk about how deterministic and how you need to prove it. So welcome.
B
I need to prove it? I thought we were talking about the industry needing to prove it.
C
We all need to prove it together. I have a quick question for just the crowd. If you work in Connected Television, can you raise your hands? Okay, so a lot of people work in Connected Television. So we're going to talk today about a couple of things, mostly talking about identity, deterministic, what that means for the industry and where the challenges lay. One of the biggest things that we've seen over the last five, six, seven years is just this massive growth. And the growth at Connected Television came with some things that we thought were solved, right? Lots of login data, all the various data points that are coming through and what that means for measurement, attribution, etc. But there's been a lot of challenges there. And the data underneath that expectation of login user data and what that means for the industry was not exactly how we thought it would be. And so Keith, my question to you is, why is it a little messy underneath the layer that we thought would address all of our concerns the last seven, 10 years?
B
We thought we were going to do it right this time is what you're saying.
C
We thought we were going to fix it.
B
Yeah. All right, so just like ad tech, Sam and I kind of put the cart before the horse here. And we came out with a topic that brought you all here today. A few bullets and glad to see there's tons of support and interest in this space. But what we didn't do is we didn't put any meat on the bone. We didn't actually plant any substance here until Sunday morning when Sam called me all panicked driving in his car all other. All over SoCal. And he was going from Target to Target trying to find. I don't know what it is.
C
What.
B
Mackenzie Childs.
C
Yes, Mackenzie Childs did a collab with Target and anyone nodding their heads. I'm married with two girls, and I was sent on a trip to go find it. So five stores, 35 minutes away from my house is where I ended up. But I got it. So you're welcome.
B
And that's the brand story that we're here to talk about, which is basically that Sam's wife, she's going to play a pivotal role today. So just for character's sake, her name is Mary. All right? And we're going to send her some mackenzie Child stuff after this. She saw a CTV ad and she was influenced by it, and she was in the right moment. She was definitely the household decision maker from what I've learned. Confirmed. And Sam was trying to enjoy either a lovely morning cup of coffee before his two daughters woke up and started dancing or getting a bunch of messages from the east coast market team. And he gets a text saying, go. And he gets in the car and he goes and he completes that purchase at his fifth target. And now he has whatever scrunchies at home. And all three of his women are very, very happy. And that, like, I'm looking in the room and there's a lot of people here. Whether you're an ad platform, you are an agency, you are a brand. That is the story that we all want to tell, right? That is what we are all looking to achieve, and that is what performance CTV is promising. And unfortunately, that, that, that is not the case of what we are seeing today. And so you have these publishers, and I don't know about you, but I pay a ton of subscription costs and have my email address or at least somebody in my family. And basically that is not being portrayed to the buy side. And so through whatever number of intermediaries or hops, that publisher has built that trust with that audience, and they completely and rightfully so own that relationship with the audience, and they don't want to publicize it so that everybody, a lot of us in this room can see that, learn from that, you know, log it and go from there. And so they keep that within their own walls. And that means that the buy side is left working with what we got. And what we got at the end of the day, which we'll get into, is an IP address. And if you look at recent research from sim, I think it was just a few months ago they came out and said 13% of IP addresses can be matched accurately, I believe was the word, or precisely, to an actual household. And, you know, we're sitting here and we've lived through the web, we've lived through mobile, we've lived through all these different times of innovation, and we're kind of stuck in the same place. And so when we say deterministic, prove it, what we're really talking about is we need to come to the table, we need to collaborate, we need to be able to have that communication not in public. And that is where Viant very much stresses our background in terms of tying all these different identifiers. Multiple emails that belong to a single person, multiple people that belong in the same household, some people who do the bidding, some people who tell you what to buy, all in one household with a physical address.
C
You ran that campaign through Viant?
B
Yeah. So this, this is where. And that was the coincidence, like a ridiculous sales pitch. But Mackenzie Child ran that campaign as an autonomous outcome campaign that we announced at ces.
C
So I got my scrunchies. Well, my kids got their scrunchies. So where does it actually break down, though? So we were talking a lot about the login data, what that means, what signals are being passed along. Is it a technology problem? Is it a supply chain problem? Is it just an overall challenge that we have in the industry that something breaks in the process? Can you just kind of dig in on that and what that means?
B
Well, we all will continue to have jobs because everything will continue to love
C
job security in my market.
B
Yeah, not sure about market tech. Sure. Ari is second row.
C
He's already replacing me.
B
So it's all of the above, which is a cop out. So we're just going to hone in on one specific topic today, which is really that the programmatic paths and the intermediaries between that publisher and the buying platform, whatever that buying platform looks like, there's way too many hops, and each of those hops represents and has incentive to arbitrage the opportunity. And the best way to do that is to maybe manipulate the user agent, the device identifier, maybe Add an extended field identifier, change the IP address, which might be a more in demand IP address that's in a third party segment, that's popular with buyers, and all of that will increase bid density, the ECBM will increase, and that is what each of those intermediaries is incentivized to do. And so that's why you see a lot of the industry focused on supply path optimization. And while SPO does result in cost savings and us as a fiduciary of our advertisers budgets, being able to spend it more effectively and have larger reach and less waste, that is actually a byproduct of spo. The main purpose of spo, especially at Vyant, is actually just to get quality signals direct from the publisher with no intermediaries, with no incentive to manipulate them.
C
So I mean, we could probably go on a whole other tangent about misrepresentation of inventory, obfuscation of data points for their benefit. It's like a whole seven hour session we can go into. But when we do get closer to the publisher, when that data is more accurate, when we start to optimize our supply path, what's the outcome, what's the benefit to marketers and buyers? And what does that look like?
B
Yeah, so getting closer between the buy side platform or the buyer in any case, and the publisher is certainly important, that can clean up any of the misrepresentation. But you do still actually have an identity problem, because whatever that mechanism is, you still need to know that the advertiser has a certain audience, whether it's first party data that they've built up themselves or they license it from another party. They have an identity that they want to target and the publisher has their subscriber information which they want to protect rightfully so. And so it really comes down to being able to communicate that effectively. And there's a number of different strategies that can do that. But the way that we've defaulted to over the years is that one of the intermediaries will license a graph. And you've seen all of these graph companies, which I have some bias with over the years, focusing historically on like hey, let's integrate with a dsp. Then their market cap is capped, their TAM is only so big, so they go and they integrate with the supply side. Then you see publishers large enough publishers starting to license them and even agencies now bringing them in house as well. And so you have all of these multiple hops. But what you really just need is a direct integration between the publisher and the advertiser buying platform so that that communication can be one to one and resolved at the household level.
C
We love attention metrics. We talk about contextual. Pretty hot buzzwords, pretty hot topics across the industry overall. We have all heard about the power of contextual, what that means for advertisers. How does that play a role in identity? How does that play a role in what's being made available to marketers today? Or does it not?
B
Yeah, I mean, so I'm identity through and through. I've known many people here for, you know, 15 some odd years. So identity number one, we're not going to budge from that. But it is certainly not sufficient to only look at the industry from, from an identity perspective. And so I hate bringing it up. I hope there's no represent representative here from Metta, but I need to give Metta a little, little shout out here because you need to have the identity, you need to have the who, but you also need to have the who in the right context at the right time. And that's what matter is extremely good at. They will take a micro moment of you pausing for one millisecond longer on one piece of content and they will hammer home that niche product that right now it's me shopping for possible and can and certain shoes that I need to buy. And it's doing an incredibly good job. And we need to be able to bring that to programmatic, whether that is CTV at the top of the funnel or lower down. And that's where context really shines and has an opportunity. But it's also one of the biggest places where we as an industry are dropping the ball. Because right now you might be facilitating a buy on CTV and you might say, hey, I want to Target Landman and I want to make sure that I hit marry at the right time so that she is interested in buying this. I keep thinking it's tableware, so I'm going to go with tableware. Yeah. So either way, I know it's scrunchies, but still, basically I keep thinking, and in Landman you have the wife. I'm horrible with character names, but you have the wife who does these incredible dinners and then she goes all out and like sets up like a theme night of Pirates. And the whole dinner it looks like a $40,000 setup. So either way, after that it might be a good moment to interact and influence Mary to buy Mackenzie Childs at Target. However, when you're buying Landman, you don't actually know that on most CTV inventory you're actually just buying Paramount and you might just be buying Disney and you might just be buying NBC. You don't know anything else. You don't know the content that is going on and the actual potential influence that you can have on Mary. We're going to sell Mary a lot of stuff by the end of the day. So the point being you're just buying something that could have a kid show or it could be a UFC fight, and you don't know the difference. And so you do need to know the context. And that is where Viant very much relies on Iris, which was acquired about 15 months ago, where we have the context ID, where just like we do a direct match with the publisher for identity, we also do a direct match with the publisher into their cms, their content management system, so that we can convey not just who it is, but what that person might be receptive to and then text their husband about the next morning.
C
In a perfect world, contextual login data identity is locked in. We're good. There's no hops, supply path optimization. We're not obfuscating.
B
Loving this. Yeah, go.
C
This is like the dream of, like, how three quarters of us will be out of work, but everything is perfect right across the board. What does that mean for a marketer? Like, what's the unlock? What's the unlock in value? What's the unlock in measurement? What's the unlock overall?
B
I actually didn't even plan this. I'm like reading my notes. This is no notes.
C
I went off script.
B
You're going, but prove it. Like, what the marketer gets out of this.
C
Yeah.
B
Is proof.
C
And this is great.
B
So basically, like, what it comes down to is like, we're all looking for attribution, we're all looking for conversion, we're all looking for approving the return on ad spend. But at the end of the day, what our industry has gotten really, really good at is actually just demand capture. And it's taking credit for that versus incrementality, which is an actual attribution event that would not have occurred without this impression on Mary. And we want to be able to do that. And the only way to do that well is to have a singular understanding of identity across the entire system. And that is from audience segmentation and targeting, frequency capping, optimization, and subsequently measurement so that we can understand that entire lifecycle. And the only way to do that is, let's see, in this scenario, we have Mary, who's authenticated into whatever hgtv, horrible at publishers on that side, creating
C
a segment right now.
B
Yeah, exactly. We're creating a segment. I'm horrible at it. So basically you have Mary, who's logged in. Mary has way more than one email address. We need to resolve that she has way more than one device. You might be watching this on the Paramount app on a Roku device that's plugged into a Samsung TV and conveyed through any number of intermediaries. You need to have that, all of that resolved plus tied into @gmail SAM architecture. And in order to do that, you have to have a strong background in understanding all these different signals deterministically or probabilistically, clustering them with good logic behind it. And then from there, and only from there, can you actually prove it in terms of an incremental uplift for that brand.
C
That was a great explanation, by the
B
way, your email address.
C
I think you nailed it on that one. I guess my final question, we have marketers, we have buyers, we have ad tech companies, we have publishers in the space. What's, I guess like the final words you would give them advice you should give them when buying inventory and what does that mean for them and what does that look like for Viant?
B
Yeah, so I kind of just touched on it. I don't want to sound repetitive, so I'm going to try to use a different picture or paint a bit different picture. But when you go to a buyer and that buyer is leveraging a third party data segment and they are buying across any number of different supply paths and they have no direct negotiation or contractual relationship with the product publisher, they've licensed one or more graphs, they, they are a hodgepodge of different ad tech infrastructure, some of which ends up, you know, underneath the same roof. And you know, you're trying to collaborate with different teams who have different incentives. You cannot guarantee that you are acting as the best fiduciary for that advertiser's budget. And looking at it from a holistic standpoint of we are going to work with you from the moment you are collecting your audience data, building your segmentation in different storylines for people who have churned in the past. Repeat buyers like you need to be ingrained in that advertiser's business. And in order to do that, you have to have relationships across the board. And if any one of those pieces breaks, then all of them break, unfortunately. And that's the hardest part, I think, of all of our jobs.
C
I think that brings us to the end. Cool. Thank you. Thank you, everyone.
B
Thank you.
Host: Sam Corey (Chief Strategy Officer, Market Share)
Guest: Keith Petrie (SVP of Data Identity, Viant)
Event: Marketecture Live
Date: June 1, 2026
This episode, titled "Deterministic? Prove It.," dives deeply into the challenges and realities of identity resolution and data determinism in the evolving world of Connected Television (CTV) advertising. Host Sam Corey and guest Keith Petrie discuss the gap between the industry’s expectations for deterministic, person-level data and the messy, often uncertain nature of the data actually available to marketers and platforms. The conversation highlights why these challenges persist, what’s at stake for marketers, and how direct integration and transparent supply paths unlock higher value and true attribution.
"There’s way too many hops, and each of those hops represents and has incentive to arbitrage the opportunity...change the IP address, which might be a more in demand IP address...all of that will increase bid density." – Keith Petrie (07:02)
“You need to have the who...but you also need to have the who in the right context at the right time. And that’s what Meta is extremely good at.” – Keith Petrie (10:18)
“If any one of those pieces breaks, then all of them break, unfortunately. And that’s the hardest part, I think, of all of our jobs.” (16:41)
This episode provides a frank, nuanced look at the ever-present challenges in digital advertising identity and attribution—with a humorous, real-world example of how things break down even when everything should work. Both Sam and Keith hammer home that deterministic claims are hard to prove in practice unless the ecosystem moves toward radically more transparent direct integrations, less intermediary manipulation, and richer context signals. Marketers, platforms, and publishers must all collaborate more closely to achieve the “proof” of value the industry demands.