Who Replaces Switzerland? Mathieu Roche on Rebuilding the Identity Commons

Watch on Youtube

Listen on Spotify

For years, LiveRamp played the role of “Switzerland” in digital advertising: a neutral identity layer connecting brands, agencies, publishers, platforms, data providers, and measurement partners. But with Publicis moving to acquire LiveRamp, can infrastructure owned by one of the world’s largest agency groups still credibly claim neutrality?

In this episode of Signal & Noise, Rio Longacre and Brett House sit down with Mathieu Roche, Co-founder and CEO of ID5, to explore what happens when one of advertising’s most important shared rails suddenly has an owner with a very different strategic position.

Mathieu argues that the industry now has a genuine reason to build alternatives—but replacing RampID is not the same thing as replacing LiveRamp. LiveRamp’s position was built through identity resolution, onboarding, publisher integrations, data collaboration, activation, measurement, and years of network effects. Recreating that requires more than another identifier. It may require an entirely new coalition of independent companies.

The conversation gets into what that new identity commons could actually look like. Mathieu explains why identity should function more like a common currency or language between buyers and sellers—open, interoperable, and usable across platforms rather than controlled by either side. He also explains why identity is just as critical for measurement as targeting: if advertisers cannot connect exposure to outcomes across devices and environments, they cannot prove performance or optimize investment.

Rio, Brett, and Mathieu also dig into ID5’s acquisition of TrueData, which Mathieu describes as a kind of “Rosetta Stone” connecting different identifiers, devices, users, households, and offline data into the ID5 ecosystem. The discussion explores deterministic versus probabilistic identity, why conventional “match rate” metrics can be misleading, and why the real question is whether an identifier can actually be found and activated against in the wild.

They also examine the publisher side of the equation: how identity can increase bid density and monetization, how ID5 works alongside companies such as Optable and Permutive, and why publishers need deeper knowledge of their audiences if they want to regain leverage from the buy side.

Finally, the conversation turns to agentic trading, CTV, and the future of advertising infrastructure. Mathieu believes agentic buying could compress parts of today’s programmatic supply chain—but rather than making identity less important, increasingly granular and automated buying may make a trusted identity layer even more essential.

If LiveRamp can no longer be Switzerland, the next Switzerland may not be a single company at all. It may be a modular network of specialists—and the race to build it has already begun.

#AdTech #DigitalIdentity #LiveRamp #ID5 #ProgrammaticAdvertising #AgenticAI #Advertising #DataPrivacy #CTV #Publishers #MarketingTechnology #SignalAndNoise

Read the full transcript below.

---

Brett (00:01.104)

Hey everybody, welcome back to Signal and Noise. Brett House joined by my co-host Rio Longacre. Today's guest is Matthew Roche, a new friend of the pod, because we've met with you twice. We've had two of these, right? And certainly a fun guy for people that have known him in the industry. He's been around. he's the co-founder and CEO of ID5.io. and he spent a decade arguing.

Rio Longacre (00:06.926)

Hello everyone.

Rio Longacre (00:14.094)

Here's number two.

Brett (00:28.66)

really that advertising needs a neutral identity layer that nobody should own outright. And for those that don't know ID five it's an identity company, an identity graph company. and you've had a lot of experience in this space. I mean you built you built the case with at WebArama when you which you co founded before I'm sorry, before co founding ID five in London, right? In two thousand seventeen.

And I and I'm not gonna say your co-founder's name because I'm gonna totally screw that up, right? Pierre Antoine Du Pier. PA yeah, du how do you pronounce that last name? Durguet? Durguet? Duge Jesus, my French pronunciation is terrible. Yeah, I apologize, I'm much better in Spanish, right? you know, so you so yeah

Mathieu Roche (01:01.338)

PA, PA for short, yes. P A.

Mathieu Roche (01:07.684)

George. Yeah. We're gonna we're gonna turn this into a French lesson.

Rio Longacre (01:15.918)

That's what he keeps saying, but we have no proof that's actually true.

Brett (01:18.672)

I lived in Spain for a year, right? yeah, so so yeah, yeah, so you started at Weborama, you co you know, you co founded I D five, right? And and the the goal was to build a shared consented identifier, not another walled garden.

you know, for this for this sort of cookie-less, very privacy specific world that we lived in, which is which has changed, you know, since you since you started. I think I met you at Cannes a couple of years back with Steve Silvers and some of my friends at the Trans Union new surf side of the things.

So yeah, so I mean you kind of made a contrarian bet, right? LiveRamp, you know, when they before they were purchased by Publicis for those that didn't see the news, you know, they had the trust in the ecosystem, they had a pseudo monopoly on identity resolution, right? Twenty five thousand publisher relationships. And then of in May of this year, and I think it's kind of finally closing or getting near closing, Publicist bought LiveRamp for two point two billion dollars for those that missed that news. and and s

Rio Longacre (02:22.423)

We cover it here on the pod. That is one of our top episodes.

Brett (02:23.852)

Yeah, and suddenly and suddenly yeah, suddenly I D five is is you know, got a a a a much bigger play. So you know, we're super interested to hear how you guys are really taking advantage of of what's happened there. and there's been a lot of commentary. We did a podcast on that. And and you guys just did a twenty million dollar series B, right? Was that the end of last year or was that in twenty twenty four?

Mathieu Roche (02:47.886)

That was two years ago. Yeah.

Brett (02:49.006)

two years ago, so it wasn't it wasn't that. And that was with TransUnion where where I came from, with S4S Ventures as well, right? and then you acquired TrueData, right, fairly recently, which was another identity graph company. So, you know, the combined footprint of ID five, from what I know, the most recent data I've read is about one point five billion people across six hundred and sixty-five million households. So very broad reach, also a lot of international reach.

which is different than I think some of the the Yeah. Yeah, and that's often a a weakness of of like traditional identity companies. Is like once you go overseas in Europe or AMEA or APAC it gets it gets weak. And you live in New York City? So how long have you been how long have you been in the city? I just in the are Yeah?

Mathieu Roche (03:18.532)

Yeah, o over over than half of that is outside the US, which is yeah, indeed very different.

Mathieu Roche (03:34.094)

I do. just a year and a half now. So still still a still a new still a new New Yorker. But I came I came at the right time with the Knicks winning the winning the you know, the championship as well. So that's it's worked out fine.

Rio Longacre (03:39.039)

Nice.

Brett (03:44.898)

Yeah, that must have been interesting.

Rio Longacre (03:45.005)

Yeah, native, native, like people group in New York, we were waiting our whole lives for that, Matt. So yeah, that was, it was, it was welcome.

Mathieu Roche (03:49.616)

I'm I'm spoiled. I'm spoiled.

Brett (03:53.177)

Yeah, yeah, and even even coming from from New England and being a Red Sox fan, I I must admit I jumped on the NYX bandwagon 'cause I've lived in New York for twenty six years and it was hard not to realize how important it was to the city. Right? It it it was I don't know if you see anything like that in Europe.

Rio Longacre (04:08.19)

Maybe next year, maybe next year the Jets. I'm kidding. The Jets are never going to win. suck. But that's the spicy take.

Mathieu Roche (04:12.514)

Ooh ooh, that's that's that's the the odds on that. I like the odds on that.

Brett (04:16.994)

Yeah, yeah. So so before we get into the thesis, well what did I did I capture everything about your background, your trajectory th from R Webarama through I D five and and and and anything else you want to tell the audience about yourself and and kind of what brought you to this point in your career?

Mathieu Roche (04:32.974)

No, I think I think you've captured it, right? I was I was at at a data company called WebRama for fifteen years and saw firsthand the the need for I mean as a data company, right? Like you you live and breathe by your ability to recognize people, right? And that's how you monetize your your data assets. And so recognizing people for a long time was cookies, right? And and and and with the the the the beginning of signal deprecation, Apple starting to block cookies in Safari and

Brett (05:02.062)

Yeah.

Mathieu Roche (05:02.724)

Higher kind of regulatory bar with you know GDPR becoming low in Europe. we, my my co-founder P and I saw an opportunity to kind of rethink how we identify people for advertising purposes. And and so it was very contrary to the time because it wasn't fancy, like nobody was nobody cared really in 2017, 2016, 2017, when we started kind of playing with that idea. And kind of I reached out to a few of my friends in industry and say, like, hey, you know.

Cookie matching is a real pain. What do you think? Or you think we sh should we could make it better? And what about like, you know, cookie deprecation insofar? Is that a problem? Everybody was like, don't care. Doesn't matter, right? All of that, all of that was built back in the you know mid 2000, right? The the engineers who've built the cookie match tables and the matching pixels and all of that are gone. and so like it's just like it just runs, right? Nobody cares, doesn't matter, right? And for years it was a lot of preaching in the desert because I knew as a user that this would be fundamental. I mean

Brett (05:39.128)

Well, yeah.

Brett (05:48.581)

Yep.

Mathieu Roche (05:59.735)

Advertising without addressability is is is useless, right? The promise of of digital advertising is that you can Yeah, exactly. Like you if you if you recognize your customers, you recognize your prospect, you know who they are, you know how to address them, you frequency cap your delivery, which means that you know the devs in an ads, you know who they are, and then you connect that exposure to some kind of outcome, right? All of that requires a way to recognize people. All of that, right? And and we've been working with like

Rio Longacre (06:04.385)

That was the whole promise of digital advertising, right? Was you could actually target people.

Brett (06:07.226)

Yeah.

Brett (06:23.652)

Yeah, yeah.

Mathieu Roche (06:27.016)

you know kind of pocket knives. We've been you know taking pocket knives to a to a gunfight by using cookies and mobile ad IDs and IP addresses and and all of those like you know bad signal in terms of of of stability, in terms of efficiency, in terms of resolution quality, while you know at the same time Google and Meta have 100% logged in users that they can recognize across devices with and Amazon with and and and they've and they've run away with the market. So

Brett (06:31.856)

Yep.

Brett (06:49.26)

And in Amazon, yeah, and they're in

Brett (06:54.084)

Yeah.

Mathieu Roche (06:54.874)

We really came to this problem by saying like we just need a better way to do this because it's so fundamental for all the publishers and all the technology companies and all the brands outside of Google, meta and Amazon, basically, right?

Brett (07:06.084)

Yeah, yeah, and I think people started to realize that how polluted the data ecosystem was in programmatic and digital media. you know, before the sort of cook Yeah. Right, it was it was inaccurate in terms of your segmentation setups and your s

Rio Longacre (07:14.541)

And how bad a lot of the cookie data really was, right? It was just not accurate at all.

Mathieu Roche (07:19.184)

I I I think I think it's it's it's important here to differentiate identity and data, because they're two very different things, right? Data, good or bad, right, is like what you know about your customers or what you know about your prospects. And it can be very good or it can be very bad. And it can be first party very good, third party very bad, or the other way around, right? If you've got a very badly informed CRM system, your first party data sucks. So it's not like first party is good, third party's bad, as people have said sometime. It's not like if it's tied to a cookie or if it's tied to an email or whatever. No, no, no.

Brett (07:25.166)

Yeah, yeah.

Brett (07:36.548)

Yep, it's a good point.

Brett (07:42.031)

Yeah.

Mathieu Roche (07:48.933)

Data quality is data quality, right? You can have good data or bad data. Then the question of how do you activate that data? How do you recognize the people that you have data on? That's identity. And you can have bad good or bad identity, good or bad data, but they're they're very different from each other, right? And it's and and very often it's like one and the same, or it's considered to be one and the same, but it's actually very different. This is a connection mechanism. This is how do I see that when you're on a website, you're on a TV channel, you're on an app, I know who you are.

Brett (07:50.671)

Yeah.

Brett (07:58.288)

Yeah.

Brett (08:03.46)

Yeah. Well is i is it how you act?

Yeah, that's true. The Yeah.

Brett (08:17.038)

Yeah. And and and ensuring that there's some level of accuracy into your ability to connect to an actual human being on their device and scale. Yep.

Mathieu Roche (08:23.256)

Accuracy and scale. Yes. Because if you do that on three people, nobody cares. You need to be able to do that on three billion people for for because advertising is a scale game, right? So you need scale and you need accuracy. That's what makes identity quality identity spine is scale and accuracy, right? Quality data, there's so many versions of that. It's like it dip it's in the out of the holder, right? What's quality data for Delta Airlines is very different from what's community data for Ford. So that's very relative.

Rio Longacre (08:25.847)

Yeah, yeah, well, yeah.

Rio Longacre (08:39.713)

Yeah, you need quality data going in. Yeah, this is all garbage in garbage, garbage out. Yep.

Brett (08:49.904)

Yeah.

Rio Longacre (08:50.157)

Yep. Well, Matt, timing, so the timing of this is good. I can tell this is going to be a fun discussion. So you joined us at Cannes and we had a really good discussion. The timing of that was good too, because that was right after the, you know, the, big live ramp deal, like where they were officially acquired by, or going to be acquired by people. So this was announced in early June. We sat down, we talked about how identity was shifting towards measurement. What, what the next infrastructure might need to support. And we were thinking today could be a nice continuation of that conversation, Matt.

So looking at like as I guess Brett now we're talking before this about how like ID five can plausibly replace part of live ramps for all certainly but live ramp really does a lot of things right you look at ramp ID and identity resolution as Brett mentioned just part of it they also have all the publisher integrations so not used just for targeting but also for measurement itself so what they do is a little broader than just an ID, spine or graph but I guess looking ahead I guess our thesis was like the

the organizations, the ad to companies that win will not be the ones that necessarily the best just replacement identifier. So let's say, let's say I like people start using alternative identity, identity spines, right? We think it would be, it would be the companies that are able to build like really a coalition of different tools, workflows, identity graphs that are able to kind of build networks around it. That was kind of our thinking here. So maybe we start out by to follow up that conversation. You said when we met then that

You'd gotten a lot of calls, which is not surprising from people interested in who may, you know, maybe were starting to kick the tires a little bit. So wondering like, like, what is the state of that? Have those discussions progressed a bit? How are people feeling in the industry about, let's say who are using ramp ID? Are they looking at potentially like alternative identity provider, like ID five? Like how are those discussions fairing today?

Mathieu Roche (10:37.902)

So there there is indeed a lot of interest in in having alternative pathways, right? LiveRamp has built an amazing business and have has has become the default rails to for using data for advertising, whether it's targeting or measurement, right? and kudos to them, they've done a great job and they've become the easy button, right? And and because there is such an important network effect, if everybody's plugged into LiveRamp, it's very hard to not use LiveRamp.

Brett (10:54.221)

Yep. Yep.

Mathieu Roche (11:07.192)

And that's been their their their main kind of success mechanism, right? this transaction creates an im an imperative for a lot of people in this space to have another way to do things. Not necessarily to move entirely outside of LiveRamp, but at least to have options because you don't want to be dependent, because you don't want to give money to your competitor, because you don't want to you want to be able to support people who don't want to work with LiveRamp anymore, right? So it creates this need that didn't exist before because you know, if everybody's on Facebook, like you're just gonna be on Facebook.

Rio Longacre (11:07.788)

Agreed.

Brett (11:30.532)

Yep.

Mathieu Roche (11:37.156)

Right. so LiveRamp was kind of the Facebook, right, with that social network or that business network effect. And so even if you know a lot of people were not happy with the service quality or the the flexibility or the contracting or the financial terms, it was which was a common complaint. And and

Brett (11:53.828)

Which which was a common complaint. yes, we heard. Because they they were expensive and they charged for both data in, which they would absorb and then they would match your your data and then data out. And yeah. But there was very few others to go with and like and like it seemed like New Star for a while was a yeah.

Rio Longacre (12:03.591)

They charge people multiple times. People complain about that a lot.

Mathieu Roche (12:10.092)

Exactly. And and and it and it it was never it was never enough of a pain to create this need for the alternative.

Brett (12:19.117)

Yeah, the switching costs were yeah.

Mathieu Roche (12:19.908)

The publicist acquisition is is enough of a pain to say, okay, I'm gonna support the switching cost, but I'm gonna switch, or at least I'm gonna build parallel rails, right? And and so if you're a brand, it's like do you want to switch or not? Because do you want your measurement company to be owned by your agency? Or or do you want you know your measurement, your your your a data management company to be owned by a a competitive agency than your agency, or to or by an agency that also works with your main competitor? So there's a lot, there's a lot of strategic challenges for brands, right? For for agencies, it's kind of

Brett (12:27.266)

Yeah yeah.

Mathieu Roche (12:49.902)

Obvious, do you want to channel money to an entity that is owned by a competitor? Probably not. but even for the technology platforms, right? The people who have been plugged into the live ramp ecosystem, the SSPs and the DSPs and all of that. Now, do they want to only depend on live ramp or do they want to make sure they have alternatives? Because some brands are gonna say, like, hey, I'm not using live ramp anymore, I need another path into you guys. Well, we need to build that path, right? And so all of those.

Brett (12:54.852)

Yeah.

Mathieu Roche (13:14.688)

Reasons did not exist before and exist now, which means there is a lot more interest for alternative path. Yes. Yes.

Brett (13:17.315)

Yeah. The real need for a real alternative because Omnicom's been out in the public and Rio B Rio works at the organization saying that within what two years they're going to or w was it even less time?

Mathieu Roche (13:24.868)

Yes.

Rio Longacre (13:28.939)

No, no, he sooner than that. He said by December, end of December, which is when the transaction is supposed to take place. Right, I think we're paying right now, like we meeting Axiom's paying live ramp. think it's a little over $50 million for identity resolution services. Just my understanding that that, I mean, this was announced publicly that that would end in the end of year. So I'm assuming Axiom is building up some kind of alternative to plug that gap. But I guess, now that you're looking at this, over the years, I've had so many...

Mathieu Roche (13:28.962)

Next year. Yeah.

Rio Longacre (13:58.807)

tech companies approached me, we have a live ramp takeout policy. We're going to be able to activate data cheaper, better, faster than them. Yet, I mean, to your point earlier, inertia played a big role, like switching's hard, it's expensive.

Mathieu Roche (14:11.392)

It's not it's not about supply, it's about demand, right? If people are not interested, it it doesn't matter that you've got a better product, they're just not interested. And they're not interested because it's the easy the easy button. The switching cost is too high, it's the easy button, right? And and and still now, and still now it's gonna take a long time for people to replicate to build alternative capabilities across the entire scope. Because Libram does a lot of things, right? They have identity, identity resolution.

Brett (14:12.271)

Yeah.

Brett (14:19.427)

The switching costs are too high to leap, so yeah. Yeah.

Brett (14:32.781)

Yeah yeah, and let's th let's break that down.

Mathieu Roche (14:35.886)

They also have a lot of data management capabilities with the Abu acquisition. They have a big data marketplace, data store. They have pipes into a lot of destination.

Brett (14:40.141)

Yeah.

Yeah. They they have the partner the partner connectivity with twenty five thousand publishers, which gives them this real time signal from a broad swath of publishers, right? The clean room which you mentioned.

Mathieu Roche (14:48.62)

Absolutely, yes, yes.

So so that's that's a lot. You don't necessarily need everything. Not everybody uses everything. And so you can you can pick apart different parts or different use cases that they support, and maybe build an alternative to this one use case. So this other one use case, right? So if you want to monetize your data, well, there are other data marketplaces you can monetize through, right? If you want to onboard your data as a client, there are other ways to onboard your data. So, use case by use case, it's it's kind of easier to build replacement path or alternative solutions, but across the board, LiveRam.

is a very a very complicated and very broad operator.

Brett (15:23.309)

Yeah, and and replacing ramp ID is not the same as replacing LiveRamp. I think that's the key point here, right? LiveRamp's a much deeper do you let me ask you this follow up question is do you think that LiveRamp their data collaboration play, which was a big sort of I wouldn't call it a pivot, it was a pivot in their positioning in market. They're like we are now we're not just an onboarding company. Yeah. It was

Mathieu Roche (15:29.102)

Yes, absolutely.

Rio Longacre (15:29.205)

important.

Rio Longacre (15:41.965)

But they made a part of it, part of their brand promise and their whole go-to-market was centered around data collaboration and Haboo,

Brett (15:47.671)

Yeah, and it was and you know, and I've heard some there's some certainly some limitations with Haboo from a technical standpoint. It was more of a workflow management play than it was a true clean room. That's what I've I've heard from a lot of product leaders. But do you think the data collaboration play was real or do you think it was much more of a position in market to change how their valuation, for example, because they were being sought thought of as like an onboarding company, right? Potentially, you know. What do you think?

Mathieu Roche (16:14.734)

I mean I I I I don't want to comment on their on the you know the communication strategy Wall Street. That's that's not like I I don't think I'm relevant to comment on that. What what I know is that when you're when you have valuable data and you wanna use valuable data for advertising purposes, you need ways to use data that are secure, that are you know that allows and and and data is is is better when it is augmented, when it is shared, when it is, you know, in in in in safe manners. Like you you always only know

what you know, right? And other people know more. And so I think it makes total sense for LiveRam to have you know data collaboration capabilities across their their customers or to enable data collaboration across their customers. So I don't think it's a it's just it's just messaging. maybe they needed to you know justify the 200 whatever million dollar acquisition of Habu. And so that was part of the justification to say we're gonna put data collaboration front and center. But

Brett (16:46.423)

Yeah.

Brett (17:10.733)

Yep.

Mathieu Roche (17:11.332)

But it it does make sense in the in the broader kind of suite of capabilities they offered to have the ability to facilitate collaboration between their customers. yeah, no, it makes sense. I don't I don't I this is one of the use cases though, right? Like it's one of the use cases. And so if you look at that, there are other ways to do data collaboration, right? A lot of people use Snowflake, a lot of people use Optible or or Permittive or whatever. So it's it's it's one of the one of the use cases. There are other use cases that that, you know, with the with the data store in particular.

Rio Longacre (17:21.069)

Yeah, Matt, think I agree.

Mathieu Roche (17:38.084)

Monetizing data, right, is a big is big use case for their customer base.

Brett (17:40.45)

Yeah. Incremental reach is another big play though, right? When you're c when you're when you're collaborating two different data sets, you're trying to find an incremental reach on top of what you both both partners have on their own, right?

Mathieu Roche (17:45.818)

Course.

Mathieu Roche (17:50.02)

Yeah. Yeah.

Rio Longacre (17:51.309)

Yeah, yeah. But looking at the acquisition, for those who didn't pay attention at the time, took place in 2024 in January of that year. It was for approximately $200 million, $170 million in cash. It was paid at closing and then another $16 million in restricted stock. so mine, yeah, for Haboo, right. So my understanding, I mean, looking back as they were saying, it's a 10x valuation. That's saying it was close to 20 million. But I've heard it was actually they were closer to 10, which I'm not sure.

Brett (18:08.408)

For Hubble. Yeah.

Rio Longacre (18:19.885)

These are just rumors I've heard. So they paid quite a bit for it. But I think that shows that they believe strategically it was important to their core business, even though it wasn't a big business. I mean, the data clean business certainly hadn't taken off as much as many people thought, even though it was one of the leaders. But they did get acquired info, something got acquired. So I agree with you, Matt. I think that they felt it was important. It would add a little bit of incremental revenue to them. Not a lot, but it's interesting. You look at Livram, they were a highly profitable company with an

Brett (18:35.329)

Yeah.

Rio Longacre (18:48.695)

generating a ton of money with not a lot of people, actually. I don't think many people realize this. So it was actually a pretty good acquisition. But then that begs the question. When you look at their business, they were considered the Switzerland of ad tech or of advertising, right? Because they were, everyone used RampID to your point. All the agencies used it. was connected to all the DSPs. had the 25,000 plus publisher integrations. it, like, so they're saying that like even post acquisition between audit controls,

Brett (19:12.878)

Yeah.

Rio Longacre (19:18.295)

client government access, firewalls, et cetera, they're going to remain Switzerland. They're going to remain neutral. Do you think that's actually a realistic matter? Do you think that, and I know obviously you're going to be a little biased in your answer, but I'm likely you're taking that. Is that realistic that they can expect that people will believe that?

Mathieu Roche (19:35.877)

I mean, I I I don't think people believe that, right? Whether whether they are doing everything they can to maintain kind of a Chinese world between them and the rest of the publicist group, mm probably. But at the same time, Arthur Sadoon in his in his initial press release said that they were gonna use they were gonna leverage LiveRam to do data co creation, which is basically training models on everybody's data to understand like this is the definition of of of risk if I am

a client of of LiveRamp and my data sits in LiveRamp and publicist says, we're gonna use that to train models and build, you know, kind of new data sets on top of people's data, like it feels kind of, you know, an antagon antagonistic with what you just said about the fact that like the data's gonna be safe and like nobody's gonna touch it and it's gonna

Brett (20:20.108)

Yeah. And yeah, and you're in your and you're chain you're train we talked about this with with Eddie Drake from Snowflake, right? It's kind of the same idea of it's data sovereignty. It's the data sovereignty issue, right? If I'm a brand and I've got data and now you're tr and I'm and and you're developing these models and patterns and you're and you're training training that for other purposes to be sold to potentially competitive category competitors, that's a big

Mathieu Roche (20:41.46)

I think I think it's a question of of alignment of interest, right? Like there is there is a lot of potential conflict of interest and the and the it's potential, but potential is enough to scare people off. And and and if you you know, like you can you can again, like you can have very good intentions if the conflicts of if the potential for conflict of interest is here, at some point something's gonna go wrong because those are, you know

Brett (20:47.031)

Yeah.

Brett (20:51.692)

Yeah.

Mathieu Roche (21:07.928)

companies that have to make money for a living and and so the

Brett (21:10.22)

Yeah, and w and what do you think the fundamental strategic objective of publicist buying LiveRamp was? Was

Mathieu Roche (21:15.524)

I think it's a very I think it's a very smart move because the LiveRamp has amazing, you know, capabilities to build to to to to to use data for advertising. Publicis is a big advertising agency, they have a lot of data because they have Lodomy and Epsilon. so they've just bought the ability to monetize, to leverage and monetize those assets at great scale, and at the same time they've kind of pulled the rug from under everybody else and put everybody backwards two or three years, which is great from a competitive standpoint.

Brett (21:30.371)

Yep.

Brett (21:39.085)

Yeah.

And and they've moved closer to becoming a product organization, right? Product product enabled we'll say. Not not a product organization, but tech enabled.

Rio Longacre (21:42.699)

Yeah. Yeah.

Rio Longacre (21:48.319)

Or tech enabled, which the agencies have always struggled with, And Brett, yeah, that's a good point. And in fact, the analogy, like the best one I heard was they took like the queen off the board. They just went and took their opponent as queen. So it's a really smart move. And I will agree with what your point, the point you said about their, strategically for them. And I know for sure. So I just talked to a friend who works at a big pharma. He said, you know, he saw the pitch from all the big hold coasts. He said the pitch from, from publicists was all around live ramp, all around. have ramp ID.

We have the pipes for the internet. We can control activation. We can control measurement. We can give you what we can get you what you want. We can get it cheaper, faster. So it is core to their pitch already, even though the acquisition hasn't gone through and they're winning a lot of their pitches. I'll grant them that. So they'd certainly have it nailed, whatever they're saying. But looking at live ramp itself, and Brett, you made the point earlier that it's not just ramp ID, right? So Matthew, looking at...

Brett (22:32.236)

Yeah.

Rio Longacre (22:42.377)

Onboarding, ID res, partner connectivity, clean rooms, activation, measurement, governance, all the things LiveRamp does, which of those can ID5 credibly replace today? And then which partners would you want to bring in as well if you're, let's say, advising a brand on their, on a strategy to address those gaps?

Mathieu Roche (22:59.844)

So it's it's not just ramp ID, but Ramp ID is core to everything. And that's and that's really important to understand, right? You can't have like everything, all of the use cases that you've just described, identity resolution, collaboration, all of that. RAMP ID is a component, it's a really, really strong component to that, right? And and and the only alternative to ramp ID in the world is the ID5 ID. There is no other kind of you know open, interoperable ID, universal ID out there.

Brett (23:23.106)

Yeah.

Mathieu Roche (23:29.22)

that has the scale that that ID5 has. We are you know distributed across hundreds of thousands of media owners. We're integrated into hundreds of technology platforms, SSPs, DSPs, data collaboration, data management platforms. So that ingredient is unique. On its own, it doesn't do anything.

Brett (23:47.458)

Yeah.

Mathieu Roche (23:47.717)

But it is it is an ingredient of most of the use cases that LiveRAM is able to support. It's 90% of the recipe in some cases, right? Because custom onboarding, for instance, CRM onboarding is basically just mapping an email to an ID, right? And so that is 90% of that we can do. For data collaboration, for data store, for data distribution, you have a lot more capabilities to build around the ID itself. But the ID is kind of central to everything that LiveRAM does.

Brett (24:01.186)

Yeah.

Mathieu Roche (24:15.682)

and that's why we think we have a really s a really good role to play in the re recomposition of the market, because we again we are the only alternative to the ramp ID in the market today.

Brett (24:25.313)

Well and and so so speaking of mapping IDs, right? There's been a you know, there's a lot of complaints when you when you bring up let's say it's a customer file, a CRM file, you've got y you know, hashed emails and you're trying to map them.

to to cookie your or bo your maids or other IDs to target across digital media, right, that you lose a huge amount of your data, 50% of it. We talked to Scott McKinley about three, four weeks ago, right, and he said that that truth sets research has shown that that even basic mapping between things like IP and email like succeeds like 16 less than 20% of the time, right? And that and that general mapping is generally 50%.

Right, so it's almost a flip of a coin of whether or not you're actually going to be able to activate or even measure against that audience that you've, let's say, uploaded into to LiveRand. Where where do you see improvements, right, in that area that are

Mathieu Roche (25:17.636)

Yeah. You know the the the you know the difficulty here is that a lot of people work with what we're gonna call static data. and and the internet is not static at all. The internet is very dynamic. Yeah, batch file just like server side, right? Like a lot of people have data that sits on servers that is tied to an ID that has been collected two years ago. And if it's a cookie ID, if it's an IP address, if it's a you know, even emails, right, they change.

Brett (25:25.164)

Yeah.

Brett (25:29.653)

Like batch, batch files?

Brett (25:34.253)

Yeah.

Brett (25:39.019)

Yeah.

Brett (25:44.514)

Yeah.

Mathieu Roche (25:44.571)

They change very often. Some change every day, some change every month, some change every year. Some are more persistent than others, but like there is a there is it it this is live matter, right? This is like this is things that evolve, right? And so when you look at it, when you look at it, and it's it's it's not just it's the quality of data, but it's also the quality of the identity spine. And to have a a a live identity spine, you need to have live integrations, you need to have connections with those those hundreds thousand publishers that I mentioned, what they give us.

Brett (25:47.211)

Some are more persistent than others, yeah.

Brett (25:53.74)

Yeah.

Brett (26:10.528)

Yeah.

Yeah, I d I D recency. I D recency is critical.

Mathieu Roche (26:13.676)

Is a real-time map. It's it's more than that. It's a real-time map of how the IP address evolved, for instance, right? Which is super valuable because IP addresses are a very important component of all of the matching that is done in the industry. So having that that that live pulse, right? That live finger on the pulse of

Brett (26:22.092)

Yeah.

Brett (26:36.066)

Yeah.

Mathieu Roche (26:42.084)

the the the the the behaviour of users allows you to have a fresh like a a very kind of up to date understanding of people's signals, of people's devices, which is the the base for recognizing who they are and where they live and and and then tying that back to data.

Brett (26:58.943)

And so so you think those match rate reports are kind of wildly overstated or you know, in terms of because because it's it's sort of maybe not apples to apples, because you might be dealing with certain you know, 'cause it's it's sort of an aggregate read of a bunch of advertisers that have done some

Mathieu Roche (27:13.314)

Match match rate is match rate is potentially the the least well understood measure of success in our industry. Everybody puts a different definition between between match rate. Yeah, yeah, but everybody has a different understanding of what that means and everybody has a different understanding of what good is, right? Matching matching a you know a list of email address of customers that date back maybe ten years, right, to an ID.

Brett (27:18.827)

Yep, yeah. But everybody asks for it, right?

Rio Longacre (27:23.009)

Yeah.

Brett (27:26.839)

Yeah.

Mathieu Roche (27:37.861)

That's easy to do, right? You can even create an ID on the fly and well, guess I've got a 100% match rate because I've created an ID for every email you send me, right? The real question is like, how many of those IDs are you gonna find in bid requests, in transactions, that you're gonna have to the the ability to to target or to measure from in seven days? That's the real that's the real criteria. How do you go from like your not not just like mapping your email to an ID, but then an ID, then finding that ID in the wild? That's the real that's the real thing. And so

Brett (27:39.105)

Yeah.

Brett (27:42.604)

Yeah.

Brett (27:49.004)

Yeah.

Brett (27:53.771)

Yep.

Brett (27:57.441)

Yeah.

Brett (28:04.074)

Yeah. To be able to activate against, yeah.

Mathieu Roche (28:06.296)

A lot of people have like, yeah, 20, 30, 40, 50% match of of of email to ID is considered not that bad because probably half of your serum database is dead anyway, right? So like, you know, like that's you know, if you look at like mapping an ID to the emails that were active over the past 30 days, for instance, I'm sure we can be pretty good with that. Because you know, the ability to recognize people's email addresses is pretty good, and the ability to map that to an ID, provided you have the footprint, is pretty good too. Then the question is, is this ID active?

Or is this a is this an ID that was created like in a in in a sandboxed web web page out of Facebook, which you know disappeared the moment the the the the web page was the the the s the the yeah the the the the browser was was closed, right? So

Brett (28:36.075)

Yeah.

Mathieu Roche (28:54.926)

The the real challenge is not just to map an ID to an email, it's to kinda find that ID in the wild and find it as often as possible. And that's where the scale really matters. The scale of the because in the end, if you can only advertise to one percent of your customer base, who cares? Right? So you need to recognize them and you need to be able to reach them. And the campaign is gonna be 30 days long. So like that's really the the the window of opportunity. That's what how we should measure match rates.

Brett (29:03.028)

Yeah.

Brett (29:07.392)

Yeah.

Rio Longacre (29:07.414)

It doesn't matter, yeah.

Yeah.

Brett (29:10.529)

Yeah.

Rio Longacre (29:17.581)

So think we, so I thought we could maybe like, cause we do talk a lot about publishers and pub tech in this podcast. thought that'd be an interesting thing to talk about for a couple of minutes. So turning to the publisher side of things. So, and I know this is an area you've been doing a lot of work. knowing that many publishers will often run multiple IDs, right? And they'll, but at the same time, it can be very difficult for them to really understand what is the net contribution, right? For each one. Like how do you advise publishers when looking at, okay,

How do you prove out incremental bid density that ID.5 resulted in, or higher CPMs, or better fill rates?

Mathieu Roche (29:53.671)

we give we give them a dashboard. We give them a dashboard and they can do A B test the dashboard has baked in A B testing capabilities. So you can say with or without ID5 IDs, what what what's my RPM, what's my CPM, what's my bid rate, what's my bid density, right? There's like a we we a big part of our distribution goes through prebid Amazon APS and then Google Direct is a kind of slightly smaller distribution shell for us, but through prebid and Amazon,

Amazon has their own, I think it's called Signal IQ. We have a identity insight, which is a pre bid module that you can install on top of your prebid ID adapter. That's gonna that's gonna do the work for you. So there's there's very clear data that supports the fact that if you're passing ID five inst in addition to other ID or just on its own, you're gonna see an incremental, like you know, you're gonna see a lift in terms of demand, in terms of bit price, in terms of density. that's that's measurable.

That's very, very easily measurable. You need to do A B testing, you need to understand how to build like you know A B test segments and all those things, but like this is something we've packaged in a tool, so that's an easy answer. Yep.

Rio Longacre (30:55.351)

So you make it easy like that.

Brett (30:55.744)

Yep. Yep. And and you guys are you guys are also partnering with Optible, I think I read, right? Right? And that and that and for those that don't know, Optible is an agentic audience and and and a data collaboration platform, right? And and you guys are the

Rio Longacre (31:09.581)

I think they were a DMP for publishers at one point, but they've recently kind of pivoted heavy into Gentic, yeah.

Brett (31:14.686)

And so and so I think the value prop there is that ID five augments the publisher's private graph, right? And operates in the and it operates inside sort of a clean room environment so the publisher's not you know, none of their data is sort of escaping. How big of a play do you think that is going forward in terms of helping publishers, you know, develop more fidelity around their own audience data?

Mathieu Roche (31:37.105)

Publishers need identity for two things. One is to better to make their inventory more valuable, right? And so that's the that's the PubMatic and the Amazon integration. It's like you just light up IDs and and by by giving more chances to the buyer to recognize the customer, you're increasing the value of inventory, right? and the other thing they need is to recognize their their visitors better because they can make different decisions if they know who the visitors are, right? And so it's like this there's a whole like audience enrichment use case.

Brett (31:42.283)

Yeah.

Brett (31:47.872)

Yeah.

Brett (31:54.049)

Yep.

Brett (32:02.186)

Yeah.

Mathieu Roche (32:05.604)

You know, it's the whole yeah, it it's it's or or or or that are authenticated but that you don't know anything about, right? So it's it's recognizing who they are and having more information about them. So like it's enriching your understanding of the fact that well, you're on a sport side, so obviously you like sport, but you're also a father of two and like you may be looking to buy a car, right? like so knowing your customers is always a good idea, right? And so helping publishers know their customers so that they can monetize them better.

Brett (32:07.188)

It's visitors that are unauthenticated that have have come to the site either blindly or or

Yeah.

Brett (32:24.746)

Yeah.

Mathieu Roche (32:32.154)

They can bring more value to the brand, they can package things differently, they can build they can build PMPs, they can that that makes that makes a lot of sense. So those are really the two use cases that that we see for publishers. and and in both cases, a a strong identity spine, a strong ID play makes makes just the use case more efficient.

Brett (32:51.82)

Yeah.

Rio Longacre (32:52.577)

You've also been doing some work with Permitive that I was interesting. was reading here when we were doing our research. lot of you can maybe comment on that a little bit because we've had Joe Rudon here, really smart guy. They're doing some innovator stuff over there, especially in the publisher space.

Brett (33:01.078)

Joe's great. Great guy.

Mathieu Roche (33:05.744)

I I mean Joe's been a been a you know a a white knight for publishers for a long time, for as long as I know him and and and I applaud him for that effort. I think yeah, they've built a great business. They've been a a very important partner of ours, like like like Optable and like many others, in in in bringing identity capabilities to use cases that matter for publishers. Again, identity is an enabler, it's not a it's not a feature, it's not something that you know you don't you don't buy identity.

You buy something that relies on strong identity capabilities, right? In the case of Publisher, you buy audience enrichment, you buy audience monetization, you buy those things, and you buy them from Permutive, you buy them from Apptable, you buy them from others. and and so those guys have have decided to embed ID five capabilities inside of their their services because it makes their services better.

Brett (33:34.155)

Yeah.

Brett (33:54.892)

Yeah, it's audience data validation, it's data enrichment, right? And and we talk about shrinking economics of of the open web all the time. And and Joe to to your point about being a white knight, like he really is leading that charge of like, hey, we need to enable

especially premium publishers, right, to to compete with the walled gardens, otherwise everything's gonna be a walled garden. Right? And and and giving them the ability to to harness and manage all of that data that they see. And then you guys are enriching it and validating it, ensuring that it's accurate, right? I is a is a power play. It's a data moat, right, for the for the publisher community.

Mathieu Roche (34:31.824)

I mean again, like you know, knowing your customers are always always a good idea, right? And I think for years publishers have been have been just like surfacing at impressions, right? And leaving the knowledge of the customer to the buyer. And it's put them at a massive disadvantage because well if you don't if you don't if you don't have anything to negotiate with, you can't negotiate and your price is gonna go down, right? And and the the world gardens have been

Brett (34:46.4)

Yeah.

Brett (34:54.314)

Yeah.

Mathieu Roche (34:56.996)

Very smart in in doing a lot of things with their own data, their own knowledge of the customer, their own optimization mechanisms. So they haven't let the buyer decide how to serve an ad on Google or Meta. They've decided themselves. And a lot of publishers on the open web have let the buyer decide. And and and and the the the the market dynamics are very different. It's not a criticism of the publisher strategy. It's just the way the market has evolved, right? It's given a lot of power to

Brett (35:08.81)

Yeah.

Mathieu Roche (35:20.42)

To the buy side because most of the knowledge was on the buy side. And we've seen over the past five years a lot of that knowledge going back to the sell side, a lot of intelligence, a lot of like audience segmentation, audience qualification, a lot of P and Ps, a lot of sell-side decisioning, right? All of that is good because it gives publishers a a a better negotiation position, right? They know more about their visitors than their than their clients, and so they can price accordingly, they can decide how to serve what to whom, right? They can be more involved in the actual

Yeah, the actual negotiation of the of the of the trade, which is which is positive.

Brett (35:54.346)

Do you do and do you see with with that in mind, I mean, because you know, I mean there were a lot of publishers that were moving from a traditional linear world, you know, like Gannett, who I used to work for a w a long time ago, that that just didn't have the core competency. They weren't necessarily digital native and they also had kind of an innovator's dilemma. Right? They're like our clients are asking for belly wraps, right? And that's what's you know, and so they were sh they were slow to shift, but then they they the

the digital natives, the Googles of the world, the you know, came in and said, hey, we realize that this new data capture capability, Facebook is in remember how much Facebook data you used to be able to see when we're like Ryu and I ran a startup. Yeah, they locked it down a long time ago. I mean we I remember seeing the amount of like PII that was available at an early day, but they realized very early on if they could build the mode around this, it will be monetizable and and then a lot of publishers were kind of scrambling to catch up, right?

Rio Longacre (36:33.421)

Yeah, they've locked that down though, though. Yeah.

Mathieu Roche (36:52.976)

Yeah yeah, I I totally agree. And and but but it's good to see that they're catching up.

Brett (36:53.033)

So there was no the

Brett (36:58.613)

Yeah.

Rio Longacre (36:59.117)

So quick questions like we've been about agentic trading. So we had PubX on recently. Andrew Moll talked about it. It's not a lot of money, but they're transacting around $3,000 per day. They haven't disclosed which clients yet, but the fact they actually have buyer and seller agents negotiating is kind of amazing. I've talked to Optimal recently with a couple of publisher clients where they're to apply, know, supply sell side agents who can then in theory negotiate either directly with agency, like hold co-agents representing brands or directly brand agents representing buy sides.

It's kind of our hypothesis is identity is going to be a very important piece of this puzzle, right? If we're going to rewire ad tech to a certain extent, right? That you need to be able to verify who people are and brands need to know with some assurance that they're targeting the right people. So I'd like to know like what discussions are you been having so far with some of these players, role of identity and how, like, how do you, how does identity end up being an important part of this without, without being really just another toll booth? Cause that's the problem people have with the current ad tech kind of supply chain thoughts.

Mathieu Roche (37:58.181)

So I I think most so most of the the concrete applications of agentic I've seen right now have been kind of around the transaction itself. It's been in the audience discovery, it's been in the negotiation, it's been in the you know, instead of sending an email and asking publishers for their for their proposal, right, sending RFIs and like stuff, like this is happening agentically. So it's it's it's media planning, it's audience discovery, it's media buying in some cases, but then most in most cases, when you want to do a targeted campaign.

You cannot you cannot execute it ru with with a with you you need you need programmatic systems, right, to decide I'm gonna buy this one and not that one, right? So right now a lot of that it's measurement, right? But a lot a lot of that that agentic transformation is happening on the edges of the actual transaction. The actual transaction still goes by and large mostly through our TB pipes and SSPs and DSPs. I think there are there's there there's the next phase of the evolution is probably to compress that supply chain. I don't know that you need

Brett (38:37.277)

Yeah.

Mathieu Roche (38:57.366)

as many hops that we have today along the the open RTB transaction, right? And so if you are able to compress that supply chain, you can compress the margins, which means the market becomes more efficient, right? The transaction itself is probably still gonna go through a system that allows you to bid or to serve an ad to one user and not to another one because you want to serve to the right people, you want a frequency cap, you want to measure on individual basis. So I I think

Brett (38:59.56)

Yep. Yeah.

Rio Longacre (39:00.341)

I hope not.

Brett (39:08.521)

Yeah.

Brett (39:20.2)

You said com you said compress the margins. Do you do you do you think that it would free up margin that can then rather than the the forty percent, the forty cents every dollar that's going to a middleman?

Mathieu Roche (39:27.854)

Yeah. I I I th I think I think that's the promise. That's the promise of of that of that of that transformation. Yes, absolutely. Yeah.

Rio Longacre (39:29.111)

More than that.

Brett (39:34.9)

Yeah. And in

Rio Longacre (39:35.277)

But who is the identity layer in this scenario, right? Does the publisher own it, the buyer, the consumer? know. Who owns it? Where does it sit?

Mathieu Roche (39:42.243)

N the the none of those can own like identity is a currency. Like the buyer and the seller cannot own the currency. Otherwise, like they can't trade with each other, right? Like why would you advise trade with yours? You want to trade with me? Like it doesn't work. You need to have a in an open, right? neutral, independent, right? I'm gonna use those words with with a lot of quotes because

Brett (39:49.31)

Yeah.

Mathieu Roche (40:04.792)

I think openness is really what matters, right? It needs to be connected. It's composable identity. It needs to be connectable to everything. You want to trade using this SSP, using this DSP, using this agent, using this data enrichment system. It's gotta be, you know, open to those different systems. but I don't think outside of Google and Facebook, I don't think any seller is powerful enough to say this is the currency you need to use to trade with me. And

And and outside of Google on the buy side with D V three sixty, I don't think any buyer is is strong enough to say this is the the currency you need to trade, I need I want to trade with, right? So that's why you have to have this open system, like this this open universal kind of identify identity system that that we that we are a part of today, and and that makes this this ecosystem scalable, right? Because again again, like unless you are

the dominant player and you can impose your and even Trade Desk as an impose UID two as a currency, right? So like it's I think that that notion that identity will be owned by one or the other doesn't make any sense today. Doesn't make any sense tomorrow because you've got agents talking to each other rather than people.

Brett (41:10.034)

Yeah y yeah.

Brett (41:14.94)

Yeah, and for this data matching and data enrichment and resolution, it it needs to happen both on the bell b there's benefits on the buy side, you know, as as as brands start to build their own data sort of infrastructures they've already built them, but which which is largely driving their sort of IP. But it also has to hap happen on the publisher side so that they've

Mathieu Roche (41:31.717)

But that then then then then it's a fidelity issue. You could imagine like having three different identity systems that s serve different roles, but then there's loss in translation every time you translate. And so that's why you need to have a an homogeneous identity spine. And the more homogeneous it is from the buy side to the sell side, the more efficient it is, because then you don't have to translate. You don't have loss in translation, right? But you y y that so so what's really important is that the the one in the middle, the one that does the handshake between the buyer and the seller, is a currency.

Brett (41:40.958)

Yeah, that's true.

Brett (41:49.066)

Yeah.

Mathieu Roche (41:59.461)

That's what it is. It's just like it's a framework. It's like, hey, listen, when you we're we're we're we're we're both we're both gonna speak French so that we understand each other. We're or we're both gonna speak English so that we understand each other. My native language is French, your native language is Spanish, we're choosing to speak English together because so that we understand each other. Identity is exactly that. I I have my data tied to an email address, you have your data tied to a mobile ID ID, we're gonna use ID five because so that we can understand each other, right? That's what identity is. It's a common framework.

Brett (41:59.625)

Yeah. It it and and you don't you're not shy about us

Rio Longacre (42:00.523)

Yeah. Yeah.

Brett (42:06.611)

Yeah.

Brett (42:15.465)

Yeah.

Brett (42:22.91)

Yeah. That's a that's a really good analogy.

Rio Longacre (42:25.473)

That's really good. By the way, that's a great definition. That's why LiveRamp has been so successful. RampID became the currency of ID across the open internet.

Mathieu Roche (42:31.308)

Exactly. And I I would I would argue that's why ID five has been so successful as well. Is because we have this kind of this notion of universal, right, open ecosystem that everybody can tap into. And so instead of having to go through complicated translation, which would be analogy for cookie matching, or or to kind of you know yeah, i it it it just makes communication better if you can all speak the same language. It may make trade better if you can all use the same the same currency. That's what identity is.

Brett (42:53.257)

Yeah.

Totally. And y and you and you don't hesitate using the word currency, which I know carries a lot of baggage, having been at Nielsen, right? 'Cause you're saying you're gonna trade on something that people can can can trust, right? And you know, there is some distrust in in some of the digital data out there as we know, and this can help hopefully solve some of that. And you you've actually I think you've been quoted as saying that that identity should be treated as like a measurement mechanism. Like like is that is that basically what you're talking about here? I mean sort of how how how would you define that?

Mathieu Roche (43:23.93)

I mean I I think I think the the the the the discourse, the message, especially around around cookie deprecation, which is when kind of you know identity became sexy, I guess four or five years ago, was very focused on targeting. But it doesn't matter that you can target if you can't measure. It doesn't matter, right? And so we've kind of we've wrongly chased

Rio Longacre (43:33.484)

All this started, yeah.

Brett (43:34.665)

Yeah.

Yeah.

Brett (43:43.454)

Yeah. Yeah.

Mathieu Roche (43:48.165)

The ability to target as the number one use case, while we should have chased, at least as importantly, the ability to measure, right? And identity is equally important to the both of them, right? You cannot measure, like measurement is measurement, measurement from an outcome standpoint, right? I'm like Nielsen measurement, right? Like region frequency measurement is different. Outcome measurement is someone has seen an ad, and then they've gone to a store, they've loaded an app, they've been to a website, right? That's like the only way to know that, well.

Rio Longacre (43:54.551)

Yeah, it's funny you say that though. Yeah.

Brett (44:07.689)

Yeah.

Brett (44:13.375)

Yeah.

Mathieu Roche (44:17.26)

0.5% of them clicked on the ad and went, and so you can you can follow the the the breadcrumbs. But most people, advertising is not a direct response mechanism by and large. Most people watch it ad on TV and they're like, that's an interesting car. Two weeks later, they shop at the dealership to see if the car is fine. Right? How do you connect the dots here? Without a way to recognize that the person that was served the ad on the TV is the same person that went into the dealership, you cannot do that. So identity is fundamental to measurement, to outcome measurement. Without identity, you cannot measure.

Brett (44:32.841)

Yeah.

Mathieu Roche (44:46.286)

And you just have to guess. You do panels, you do things like that, which is guessing. It doesn't work. So so yes, targeting is important, but measurement relies on identity.

Rio Longacre (44:48.087)

Yeah. Yeah. Yeah. Well, that's a great point, by the way. Yeah. In fact, clean rooms, clean rooms, which we discussed before. think a lot of people, when they first came out, was all about targeting, right? But I think most importantly, the use cases, people who got the most value out of initially and probably even to date were actually related to measurement. I think that's a really good point about how, how,

Having a currency enables you to do that, but then be able to match that across different properties and different platforms is very critical if you want to have any chance of measuring accurately. I like that a lot.

Mathieu Roche (45:23.618)

And and and and and again, the way the reason why it's important is if you don't measure, you cannot optimize and you cannot justify the investment to the client, which means that the minute they're they they need to cut costs, they're gonna cut you.

Brett (45:24.168)

Yeah.

Brett (45:34.652)

Yeah. And and so do you think this has any role to play in in sort of multi-touch attribution? I know it's a it's a it's a

a topic that's probably been beat to death. But but there was a lot during the you know the age of data deprecation, which was you know about si maybe seven years ago and and all the brands were freaking out that their you know that their house of their house of data cards may just collapse, right? We used to do a lot of that consulting on the New Star. It was sort of like h h helping educate these clients that they're building an infrastructure that will stand the test of time. but but you know there's been an argument that MTA sort of collapsed

collapsed under maybe the weight of data deprecation, and that you couldn't really connect all these signals and really come up with a a read on incremental contribution of a particular channel or a particular campaign. What what are your viewpoints on that and and compared to like more broad strategic models like MMM, like how would you think of that in the identity world?

Mathieu Roche (46:36.548)

Between MT and MMM, I think this is a I mean, this is a a a specialist debate. I don't I think identity is important for both, right? I'm not I'm not gonna I d I don't know that I have the the the the qualifications people to tell you that MT is better than Mm. I think you in both cases you need to know who's been exposed and and who's done what afterwards, right? And so identity is important for that. Signal deprecation is an issue because if you don't have a signal to connect between the the the the exposure and the outcome, you cannot do that.

Brett (46:44.158)

Yeah.

Mathieu Roche (47:03.124)

but that's exactly what ID five exists. It's to kind of replace the signals that have gone because Google and Apple have decided to remove the signals for everybody else and keep them for themselves. and and that's what we're here to to to compensate for. so I again like if you don't measure it doesn't matter, right? And and we need to make sure that measurement is front and center and and in a context where there is it is harder technically and legally, right? Both

Brett (47:12.264)

Yep.

Brett (47:17.726)

Yeah.

Brett (47:30.846)

Yeah.

Mathieu Roche (47:31.652)

to to collect signals from devices, be able to recognize who people are. it's really important that we invest in those capabilities because without that n nothing matters. Like again, if you if you can't justify the the performance of the of the investment that the brand is making with you, whoever you are, whether you're a platform, a c a publisher, whatever, like you need to be able to to justify the performance of your investment.

Brett (47:45.577)

Yeah.

Brett (47:52.383)

Yeah, either they're gonna be flying blind or they're gonna be reducing investment because they can't prove and they're gonna be putting money back into things that they know they can quantify, like direct mail and things like that that they they have l you know, d deterministic data. Or Google and Meta. Yeah. a question about this is that you you know, you talked about in the sort of the beginning of the hour, the top of the hour I should say, that you guys have a pretty broad international footprint. Would you say fifty percent of your total graph? What was what was the number?

Mathieu Roche (48:02.785)

Or or or or Google and Meta.

Mathieu Roche (48:18.684)

it's probably over fifty percent of the graph is outside of the US, yeah. Yeah. We cover Europe and we cover a few countries in in in AIPAC, Japan, Australia, New Zealand, Singapore, yeah. So yeah, it's mo mostly outside of the US.

Brett (48:22.642)

Yeah and and

Brett (48:30.492)

Yeah, and w and when I was throughout my career, but I saw that as kind of a major problem, for the for the vendors that I worked with. The excellent Nielsen and Nielsen Marketing Cloud, at New Star, it was like once you started to go either

Rio Longacre (48:41.143)

Well, for LiveRAM, too, Brett, I mean, they were mostly US. think they didn't.

Brett (48:43.654)

Yeah, and and Library was mostly almost almost predominantly like ninety-five percent. And and you know, that you had these global brands, you know, 'cause 'cause our most of our client base was very large brands that had multiple locations and we just couldn't didn't have the data fidelity for either activation or or measurement use cases. How are you guys what what are some of the differences in these international markets considering some of the privacy landscape, like GDPR?

Mathieu Roche (49:05.124)

We're not we're not the the main difference is that we're not an American company, so we didn't have the the the the luxury of starting with a huge homogeneous advertising market and and like and and then we started with the hard part. We started in Europe, we started on the the year before GDPR became law, like we started with like you know, this you know different jurisdiction with like different applications of GDPR. so yeah, a hundred percent.

Brett (49:14.163)

Yeah.

Brett (49:28.274)

It's a little privacy by design, right?

Mathieu Roche (49:31.509)

the reason why we started ID5 was you know probably to kind of make identity work better, but also to make identity compliant with with new regulations, right? There's like they're equally important to to what we do. so I guess you know it's it's easy to be when when you do the easy thing first, which is the US, where data has been accessible at greater scale and the market is so big and the budgets are so big and stuff that it's it's easier to make business in the US than it is outside of the US, right? When you do the easy the easy part first.

Brett (49:57.45)

Yeah, it's like the Wild West over here.

Mathieu Roche (50:00.229)

Then like it's harder to go to Europe because you're like, everything has been built for easy, right? We built everything for hard. We built everything for hard. And so for us, going to the US, going to Brazil, going to Japan, going to it's it's easy. It's easier. We don't we don't have we don't have we we've built for that for for that bar that's here that is like explicit consent, transparency by design, opt-out management, all of that is like has been baked into ID five capabilities since twenty seventeen.

Brett (50:02.226)

Yeah.

Brett (50:05.575)

Yeah. Yeah, that's a really

Rio Longacre (50:14.253)

Yeah, it's funny, had to see it.

Brett (50:27.496)

Yeah.

Rio Longacre (50:28.429)

We had a CF upload, remember that Brett, and he said that he compared Germany versus here. said, the US when he came, it was easy mode over here compared to like, he's, I didn't realize it was in a hard mode over there, but that is a good point you made about GDPR and how it's, that was the first big global privacy statute. It can be, can be pretty tough to navigate sometimes. And so I think, so your point is you were forced to comply with that and you weren't, so knowing you already were able to comply to that, different countries with different interpretations when you came here, it was easy, right? It was easier.

Mathieu Roche (50:48.952)

And and and a a big reason

Mathieu Roche (50:57.102)

Yeah, yeah. Well now we have to deal with the different the different states, but it's the same it's the same you know, it's the same problem. We've already solved it. I think a lot of people were operating in in Europe prior to twenty eighteen and then left. And that's why the market is so

American-centric. American companies have left Europe, right? Companies like Tapad that was acquired by Experian later or or Drawbridge that is now part of LinkedIn have left Europe. And so that's that's left a a big opportunity for people like us who are able to deal with with the European data laws. but it's also made the data market in the US very kind of self-centered in a way. And most of the big data companies are American companies that only operate in America, which is a problem for a lot of their clients who are global companies.

Brett (51:17.96)

Yeah.

Brett (51:34.046)

Yeah.

Brett (51:38.247)

Yeah. Would would you argue that's significantly more competitive in the US versus and and do you have have you seen you know true market share advantage in terms of like client penetration in other markets outside the US just because of the the the less competition or or or is that Yeah?

Mathieu Roche (51:54.561)

hundred percent. Hundred percent. We're pretty much the only the only show in town in Europe, right? So yeah, yeah, yeah, yeah. And and yeah, it's I mean d again, much easier to operate in the US. Access to data, right? Fli liquidity of data, data marketplace, like size of budget, everything's easier in the US in terms of on the data market. So there's a lot more competition because it's easier.

Brett (51:58.366)

Yeah, that's what I was thinking.

Rio Longacre (52:00.781)

You have it?

Brett (52:11.454)

Yeah.

Rio Longacre (52:15.095)

Yeah, think like having worked with many clients in Europe, we worked with lot of German publishers and big CPG brands. A lot of them had been very afraid even because of privacy laws and potential penalties of even storing data in the first place. So they actually had less of it. So mostly for his compliance. So yeah, that makes sense. So go ahead, Brett.

Mathieu Roche (52:26.255)

Yeah.

More more more for us.

Brett (52:32.849)

Yeah. S s good. No good.

Rio Longacre (52:36.557)

Okay, so maybe looking at true data, I it took place a little bit, a little while ago. You talked about it on, on, on, on, on, on, on, on, It was described as the Rosetta Stone. I'm curious, like, what did add to your capabilities that you didn't have? in terms of activating those and making them part of what you do, like, where are you in terms of integration right now?

Mathieu Roche (52:56.942)

Yeah. Yeah, yeah. So ID five was an ID, right? And that's the that's the currency that I mentioned in the transactions, right? But I also mentioned that on the sides of the transaction, right, you might have your IDs tied to an email, I might have my IDs tied to a cookie. And so if we want to talk to each other, we need a way to translate or to use something that we both have, right? And ID5 the ID5 ID was that. But the difficulty for our client was that they needed a way to connect it to the ID5 universe.

Brett (53:04.574)

Yeah.

Mathieu Roche (53:23.064)

And so what we bought with TrueData is a graph that is effectively that Rosetta Stone that I talked about, which is a multi-ID translation layer, right? Whatever ID you have on your side, we we're we're we can connect that at great scale with a great match rate to the ID5 ID, which becomes the biddable, the transactionable ID, right? And so we've really combined an ID with a graph. The ID on its own was great, but n had like integration challenges. The graph on its on its own was great.

Brett (53:29.735)

Yeah.

Brett (53:39.838)

Yeah.

Mathieu Roche (53:50.095)

But was just another graph on a very competitive market. Now we have a graph that has the ID5 ID, which is a unique asset, and we have an ID that is connectable to any source of data because the graph is kind of this multi-ID connectivity layer that gets us not only online data, but also offline data with the addition of emails as one of the components in the graph. So it's really a way to make the ID5 infrastructure a lot more interoperable, a lot more open.

Brett (53:53.341)

Yeah.

Brett (54:10.555)

And what and what made you buy versus

Brett (54:16.552)

Yeah.

Mathieu Roche (54:16.656)

to different operators on both ends of the spectrum, buyers and sellers, and to really augment our ability to convert offline data to online device to connect offline data to online devices.

Brett (54:26.333)

Y yeah, the ho the to so it's it's directly directly to household, right? It's digital IDs to to household. Yeah, and and so so was it because they had a ton of household specific data or was it more they had a a kind of a machine learning underpinning that that made that you didn't want to build yourself?

Mathieu Roche (54:30.478)

Yeah, device, user and household. Yeah, absolutely. Yeah.

Mathieu Roche (54:41.454)

Both, both, both. They had they had a graph that was already, you know, already built, that was already covering like most of the American population.

and they had the know-how, right? How do you buy data to be able to build the graph? Like what is good data, what is bad data, what are the models that you need to use to build that graph, right? And so we we we bought not only the the customer base, but also the product and the team and the experience expertise that came with the team and really merged that with the with the kind of the legacy kind of ID5 culture and product and value proposition of that transactional ID to to to really be be more efficient and easier to

Brett (54:54.376)

Yeah.

Mathieu Roche (55:19.804)

Yeah.

Brett (55:21.191)

Yeah and I I don't think

You know, people may not understand how difficult it and and what a challenge it's been for for decades. I mean at least back to you know the to the mid two thousands of connecting household data to individual data and resolving that in a in a predictable way. Like Nielsen struggled this. Nielsen bought Excelate, which had all this cookie data, increasingly made data, device level IDs, etc. And Nielsen had all this buy-side data which was household based on the buy-side of their business where they were tracking purchases and you know

you know, retail investments and and they just could never bring the two together because they were just completely different data formats, data types. It seems sounds like and and I used to talk to Mark X Mark Mark Zarosi about that, about how that is like the golden goose, right? If you can connect these two worlds

Mathieu Roche (56:08.024)

I think yeah, the b the big difference is that we are in people's devices. Again, back to my my my point earlier, like you've got a ton of data on server, but it might be super stale, right? And if you don't see how that data lives, you don't really know. Like we're in people's devices. We see five or six billion devices per day. Like connection to our servers, right? Five or six billion devices connect to us every day. So we we have our finger on the pulse of people's, you know.

Brett (56:17.127)

Yeah.

Brett (56:25.095)

Yeah.

Brett (56:30.502)

Yeah.

Mathieu Roche (56:35.694)

Posi behavior, right? In in we we we so we we don't know anything we don't know anything about them. We don't know anything about them. We don't know what they who they are, we don't know their name, we don't know we just know we're just able to recognize them over time. We're just able to s to follow to follow like the

Brett (56:37.542)

Yeah. In their ge in their geography, right, down to the latin latitude and longitude.

Mathieu Roche (56:51.162)

how the how they behave, right? And and and that's how our customers use our services to be able to leverage the data that they have on those people. So it's it's and but that that that link to the live kind of behavior of of of of consumers and people out there is is really what makes us different and what it gives us the ability to then model right belonging to a household or or or being a user of a certain type.

Brett (57:00.551)

Yeah.

Brett (57:16.22)

Yeah, and that's where that's a that's probabilistic, right? You're saying, you know, based on all of this data, including geodata, you're able to model in in predictively, you know, they're

Mathieu Roche (57:24.684)

I I I I I wanna be clear that we're not using any geodata in in Adify's model, right? There's no geodata. No no no no no. Yeah, yeah, no, no. No.

Brett (57:27.686)

there is no j data. Okay, okay. I j want to make sure that. Yeah, so you're not getting down to the lat l you know, but you're saying you've got a certain number of devices. H how do you actually align and does that what true data helped you do? How do you actually align all that device data and online cookie data, et cetera, to an individual household? Like what like how does that happen, probabilistically or or deterministically?

Mathieu Roche (57:47.461)

Well both. Both. the the the again, like there are there are you know, if if I'm I'm on an IP address here that is pretty stable, then I've got a computer, an iPad and a phone, right? And if I see those devices connected to the same IP address if within a very short period of time, I'm very likely to conclude that this is the same person and and and so or at least that those devices are in the same household. So that is

Brett (58:07.463)

Yeah.

Brett (58:12.017)

Yeah.

Mathieu Roche (58:13.046)

observed, inferred, like there are different methods to doing that. The question is like what's the confidence interval, right? It it's really 100%. Even if you have like you know two email addresses popping up on two different places, you're never a hundred percent sure that it's the same person because it could be, you know, my daughter using my email address to connect to Netflix or something. But but it gives you like a higher level of confidence that it's the same person. Similarly with other signals like IP addresses or cookies or maids or whatever. And so our job is to make sure that you know when we process that information

Brett (58:18.055)

Yep.

Mathieu Roche (58:41.262)

we have we have a a high enough level of confidence that that we can make those connections, we can perform that reconciliation, because yeah, the the the signals are are strong enough for us to do that.

Rio Longacre (58:54.539)

So look at our CTV you mentioned that just for a moment for the viewers, CTVs or connected television has been growing. It's been the biggest growth area the last couple of years. I think it's predicted to grow 18 or 19 percent this year. Right. I think we covered that previous episode. So I read that Future Today applies ID5 across more than a billion CTV impressions, including viewers who may or may not be logged in. So wondering, how do you like how do you establish a good

Brett (59:07.634)

Yep.

Rio Longacre (59:22.125)

confidence level that you're targeting the right people. just mentioned it could be different people and like how is the true data acquisition like because you have household data. that help you grow your CTV business thoughts on that Matt?

Mathieu Roche (59:36.437)

so our it our C T V business grew because I think the market is is is getting more mature in how identity and data play a role in in CTV advertising, monetization, measurement, et cetera, et cetera. the the the the true data kind of being a a a part of our value proposition, right? That acquisition being a part of value proposition has helped. I don't make any direct connection between the fact that

We have a graph and like C T V as a household gonna make like I I think that's a bit not that d yeah, it's a bit of stretch. but I think what really what really changed over the past probably twelve months is that C T V or streaming has become a lot more targeted and a lot more measurable.

Brett (01:00:05.606)

A stretch.

Brett (01:00:20.253)

Yeah.

Mathieu Roche (01:00:20.278)

And and and especially when you think about performance TV, which is a big trend, right? a lot of that requires outcome measurement, which rarely happens on the same device, right? Like you watch an ad on TV and you go buy something on on on a phone or on a website or or and so do do you

Rio Longacre (01:00:36.599)

How would you define Performance TV? Is Performance TV putting a QR code in there and tracking it back? it doing conversion API where you're actually like how?

Brett (01:00:41.468)

Like shopper marketing.

Mathieu Roche (01:00:43.926)

It's it's whatever whatever whatever whatever campaign that aims to drive an outcome. An an actual kind of business outcome, right? So a a sale, a registration, a subscription, right? That's and and and and

Rio Longacre (01:00:55.949)

There's more lower funnel or below the line.

Mathieu Roche (01:00:57.944)

Yeah, it's it's l it's lower lower funnel advertising. Yeah. Which is very new for T V, right? T V has been all about upper funnel, all about brand building awareness, building brand preference. And now there's a lot of advertising that dro that aims at driving performance, whatever the mechanism, whether it's a QR code or whatever. but

driving outcome is new for TV and and because the outcome happens on a different device, it requires that identity connection between the TV and the device where the outcome happens. And that's where we've seen a lot of a lot of traction, a lot of demand, alongside the ability to bring more data to the TV set itself so that you know more about the customers, right? I've I've been targeted on my TV on by a live ramp ad. Yeah, that

Brett (01:01:25.596)

Yeah, yeah. Yeah.

Brett (01:01:42.391)

really?

Mathieu Roche (01:01:44.932)

You could say it's good or bad at bad targeting. I don't know. But like I I don't think my neighbors who don't work in ad tech have seen live ramp ads on their TV, right? So

Brett (01:01:50.184)

And and was this outside the YouTube environment, or were you in the like a Netflix or like a Hulu or a. Okay.

Mathieu Roche (01:01:54.981)

Yeah yeah, I was on I was gonna say I was on Pluto. or yeah, Pluto or Ulu or something like that. Yeah, absolutely. so so you know, it's it's being able to recognize people and and address them with a message that we think is relevant for who they are is is always makes advertising better.

Brett (01:02:01.03)

Yeah.

Mathieu Roche (01:02:13.902)

Right. And and T V was kind of immune from that for a long time because you were like, Well, it's a it's a it's a football game, so we're gonna advertise to men and it's it's a it's a soap opera, so we're gonna advertise to women. not really, right? There's a lot more granularity that you can bring to this. That's what data brings, but like that's why you need identity, because you need to be able to do that at a granular enough level at a household user level.

Brett (01:02:14.149)

Yeah, but

Brett (01:02:25.062)

Yeah.

Brett (01:02:34.031)

Yeah, well and i and I think there's this promise with C T V and I think we we've kinda concluded Rio and I have talked about this a bunch, is that is that the C T V advertising ecosystem still kinda sucks, right? I j I think the user experience is pretty bad. Yeah, i frequency capping's a problem, right? They don't seem to they don't seem to be they don't seem to be selling all their inventory.

Rio Longacre (01:02:46.983)

It's still young, it's still early, yeah.

Terrible.

Mathieu Roche (01:02:51.907)

Identity.

Rio Longacre (01:02:53.741)

Yeah, and a lot of the, but yeah, 40 % goes unsold. think we covered it in a previous episode. So that's why you get it.

Brett (01:02:58.157)

Yeah, and so and so I don't I don't get anything targeted or relevant to me in C T V ecosystems outside of YouTube. Right? So I'm gonna put YouTube in the C T V when I'm on YouTube now that's a level and that's you know, you understand why they're they're good at that. But when it comes to other like the Hulus or the I I'm just finding it a much worse experience than what linear TV delivered from an advertising perspective. Even though there's all this promise that we've got, you know, accuracy and data and i it et cetera, et cetera. I'm just you and maybe that's just

Mathieu Roche (01:03:27.022)

Yeah, it's it's it's disappointing that it's not incremental. Like if anything, it should be at least as good as as as linear because you have all of the same information about the show and the the you know the the population that engages with the show from panels. So so it should be at least as good and then everything else should be incremental, right? Your ability to overlay data, your ability to, you know, better better frequency cap. but I agree with you. I think it's still a new it's still a new world, it's still a new mechanism.

Brett (01:03:33.627)

Yeah.

Brett (01:03:39.193)

At the very least, right?

Mathieu Roche (01:03:53.161)

and and and it's it's a world that needs to learn. And same thing as the you know the innovative dilemma that that we had on the on the web with like newspapers and everything twenty years ago is still very, very much alive here in T V as well, whereby, you know, you gotta you gotta trade your your linear revenue for streaming revenue. It's it's a brave new world and it's not easy to compete with yourself. So I I I give them a lot of a lot of credit for you know

Brett (01:04:01.177)

Yeah, yeah.

Brett (01:04:11.398)

Yeah.

Mathieu Roche (01:04:19.172)

taking the taking the steps toward doing that, right? You see Wanabros, you see LG, you see people like that are really investing in their ability to do that. but it you can't you can't you can't build Rome in one day.

Brett (01:04:26.661)

Yeah. It's an inevitability, right? It vi Yeah, it's an inevitability and you just hope that that the ecosystem that rolls out for video and in C T V in general is just a cleaner, better user experience than what happened, I think, in the ch in the traditional digital ecosystem, right?

Rio Longacre (01:04:44.301)

Well, it's getting better. I mean, it's definitely improving like the ad experience. Certainly. I mean, it's not great, right? But it's certainly gotten better in the last couple of years. And funny, I just cut the cord finally. So we had cable up until like a few months ago. We finally cut the cord and I don't, I don't watch any, any, any linear television anymore. Yeah. It actually has helped. It's funny. Yeah. it has improved since then. We also upgraded our package to, Comcast was not easy to deal with doing that,

Brett (01:04:59.237)

That's why your internet's better. You bit y

Rio Longacre (01:05:10.583)

So Matthew, the title of this was going to be who replaces Switzerland. And I think you're like, so the, and the, the assumption was that because of the acquisition live ramp probably can't be Switzerland anymore. I know that they're saying they can be, but that's certainly debatable to say the least. So I think your proposition is ID5 to replace Switzerland. So let's say you're talking to an organization that wants to evaluate their, their, their ID and identity resolution strategy. What would be your advice to them to start this process?

Mathieu Roche (01:05:40.461)

Understand your use cases, all right. Understand why identity matters to you because again it's not it's not as easy as as as to say like this is the use case, this is this is the the feature that I'm buying, right? Like you're buying a you're you're buy you're buying a car to take you from A to B. That's an easy that's an easy use case, right? Identity is the wheels. The car doesn't doesn't run, doesn't you know drive without wheels. Like it it cannot go, they cannot take you from A to B. But you're not buying wheels, you're buying a car. It comes with wheels, right?

Brett (01:06:10.022)

Yeah.

Mathieu Roche (01:06:10.52)

And so understanding like where do I need wheels? What use case do I need wheels for, right? Understanding that and then understanding what is the characteristics of the wheels that are important for me, right? Do I just want to go with whatever wheels are by default on the car? Or do I have like because I've got different cars and I need I it's a stupid analogy. I need a one set of wheels for all my cars, right? They need to be like of of of a certain a certain type. so it's understanding the use cases that are supported by identity.

And then understanding what makes identity valuable to support those use cases, and then who can deliver that valuable identity spine. So I think that's really like starting from the use cases, starting from the applications. Are you targeting? Are you measuring? Are you frequency capping? Are you onboarding data? Are you sharing data with others? Like why what are you doing that is important for your marketing strategy that relies on an identity spine? Once you know that, once you understand that, you understand what you need from an identity spine standpoint.

Brett (01:07:03.824)

Yeah.

Brett (01:07:08.015)

Yeah, and that and that's gonna help if a brand has to make a decision, if they want to make a decision, to move off of LiveRamp in X amount of time, twelve months, that's the type of analysis that they're gonna they they should they should understand their own sort of data house and and and their primary use cases before they they sort of switch. But so what do you think a credible alternative looks like? I mean, is it a is it a composable stack with somebody like an ID five as like your identity translator? I mean, what do you what do you

Mathieu Roche (01:07:30.488)

Yeah, I I don't think again, r r right now right now there is no other kinda country you can move to, right, that is the equivalent of Switzerland if Live Ramp is Switzerland, right? There is potentially a coalition of countries, right?

that that you can move to and that coalition of countries will have some obviously identity kind of underlying identity spine right an ID and a graph which ID5 is is very well positioned to be but then you also need data capabilities, data management capabilities, data sharing, data enrichment, right? You need data monetization capabilities if you're if you're using data store currently with LiveRamp. You need connectivity into different destinations. Are you doing programmatic advertising? Are you advertising on on World Gardens, right? So if you if you're able to kind of

Brett (01:07:45.382)

Yeah.

Mathieu Roche (01:08:12.452)

You know, build rebuild that puzzle, right? The the the different bricks or the different pieces of the puzzle, you can find. They all exist in the market. There's a ton of data management or company who offer data management capabilities. There's a bunch of of kind of data marketplaces out there that you can use to monetize your data, building pipes or or or using systems that have pipes into the destination or the the places where you want to send your data. There's a lot of those, right? There isn't a lot of like identity.

Rio Longacre (01:08:37.933)

Yeah, there are companies that build those connectors. Yeah, for sure. You're right, those exist.

Mathieu Roche (01:08:40.142)

Yeah, exactly. There's not a lot of identity kind of underlying I IDs and graphs, right? That that's where I think ID five has a has a very good role to play. But but on top of this, right, again there there there is a lot of of data management and data activation solutions out there. and so I think the solution will be different for different types of clients, different agencies, different brands, but like they will rebuild their path to activating data or to monetizing data using those.

pieces that exist. And then we'll see. The market will will will work its way and like those PCs might consolidate. Those PCs might might kind of come closer together. You might have like you know relationships like we have with Optable or Permutive, right? Preferred pre pre-built, right, native integrations that just make it easy. But I think that's how that's how the market is going to reorganize and and restructure after the LiveRamp acquisition.

Brett (01:09:24.933)

Yep.

Brett (01:09:31.61)

Yeah. So

Rio Longacre (01:09:32.629)

So Brett, just like you said, more like composable or actually modular. I think that's a better approach where you piece together different, like kind of best in breed solutions, depending on which ones you need, depending on your existing tech stack, you know, for data management, maybe you have like a snowflake or data bricks, right? I mean, if you're, if you're collecting cookie data, maybe you need, maybe you need like a permutative, right? Especially if you're a publisher, right? If you're, you know, for activation, maybe you could maybe use a high touch or something, you know, one of our sponsors to the show.

Brett (01:09:37.026)

Modular's a better word, I think. Yeah.

Rio Longacre (01:10:01.165)

you for all the connecting pipes, maybe use like a MedConnect, right? Cause you know, there's lots of people building up pieces of this, this mat. So I really liked that. And you piece that together and you, and you, and if you want to switch something, you swap it out just a little easier. it's, it's, you know, I, I, I'd be hard to disagree. That's the direction people are going.

Mathieu Roche (01:10:14.804)

On on a on a on a use case on a use case by use case basis, because again you cannot just like take everything you're doing with LiveRamp and put it here and it works, right? It's like you gotta understand like why am I using that that that company, those applications for, what are the benefits I'm getting from them and how can I recreate those benefits with other players.

Brett (01:10:31.524)

Yeah.

Rio Longacre (01:10:31.681)

What percentage before we I know, but you probably want to go to quick hits, but like what like what percentage I mean, are you willing to make a guess at what percentage of live rounds business based on what you're hearing or just your intuition? Do you think is potentially at risk?

Mathieu Roche (01:10:43.92)

I think up to fifty percent of their business because there there's a lot of people who don't want to depend on on on either a measurement company that belongs to their agency

Or a measurement company that belongs to another agency that has like a vested interest in making their business look better than it is. exactly, grading your mark. And then a lot of people are just like, yeah, they have a problem with working with other, you know, company that belongs to an agency that also serves other clients that might be competing. There's a lot of, you know, a lot of people change agency because they don't want to be serviced by this the the agency that serves their core competitor, right? It's very unlikely that.

Brett (01:10:58.448)

Creating homework? Yeah.

Mathieu Roche (01:11:17.348)

Pepsi just went to Publicist, congratulations to them. It's very unlikely that Coke is gonna go to Publicist too, right? Coke and Pepsi very unlikely to be in the same agency, right? Is is is you know, Coke, if they go to WPP or Omnicum, gonna still want to use LiveRam be belonging to Publicist who now it represents the the Pepsi accounts. I don't know. That that that I'm not saying they're doing anything bad, I'm just saying like it creates the the the it the doubt that

Brett (01:11:22.363)

Yeah.

Rio Longacre (01:11:22.403)

yeah, they are there.

Brett (01:11:45.029)

Yeah.

Mathieu Roche (01:11:46.158)

And it's data. It's really, really important asset for a lot of brands, right? Like do you wanna do you wanna do you wanna yes.

Brett (01:11:49.498)

Yeah, it it adds a lot of risk. Maybe they won't be bad actors and take advantage of it, but it does add the potential for this to happen.

Rio Longacre (01:11:55.469)

Yeah, well, I think ultimately agencies and consultancies like they do, they're supposed to do what's best for the client. So if the client says we're going to use LiveRamp or anyone else, they should say, okay, yes, that's, we'll help you do that. But the counterpoint to that is, I mean, you're right. Like, let's say if you're, you know, let's say, okay, on the Coke, Coke, you know, the, on the Pepsi account, right? For example, do you think the, the ad sellers, right? They're not going to want to

work with the technology that their competitors using, So like, so you think about it, like the people on the agency side, right? So let's say you're implementing, let's say your media team is working with a company, you're not going to want, if you can prevent it, you're going to lobby them to use, not to use a competitor solution. Cause you know, and I'm sorry, like publicist is already including live ramp in their pictures. I'd be shocked if they don't actually rip through the client list the day that deal is done to start calling everyone and saying, well,

will give you a special deal. mean, I'm not saying I know they're do that, but I'd be shocked if they didn't, right? Because that's how, I mean, when you acquire a company, it's supposed to be this, you know, that's supposed to be more than the sum of the parts, right? And part of that is unlocking like the value you have. So we'll see. know, 50 % is a high number, but that's not crazy to me. think, you short term we could see 15, 20 % right away. But time will tell.

Brett (01:12:57.262)

Yeah.

Brett (01:13:18.202)

But as we said, nobody's gonna know because unless we hear like client proclamations because this data is all it's a publicly traded company. No, well they are they are publicly traded, but they're not gonna share that data around LiveRamp, are they? In terms of revenue number.

Rio Longacre (01:13:29.78)

Probably not. I doubt it. Yeah.

Brett (01:13:31.979)

cool. So let's let's move on to quick hits. I think we're we've we've made it well past an hour and thank you. this has been awesome. you know, appreciate appreciate your candor. It's been a it's been a fun conversation. And welcome to New York City. I know it's a year and a half. we'll we'll talk sometime over a beer to find out yeah how you compare that to London and to Paris and other places. because we we do like to talk about that. It's I live here for a reason, but I do love London and Paris. so quick hits. one out

Rio Longacre (01:13:40.055)

This has been great.

Brett (01:14:01.832)

advertising metric you'd remove from every dashboard tomorrow.

Mathieu Roche (01:14:08.056)

Advertising metric of click through rate? Yeah. Yeah. Aaron, that's that's an easy answer. This is the one that I think we've we've been we've been dependent from from way too long and that doesn't make any sense.

Brett (01:14:12.6)

Well I'll click the rate. Yeah, yeah, of course. Of course.

Rio Longacre (01:14:22.677)

What's one misconception about identity that pops up again and again and again that you just want to squash?

Mathieu Roche (01:14:29.454)

Identity and data are the same thing. They're not. It's very different.

Brett (01:14:38.66)

So the most overhyped identifier in the market right now, like in terms of, you know, specific A D type.

Mathieu Roche (01:14:50.916)

Email.

Brett (01:14:52.575)

that that's interesting. Why?

Mathieu Roche (01:14:55.556)

Because everybody thinks it's the gold it's the the the the gold standard. it is it is it is a lot less

Brett (01:14:59.332)

Yeah, the hashed email, right?

Mathieu Roche (01:15:07.32)

It is a lot less valuable than people think from a r an identity reconciliation standpoint.

and it it only applies to people you already know, right? Like by definition, if you don't have an email because they're not your clients, well that then they're you know they might there might still be a prospect. And if you only reach people that you know the email for, you're only reaching your c customer base. there's a lot of there's a lot of fake email, right? Batman at gmail.com is a is an address that we see very, very often. It's definitely not the same person, and I don't think it's Batman actually at all. I don't think he's he's gonna sign up with his his real email.

Brett (01:15:31.91)

Yeah.

Mathieu Roche (01:15:39.125)

so it's just like it's it you know there is this this and it ties to the notion that deterministic and probabilistic, right? Like people think email is the best, and like if I don't have email, I'm gonna do other things. And it really depends on the use case. So like those kind of absolute like statements like probabilistic prop you know deterministic is better than probabilistic or email is better than cookies or or IP address, whatever. It it all depends and people do like massive shortcuts. It does it I I can I can give you twelve different counterexamples of why.

Brett (01:15:59.579)

Tall relative.

Mathieu Roche (01:16:06.244)

You shouldn't use email for certain applications, or why probabilistic modeling is better than deterministic recognition. Those those are the shortcuts that that that people make in a very difficult technical I I I totally understand. and and I'd like to I'd like to crush that.

Brett (01:16:23.213)

Ha ha ha.

Rio Longacre (01:16:23.957)

All right, last one here. Does agentic trading make identity more important or less relevant?

Mathieu Roche (01:16:32.908)

you you you don't expect me to say to say less relevant, obviously. I think listen, no no I think no no but yeah the the why is important, right? I think a Gentec is gonna make everything programmatic, meaning everything user user level, right? Like right now there's a big part of advertising that is not programmatic, that is still served on a bulk basis, right? Because because we can't.

Rio Longacre (01:16:36.493)

I think I know what you're gonna say, I'd love to hear why.

Brett (01:16:37.082)

Ha ha

Brett (01:16:50.297)

Yeah.

Mathieu Roche (01:16:52.824)

Right? I think agenti gives you the ability to change your your purchase order every minute if you want to, right? And so you can say, I want to serve this out, I don't want to serve this out. It's almost like it's not it's not RTB, right? It's not real time, but it's it's getting closer to real time. We need which means you need you can change the rules of of definition, which means you can get closer to being programmatic, at least like not serve the same not not apply the same ad serving principle to everybody for a week, but have

Brett (01:16:57.894)

Yeah.

Mathieu Roche (01:17:21.604)

Daily or hourly rules, right? And so, and that's gonna make identity more important for everything, not just pro what is programmatic today, which is maybe 40 or 50% of the world, but the other 60 or 50% will also become more user-driven because we can. Because with agentic transaction, agentic buying and agentic planning, we can. So it a greater degree of granularity, yes.

Brett (01:17:24.037)

Yep.

Brett (01:17:36.794)

Yeah.

Brett (01:17:41.402)

Yeah, it's it's a much greater degree of relevance because there's th it's more recently updated. Yeah, yeah, for of granularity and then you can also align the creative

variations because that now can be scaled and automated from a creative generation perspective because that was also a blocker in the past, right? You just couldn't, you know, brands didn't have a thousand variations of ad creative, right? Outside of the gaming industry maybe or the or the mobile app industry. But yeah, this has been awesome. Hey, thanks for thanks for joining us. Thanks for spending the time. We're we're we're putting you in the friend of the pod category now because you're you're a re you're a repeat guest

Rio Longacre (01:18:10.657)

It great. It was fun.

Mathieu Roche (01:18:17.134)

Happy very happy too. I I I I'll I'll wear that badge proudly.

Brett (01:18:21.188)

Yeah, yeah, and yeah, I g can't wait to meet you again in person. we'll probably see each other in New York City soon enough. And for everybody that's made it this far, visit us at debdebdeb.signal and noise.ai. Yeah. we are growing. and you can find us on b basically every platform, both social media and anywhere you listen to your podcasts, and read our articles and subscribe to our newsletter. And we will see you next time. Thanks everybody.

Mathieu Roche (01:18:30.768)

Congratulations.