← Marketecture: Get Smart. Fast.
Marketecture: Get Smart. Fast.
Episode 147: Mathieu Roche on the TrueData Acquisition and the Modern Market for Identity
00:59:46
Loading summary
Transcript207 lines
- [00:01]
A
This podcast is brought to you by Maloco. If you're an app marketer concentrating most of your budget in Google and Meta, you're not alone. 88% of consumer app ad spend stays inside walled gardens. Meanwhile, there are billions of daily users spending their time elsewhere in gaming apps, dating apps, productivity tools, you name it. That's where Maloco comes in. Maloco's AI powered platform connects you to those users, reaching 2 billion daily users across 3 million apps. Ready to diversify. Visit Malocoads.com and start scaling with confidence again. Go to Malocoads.com to learn more. This podcast is brought to you by Adelaide Media. Verification and measurement are undergoing major disruption. Legacy players are pivoting to performance. Advertising AI is reshaping, brand safety and attention is replacing viewability. Adelaide is leading the shift with au, a new way to assess media quality that scores placements based on their potential to drive attention and outcomes. Before your ads run, think of it like a credit score for media. Finally, a clear view of quality. Before you buy, take the guesswork out of your investment strategy and try Adelaide AU on your next campaign. Welcome to the Market Tech podcast. This is Ari Paparo. I'm here today with Eric Franchi and our special guest, Matthew Roche from ID5. So, Eric, you've known Matthew for a long time, right?
- [01:34]
B
Yeah, my fund, Perim has been an investor in the business since 2019. So I've been with Matthew on the entire journey and they had a big announcement this week. They acquired another company we had a small investment in called True Data and together they've built, I think what is a very interesting and modern approach to identity. So we're going to hear all about it.
- [01:55]
A
Yeah, I love these private to private deals. There's like a trend going on. There was the Rembrandt deal. Sign of the times. Yeah. Um, it's a, it's good, it's a good way to bulk up as, as, especially if you can figure out a, a way to get over the concerns around valuation and stuff like that.
- [02:11]
B
Yeah, absolutely. So this would be a fun one to talk about. Matthew is, is just, I mean he looks at everything from the perspective of identity. So even like, you know, we'll get into the news getting his take on things like earnings and AI and how it relates to identity, I think is, is a real interesting dimension that the user, the listener, is going to enjoy.
- [02:33]
A
Yeah. So yesterday I spoke at the pre bid summit with Joe Zulawski, your partner. So they do this thing second year in a row where we call it between two ferns.
- [02:44]
C
Oh, the Bid switch Summit.
- [02:45]
A
The bid switch. Yeah. What did I say?
- [02:46]
B
Yeah, you said pre bid.
- [02:48]
A
Oh, I'm sorry. Pre bid summit was a couple weeks ago. Yeah. So Barry and the team from Bidswitch had us on, and the conceit was that they give us the questions to ask each other that we've never seen before. So that was kind of fun. And we talked about your podcast. You do three podcasts now. Why don't you plug your. You do this podcast, Open Market and the Imperium podcast.
- [03:13]
B
No, I let Joe and Corey have the Imperium podcast. So I'm on too. So you guys were making fun of me.
- [03:19]
A
No, I was just wondering because I think in this podcast, we have a good dynamic where you're the straight man and I'm the unhinged one. And I was just wondering how it works on the Aperium podcast.
- [03:30]
B
Those guys do it.
- [03:32]
A
I'm not totally straight.
- [03:35]
B
I have, like, overexposure already. I feel like with two pods. So it's two straight men.
- [03:42]
A
Right. Okay. But if you're not getting enough Eric content, you could go to Open Market and listen to that.
- [03:48]
B
Yeah. On that one, I have Joe Zappa, who, similarly, I'm a straight man, and he is unhinged at times.
- [03:54]
A
All right, that's what you need. That's how you get the ratings in the podcast game. All right, let's jump in with Matthew Roche from ID5 talking about the. The acquisition of True Data and the whole identity market. We'll be back in a second. All right, welcome. Matthew Roche, the CEO of ID5. Pretty exciting week for you. How. How's it feeling?
- [04:15]
C
It feels. It feels like a relief. Right. I. I posted about this on LinkedIn that you're. You're schizophrenic for, like, four months when you're going through a deal because in your head, you're already in the post acquisition. Right. Like, you know what you want to achieve. The organization, the pitch to clients, but you have to keep it super confidential. There's only like six or seven people on the team who knew about the loi. Due diligence and everything. And so you're like. You're living in two worlds.
- [04:38]
A
Right.
- [04:38]
C
Which is terrible. It's just like, I'm happy to be able to finally come out and tell the world how excited we are about the acquisition of True Data.
- [04:48]
A
Yeah. So before we get further disclosures, I'm an investor in ID5. Eric's investor. You're an investor, right?
- [04:55]
B
Eric appear in my fund as an investor in ID5.
- [04:58]
A
Okay, so we're talking up our book here. ID5 is the best company on earth. Let's start with the facts. What happened? You acquired true data. What is that?
- [05:08]
C
That's right. So ID5 is an identity company, but very focused on IDs, right. Our job is to make devices addressable, recognizable over time, and true data. And we've kind of with, we've played with identity graphs a bit, but it's not where we come from. Right. TrueData is an identity graph company. Their job is to connect all sorts of data at a user or sort of devices together at a user and a household level. And to be the Rosetta Stone. So that if you have an email or a cookie or a maid or something, they can stitch it together and tell you this is this user, this household. Right? And those are the other IDs that this user and this household belong to. And so we're bringing together basically a graph and an id. That's the pitch behind the acquisition.
- [05:55]
A
If you had asked me last week, I would have said you already did that, so maybe I'm a bad investor. So how did it work? Before the deal, ID5 would resolve each of those devices to its own ID that was consistent, but it wouldn't stitch them together. Is that the difference?
- [06:11]
C
No. So when you think about how clients use us, right, clients use us to recognize devices. And most of the big clients that we work with, they add rid to their graph, right? And now we've got like an end to end proposition. They don't have to go source the graph elsewhere or to build their own graph themselves. We see more devices than probably most people in the space. And so we have more signals. We're able to build better graphs and connect those graphs with the ID that sits on devices that nobody else is able to recognize because there's no cookies in Safari, there's no maid, and like some, you know, iOS devices that don't allow that, you know, so those are the type of things we can bring together so that basically we can increase match rates between the data that people have about their customers and the devices that they want to reach those customers through.
- [06:57]
A
Right? So your customers are whom they're other ad tech companies.
- [07:02]
C
So a lot of ad tech companies, a lot of publishers, and increasingly, especially with this acquisition, more and more brands and agencies. Identity is not a feature, right? It's not something you buy, something you use to get to a point to deliver something. Right. Whether it's targeting or optimization, frequency capping, measurement. Right. All of that needs the ability to recognize people and to overlay data on, on those, on those, the devices that you recognize. And so we kind of power a lot of those capabilities, but in the background. And a lot of those capabilities are leveraged by your DSPs and your SSPs, your CDPs, your clean rooms, your publisher, your publisher text. So it's really powering a big part of the ecosystem, but kind of in the background.
- [07:47]
A
So from a customer perspective, they probably already were taking your IDs and doing some matching with a graph of some kind. So potentially if a customer wants to use the true data graph, they're being able to reduce their vendor count and their cost by one.
- [08:03]
C
Well, we're, no, we're going to call that upsell. But yeah, there is, there is certainly economies of scale here, but it's mostly, it's also like, you know, the more you stitch the finished product, the less kind of stitching capabilities you have because a lot of the original signals are lost in the finished product. Right. And so we're elevating that so that we're able to make that stitching at the most granular level, which means the scale that we're going to be able to deliver with the graph and the ID combined will be really unique.
- [08:32]
A
And is true data a deterministic solution? Is that a way to think about it? So, and ID5 was historically a probabilistic solution.
- [08:41]
C
That's, that's an age old debate. The ID5ID is stored in the device. So in this sense it is very deterministic. Right. When the device calls us, we read the ID and we respond to the id. However, there is a lot of algorithm happening in the background to be able to recognize that this device and that device are the same one when you know, they are connected on different IP addresses in the morning or the afternoon or where they are, you know, you've got different browsers or you're like you're in apps one day and you're in a browser the other day. Those are the type of algorithmic processing that we do in the background. But there is definitely a deterministic foundation with the true data graph that we're able to expand probabilistically. Now with this acquisition.
- [09:25]
A
What's the current state of that debate? The deterministic probabilistic debate? First of all, is it a debate? Second of all, are customers or platforms choosing one side or the other and, or is everyone doing everything?
- [09:40]
C
I think it is still a debate because we love a debate and we love to kind of throw like, you know, complicated words at each other. I think when you, when you look at it from 30,000ft, what matters is am I able to recognize enough people, right, to kind of be able to run my campaigns and I'm able to use that recognition capability to target, to optimize and to measure. Right. And so the methods that are underneath, as long as they're legal, right. And they drive efficiencies, it's almost irrelevant. Some people say probabilistic is better because it gives you more scale. Deterministic is better because it gives you more precision. In the end, it's always a trade off. Nobody cares. We're an audience of one. But if you're going to give the same idea to everybody because your algorithm sucks, nobody cares either. So it's a matter of how do you find the right, you know, equilibrium between scale and precision? And this is exactly how we train our algorithms, right, is that, that's, you know, precision and recall trade off.
- [10:37]
A
And how do you.
- [10:38]
C
But what matters is that, is that it works. And, and all of our clients benchmark us against some form of two sets, right? Some form of like, you know, thing they consider being the truth. And, and if we're not good enough, we're out the door. So that's a very clear, there's a very clear measurement.
- [10:55]
A
Do customers do a test generally before they onboard you and what's the truth set, like, you know, known users with emails and logins or something like that?
- [11:04]
C
It depends. You can use third party cookies, you can use logged in users, you can use customer databases, you can use another graph you can use. Yeah, it depends what your, what your reference is. But yeah, there is always, I mean, most of our contracts start with a POC and then convert into like a long term relationship once the POC has been successful.
- [11:24]
A
So how does this position you against Liveram? Are you head to head with them?
- [11:29]
C
I mean, that would be a bit presumptuous for me to say that Liveram's built an amazing business over the past 20 years or so and it goes above and beyond identity. Right. They operate as a data exchange, they operate as a clean rune. But we think that the world has evolved a lot over the past 20 years and that we have a very strong value proposition when it comes to recognizing people on a global scale. First, I don't think Liveramp operates necessarily globally. We have like, you know, a very strong footprint in North America, across Europe, right. All of the EU and the uk. We also operate in Australia. We have a graph in Japan we have, you know, we're launching in latam. So we really have this global scale and we have, you know, the benefit of adding an ID that exists in more devices than any other id. Right. And so I think in the end it's a question of match rates and we've been very focused on optimizing the ability to match offline data to online devices. And I think we have a very strong value proposition to marketers and publishers on that front.
- [12:26]
A
So the leading IDs out there, we did a top 10 IDs like that was might have been more than a year ago with Michael Sullivan.
- [12:33]
B
It was a while ago.
- [12:34]
A
It was, yeah. But I think if I remember correctly, it was like probabilistic IDs were the leading ones were, you know, ID5 was, was definitely when we called a probabilistic was definitely up there. And then deterministic was ramp ID UID and maybe load me, I forget what they call it. They have an ID name or something like that. Is that still the kind of the, the shape of the market?
- [12:58]
C
I mean, you know, Mike, Mike and Sincere have been our, almost our best, our best marketing platform. Right. Because they've shown that ID5 was consistently the most adopted ID by publishers across their, whatever, 200,000 websites that they track. Yes, I think LiveRamp UID2. Right. Load and you was up there. There's a very small group of IDs that have really scaled to a level where they become a usable transaction id. Not to be confused with the TID like occurrence in the transaction. Right. Because they have scale, it's like a currency. Right. If you're the only one using rubles, well, nobody cares. Nobody's going to use Wobbles because they can only trade with one. So there's a really strong network effect to what we do. The more publishers adopt us and decorate their bids with ID5ID, the more valuable it becomes to the buyers. We announced that Spotify, for instance, is integrated with ID5. So now you can buy Spotify inventory using ID5 IDs. Right. Announced the Guardian a few months ago. Like there's a ton of very, very valuable inventory that can be recognized and bought using ID5 IDs. And that drives a lot of demand. Right. And so then, you know, big DSPs and like retargeting platforms and data companies all of a sudden are interested. And so this, this DYN really accelerated by adoption. And that's why you have a very small number of IDs that have, that have some level of adoption because Again, there's a natural kind of concentration on the top three, four, five or whatever.
- [14:23]
A
Right. It's impossible to start in the first.
- [14:25]
C
One for a long time.
- [14:26]
A
Hard to start a new id. No one wants to start a new id. So tell us about the deal, like how many people came over, how tell if you. To the extent. We'd love details about the negotiation process without however you want to sanitize it because, you know.
- [14:41]
C
And nobody got killed as a result of the transaction. Nobody got armed. So we can tell you about the deal. A lot of it is confidential, as you can imagine, but it's an increase of about 30 to 40% of IB5 in terms of like staff and revenue. Okay, so that's a significant acquisition. I mean, we're not very big company. We're 60 before the acquisition, we're 80 now, or a bit over 80. And so it's a pretty significant endeavor to go on the process of, well, first discovering the opportunity and starting discussions and then getting the board to support the idea and then issuing an LOI and then going to due diligence process, which was the better part of summer, like July. We started in July until the end of September. Holy October. Then the closing process. Right. With all of the. I think at some point we are four or five law firms working on the deal. Right.
- [15:34]
A
Sounds expensive.
- [15:38]
C
We're negotiating really hard. But no, but there's a ton of things to do. Right. There's a ton of things you want to check, especially in our space. Right. Identity data. Like it's sensitive. It's sensitive. And so we had to look at a lot of things through the due diligence process. And also it's a transatlantic deal. Right. True Data is a US company, ID5 is a European company and we hire people everywhere. I think by now we're over 20 countries. So there's a lot of labor, tax, IT, financial, all of those things. So it's certainly been an amazing process for us. We've learned a ton. I was happy to see. I don't know if I was happy, but the press. Well, Alison Schiff and Ad Exchange recovered the deal and she said their first ever acquisition. As if there's going to be a lot more. I don't know about that, but we certainly learned a lot. And the exact team, the 85 exec team and the True Data exec team has been super involved and super helpful in going through this process. And it's been, all things considered, quite seamless. I think. So, yeah, it's been a great experience.
- [16:56]
A
As A French CEO. You must feel great to have American employees. Like, you can fire them, you can make them work on weekends, you can take away their health care. I mean really, you can do anything you want, right?
- [17:08]
C
Yeah. So you know what we've done? I'm going to pitch our HR policy for a bit. We've normalized everyone, right. So everybody's got the same benefits in terms of healthcare, in terms of holidays, in terms of severance policy. Oh, wow. I think our American employees are very happy because they get the best of the European benefit package. But yeah, the benefit is really to have this kind of multicultural international presence. It's just there's a ton of. It's not a political statement, but diversity is good, right? As a company, when you want to be innovative, when you want to find like, you want to disrupt markets, you want to find new ways to do things. I think people with different backgrounds, different perspectives, different point of views, different experiences, different ways of doing things. Right. We've got like German engineers and Italian marketing persons and American finance and all of that comes together and it creates a ton of innovation, new ideas, new perspective. So it's, it's. And also we can, we can hire from pretty much anywhere in the world, which is a, which is a benefit. It's, it's, it's more creative out for us.
- [18:32]
B
In all seriousness, Matt, was the fact that this was a US based company that much more attractive for you from a strategy standpoint? Because the U.S. you know, by far largest market, most of your customers are here having all of a sudden now, 20 more employees, largely in the U.S. has got to mean good things for you from just a coverage standpoint.
- [18:55]
C
Yeah. American, American presence is obviously very important, as you said, because it's already our biggest market. Right. I think ID5's revenue is 60, 60 to 65% coming from US plants. But being more present, having more people on the ground. Right. We have a new CRO, John Durkee, based in New York, that helps with my time personally, but also just like establishing our credibility and our presence in the U.S. it's also interesting because the U.S. market is quite open when it comes to identity. There's a lot of identity vendors in the US Very, very few outside of the US and we have this history of being able to operate in Europe and Asia. And so for us, buying a US assets is great for our US presence, but it also gives us a lot of experience and relationships and know how that we can then expand internationally. Right. And so we're going to, we're going to, we're going to expand what TrueData has been doing in the US in Europe, in Asia. We're going to expand our graph over there with what we're learning from the True Data team. And that's net positive, right? That's like, that's clear upside to how we, how we operate building.
- [20:00]
A
So now it's integration time. Right. So a little bird bee told me that you had your off site like in the middle of nowhere in Rockaway, Brooklyn. Is that true?
- [20:11]
C
So we had a. We. So we do a lot of get together for a remote company. We think it's important. So we had a full company off site which was just before the acquisition that was in Mallorca, Spain. So.
- [20:21]
A
Oh, that sounds nice. That sounds a lot nicer than Rockaway.
- [20:25]
C
But then I brought the executive team together to Rockaway because we were not there for the food or the weather. We were there to work earlier this week. So yes, Rockaway Beach November 3rd and 4th.
- [20:38]
A
Oh my God.
- [20:39]
C
What we treated the exec team to. And you know what? We had two amazing.
- [20:43]
A
Did you like get Russian food or something?
- [20:46]
C
So the food options in Rockaway are quite limited to be honest. But the Rockaway Hotel, right. Free plug here. I'm not paid, I didn't get a discount. But it was a really good destination for. And the best benefit is 20 minutes from JFK. He had. Out of the nine people on the exec team, seven of them were flying. So proximity to the airport was critical.
- [21:05]
A
You could take the AirTrain right to Howard beach and just walk there. It's like a magical part of New York that most people don't see. So yeah, so I guess what's next? So obviously you're trying to integrate these products, sell them together but from a customer's point of view, do they know what to make of this? Is it obvious? Just check the box and get the combined product or is there work and what's the transition going to be like for customers? If you're assuming you're going to upsell them on that?
- [21:36]
C
Yeah. So I mean there's actually in the market, I think there's a better understanding of what an identity graph is than what an ID is. Right. I think historically the market for identity graphs has been there. Right. You have people like Drawbridge back in the days or Tapad Rap prior to the acquisition they kind of pioneered the notion of identity graph. So it's not a new thing. There is a market for that. There is an established set of vendors. Most people have been building Their graph on cookies and maids and emails. Right. And so the legacy graphs kind of have a lot less reach because those signals are less available. So what we're going to do is actually augment the graph with all of the benefits of having an ID like ID 5, which is available everywhere across all the devices in all the countries. But we're going into a market that is actually more established and a bit better known, if you wish. So I think it will be a very logical step for our clients. A lot of our clients are actually, you know, we have a lot of clients in common through data and ID5, so that's going to be an easy kind of tie in. And, and, and in the end, the, the, the key success factor is the metrics, right? And by bringing the best idea in the market with one of the best graphs in the market together, you just augment. You just, you know, increase the match rates you can provide, and that's the only thing that matters, right? People want to be able to reach their customers at scale, want to be able to apply data at scale, want to be able to recognize as many devices as possible at scale. And so that's what we're going to deliver with better match rates than everyone else in the market.
- [23:06]
A
Any thoughts on the company name? You're keeping an ID 5. Switch to true Data new name. Throw an X in there somewhere.
- [23:13]
C
It's ID5. It will be ID5. We're going to retire the True Data brand and everyone has an ID5.5 email address. And soon we'll have an ID5 sweater. So we're sticking with the French play on word.
- [23:29]
A
I'm sorry, what is the French play on word?
- [23:31]
C
You don't know about the 5? So 5 in French is sink. There you go.
- [23:38]
A
See? Wow. Oh, wow. Mind blown. Did you know this, Eric? Or you were totally on top of the French playing words. Okay, my bad. All right, with that, let us take a break.
- [23:49]
C
I had to pitch him, right, to get a pyramid on board, so I had to deploy the charm.
- [23:55]
B
He's a great storyteller as he weaves it in.
- [23:58]
A
You and Elon musk with the sink jokes.
- [24:04]
C
I'm not sure I like the analogy. I prefer the reference to library better, I think.
- [24:12]
A
Well, all right, let's take a break. That was a great conversation. We'll be back with the news of the week. As always, lots of AI, lots of earnings, other stuff. So we'll be right back. This podcast is brought to you by playwire. Tired of ad monetization? Platforms that feel like black boxes even when they promise you complete control. Playwire is launching a truly self service ad platform that gives you actual control. Manage the components of your ad tech stack you want or let their AI handle the heavy lifting where you don't. Your choice real time analytics, transparent reporting and zero hidden fees. Mention market and get 50% off your first year. This offer is only available through the end of Q1 2026, so act fast. Visit playwire.com ramp self service to take control of your revenue. That's playwire.com ramp self service.
- [25:04]
B
All right, we're back everybody with the refresh. Hope you're ready. We have a lot of earnings but this is interesting stuff. Some themes that we're going to continue to pull out from a lot of AI and some other announcements. We'll see how much we can get. All right, as with last week, I'm going to tick through the earnings pretty fast. You guys just kind of keep track. Give me your reactions after we go through it. So we'll talk about six companies. First was Reddit. Reddit revenue up 74% year over year. Ad revenue up 74% year over year. Just a rocket like well done Reddit. Interesting thing is that they said they aren't seeing traffic from generative AI. So these are mostly users that are just going to Reddit. And even with all of the deals that they're doing with Google and such not having an effect. Amazon super strong quarter for ads up 24% year over year to 17 billion. Like their peer set with Cloud AWS up 20% year over year it is on $132 billion run rate. They subsequently announced a big deal with OpenAI to send the stock up. Magnite solid quarter revenue up 11% CTV up 18% Interesting thing and I want to ask Matt at some point something about this. They said DV plus which is basically all things not CTV Audio is the fastest growing segment of that Roku Solid quarter platform revenue which IS ads up 17%. They also released an ADS API Mountain. This is now the second quarter for Mountain and a solid quarter $70 million Q3 up 31% year over year. SMBs went from 6% of customers in Q4 to 15% this quarter and then finally this one snuck in right before we got on the call. Stagwell rocketed on both earnings and a partnership with Palantir. So on their earnings they disclosed that it's developing AI driven marketing platform that is designed to integrate large data sets from Palantir and unlock potentially hundreds of millions in revenue over time their stock went up 31%. So what do we got here?
- [27:17]
C
We got.
- [27:18]
B
I'll let you guys talk in a second. Continued strength in advertising businesses doesn't seem to be sign of recession. Continued strength in cloud and a couple of interesting things getting pulled out here. Right, SMBs up, audio up. What do you guys think?
- [27:34]
A
I'll jump in quick. So Reddit's interesting, doing really well. People don't really acknowledge that Reddit is a walled garden. You can't buy their ads using any system. Reddit, they had programmatic years ago, I think. I predict they're going to move to more of a hedge garden over the next year or so. They're definitely now they've got their sea legs on advertising. It's a little bit easier to let other people bid in there and you know, it's becoming the most important source of news, you know, text based news. Other people have sort of abdicated their role there on text. Matthew, you live.
- [28:09]
C
Yeah, I think it's interesting because also 75% are almost growing as fast as ID5, which is great.
- [28:15]
B
Not quite though.
- [28:15]
C
Not quite, not quite, not quite. It's interesting because there's a lot of talk about publisher and content being less important because people are going to consume it in chatbots. Reddit is a good example. You cannot predict its conversation. It's news, it's like it's fresh. And so I was at a tech Lab event this morning and John Roberts from People Inc. Was talking about the importance of content and new content because if the models are trained on last year's content, well, it's only going to be interesting for the thing that doesn't move. Everything that moves and conversation on Reddit are by definition things that move. Right. News, fault scores, all of those things. You can't predict that. You have to train the models over and over. So it's nice to see this is not all doom and gloom for publishers. There is a type of content that is not going to be disrupted by AI models because just you need people's contribution or journalists contribution for that content to be created.
- [29:06]
A
Yeah.
- [29:07]
B
On the other side of it there's just one Reddit, right?
- [29:10]
C
Yeah, but there is, you know there is, there is, there is a lot of like again sports coverage, like scores, like you know, new, new, new news. That news is happening, right. It's not just on Reddit, it's, it's happening everywhere. And so having journeys cover that is important.
- [29:22]
A
We're going to talk about how people and its licensing deal with Microsoft AI, but what's interesting is about this AI licensing stuff is that it's, you know, the easy to understand example is the latest news, right, that you'd have to license that. But actually old stuff is also licensable in real time because from a AI perspective, you need rag. So if you, if you want to analyze a bunch of content, not just get the stack, the answers from the already trained model, you need to get that stuff on the fly. And so there's a license model for legacy content as well that Reddit could really benefit from. The other ones. I went to the Roku event yesterday. So Roku had a event here in New York. It was focused on performance. They're very focused on performance advertising. I think the strategy there is that kind of a little bit like universal ads, they have a UI for performance advertisers that they control every aspect of and an API now which they announced yesterday. And then they obviously benefit from the broad ecosystem of programmatic for, you know, more upper funnel type stuff. So it's a nice, interesting, balanced, balanced approach strategically, you know, the pro on the product side.
- [30:37]
C
It's interesting to hear performance and TV in the same sentence, Wes, because it wasn't always the case. Now you have, you have the mountain news, you have Roku that's doing a lot of performance and I think it's, it's, we see a ton of traction on that front as well because a lot of that performance happens outside of the tv. Right. To be able to connect exposure on the TV with app, download on a phone or website, visit on a, on a, you know, visit on a website or even like in store transaction, like connecting the dots are the heart of it. There is an identity spine, there is an identity graph that has to power that. And I think that's going to be a really strong kind of movement going forward is how do we, how do we bring more outcomes into tv? Planning, buying, optimizing. That's a really exciting part of the world for us.
- [31:17]
A
Yeah. And that was a topic of conversation in the kind of, you know, social parts. I was talking to various employees of Roku, other people, and they were all talking about how, you know, IP address was used for so long for performance in ctv and it's not, it's not as reliable as it used to be. There's privacy issues and that you really need identity in CTV to make the performance happen. Their approach is they have that because it's their device, but not everyone has their own device. So. I think you're right on there, Matthew.
- [31:46]
C
I don't think it made the news, the refresh for this week, but there was a report released yesterday by Sim and Goaddressable, I think, which was measuring like the persist of IP addresses. The use of IP addresses. Right. And their persistent. The ability to connect IP addresses to other types of signal like emails or whatever, showing it's pretty poor. I mean it's kind of a dirty little secret. But like IP addresses rotate all the time. Right. And if you only use IP addresses as a way to recognize the device you're going to be, you're going to make a lot of mistakes. And that's where our integrations with devices that are basically a device ID. Right. That's what the 8580 is, gives us the stability that even if the IP rotates, even on TV, it rotates a ton. Obviously on mobile apps, on mobile phones, it rotates even more. But like being able to connect an IP to a device on a stable level and over time is really critical. Otherwise you're going to lose a ton of customers by just tweaking their IP address.
- [32:43]
A
Yeah. The rule of thumb I had heard was 20 days was the average length of an IPs in a household. Now obviously in a public space or a phone, it's going to be totally much faster, but in a household it's 20 days. But I've also heard that it's declining pretty rapidly because of some technical reasons like the move from IP4 to IP6 and things like that.
- [33:06]
C
Yeah. And the use of mobile networks to connect households now. Right. With certain isp, certain telcos actually plugging the home kind of WI fi into a mobile network, which then rotates the IP address much more often than traditional cable or land ISPs.
- [33:24]
A
I think with Matthew here, every conversation in the news is going to turn to an ID conversation. I think we have a little bias in our conversation today. One tree pony.
- [33:35]
B
I'm not mad at it. It's not even because I'm an investor. You're bringing us along for some pretty sophisticated info on Roku. This is the manifestation of a couple years of work. I would note people forget they invested in TV Scientific last year. Yeah. They've been all in on performance for some time. So it's great to see they're continuing on this road.
- [33:58]
A
Yeah, absolutely. And it was interesting. They announced an API for their ad manager and it was a little bit like, oh wow, they have an API, like join the party. Right. But you know, they're investing. It's good to see investment. Right. The, the Other thing, Stagwell, Palantir, it's kind of like one of those like two companies that are, they're buzzworthy but that no one can tell you what either one of them does.
- [34:27]
B
Stagwell does.
- [34:29]
A
Stagwell is an agency, right. They're like the numeral.
- [34:32]
B
They're, they're, they're an agency. They've got a collection of like really good brands like Code and Theory and they've got the sports business. They've got a bunch of like, you know, I think good disparate assets. But for sure, Palantir is a hot company. The hottest, hottest company in adtech besides id 5.
- [34:53]
A
They called them an ad tech company. Yeah, I like the idea of the, you know, merging the advertising ad tech world with the military complex. Like it's just targeting. Right.
- [35:09]
C
We're in a bit of hell scare and like we're golden.
- [35:11]
A
It's like, you know, someone pulls up an app, looks at a game of Candy Crush and either they see a 30 second non skippable ad or a hellfire drone kills them. And we decide on the fly. The algorithm decides. All right.
- [35:30]
C
How do you regard from that? I don't know. I have nothing to say. I don't have an ID joke that comes up right now.
- [35:36]
B
Welcome to the pod. This is why everybody's gotta, you gotta tune into YouTube just to see some of the reactions here.
- [35:44]
A
When's your cat coming back on the pod?
- [35:46]
B
He's locked out of the room.
- [35:47]
C
I'm sorry, we have to do a.
- [35:48]
A
Holiday episode or interview the cat.
- [35:53]
C
No problem.
- [35:54]
B
All right, let's talk about AI. So Matthew, talk about what People Inc. Said this morning to set the stage and we'll talk about their Microsoft deal.
- [36:06]
C
Yeah, so I mean John Roberts has been on a bit of a bit of a roadshow, right, about people, which is great because it's really helping educated the marketplace about how they can engage and be kind of offensive when it comes to like embracing AI and like interacting with AI LLMs. Right. Rather than being just like, you know, defensive on that front. And so they've done a bunch of deals with LLMs. They've been. I don't want to put words in its mouth, but I think they were early with cloudflare in the blocking of unauthorized or unmonetized content scraping from LLMs. Its position makes sense. This is just stealing content from publishers. If you just crawl the content, train your LLMs on the content without sending them traffic. At least Google does that and then sends traffic to publishers with search. There is a Monetization, they still keep 50% or more of the value in the process. I agree with you. But like there is at least a form of compensation. There is a, it's a give and take. Right. I'm going to take your content, I'm going to index it and then you're going to get the benefits of traffic going to you. Well, you've got all of the calories, none of the good taste with the LLMs because they take the content, they build that product and they keep the users for themselves and you don't get the traffic. So like that doesn't work. And John's making a very eloquent ace for publishers to protect their content, their number one asset, their content. Right. And license access to the content to people either via Cloudflare. I think they've done it here with OpenAI now they've announced this over Microsoft and there is a real marketplace to be built here. Right. Like how much we know the value of content from an advertising standpoint. Right. It's the CPM that people are ready to pay to reach users on that content. What's the value of content for AI engines? I don't know. We need to find that out.
- [37:51]
A
Right. There's also a cloud play here which is that the model of a big LLM used by hundreds of millions of consumers is one model and we want those, I think we could say we want those folks to pay for the content. But there may be thousands of enterprises using LLMs in different forms and they may need up to date content from publishers to make their LLMs function properly. And maybe they are paying on a per user usage basis as well. And that's an area where, so Microsoft is particularly well, well suited where they can provide the cloud resources, they could provide the base models and they could potentially provide a, a marketplace by which your models are have access to the latest content.
- [38:37]
B
Yeah, which is exactly what this announcement from people is about. So they signed a deal to be part of Microsoft's content licensing marketplace. It's called PCM Publisher Content Marketplace. And this is interesting from two perspectives now. One, it compensates pubs on a per use basis. All of these licensing deals today, largely from OpenAI and some others have been kind of all you can eat. They just want to have like open firehose access to content, which is hard to value. Right. To Matthew's point, like, you know, so it's a little bit of just, you know, what you can negotiate. This is the first of what I think is the model, you know, of the future of pay per use. So, you know, if you have good content and you know it's in demand or it's super niche and valuable, you get compensated on a pay per use basis. So I think that's interesting. The other side of this is they have some installed demand from Microsoft Copilot as the first customer. So presumably People Inc. And I think Gannett or now the USA Today company is the other customer. They're going to get paid from day one. Now, copilot is nowhere near OpenAI ChatGPT in terms of the volume of traffic, but it's just interesting experimentation. I think there's a real upside to it.
- [39:53]
A
Yeah, sure. I mean, I could be using GitHub to complete some autocomplete my code and it'll just tell me the latest fashion that Brad Pitt's been wearing on the red carpet.
- [40:04]
B
Maybe we're on an ID5 sweater.
- [40:06]
A
Yeah. Just crossing streams, getting the. The pop culture into the coding systems.
- [40:12]
C
It's always useful to have pretty code.
- [40:15]
A
It is, for sure.
- [40:18]
B
All right, those who subscribe got your Ari's epic article or write up about ADCP earlier this week. On the heels of that, the IAB introduced ucp, which was donated by Liveramp. I think people are trying to figure out exactly what that means and there was a bunch of conversation and questions around it. Maybe you can take us through Ari, ucp, adcp, how these work together.
- [40:47]
A
Because they do work together, man. This one, this one, I don't know. I'm not sure I could do it. There's an ID thing, so maybe Matthew can.
- [40:56]
C
Hey, Matthew, listen, somebody said id. Yeah. UCP is user context protocol. So it's like, it's basically a standardization of how you transmit user information. Right. In between agents. Right. So it's think of it as like the buyer UID or the extended ID field version of the ADCP protocol. Right. Buyer UID, extended IDs. That's RTB. That's how you pass user IDs in the RTP transaction. UCP is, and I'm oversimplifying here, but UCP is a way to kind of standardize how you share ID information between buyers and sellers in an agentic transaction. So it's a protocol. Right. It's great. That light ramp is kind of contributing to the open source effort and donating this to the tech lab. And it's just like a way for us ID providers and identity companies to be able to normalize how we pass identity information between buyers and sellers in an agentic world.
- [41:55]
A
Yeah. So ADCP Had a segment finder effectively that allowed you to interrogate the library of a given data provider. So this is at the user level. I'm still having a bit of a hard time understanding how this works. Maybe a future newsletter. But like, you know, in my mind. Explain this to me. Like you have a user and they're in one system, you want to get it to another system. How does AI fit into that?
- [42:20]
C
Oh, the fact that there is an agency transaction at user level is still a question mark. I think CP is a great planning and measurement solution. Is it going to replace RTB for like one to one transactions at a user level? I don't know about that.
- [42:33]
A
Right.
- [42:34]
C
But if it is ever going to happen, whether it's, you know, programmatically or I don't know if you can, you know, or guaranteed buys, for instance. Right. You could still do that at a user level. If you can get to the velocity level that you need goal to do individual transaction, you need to be able to pass user information in that call. And that's what UCP will help you normalize.
- [42:58]
A
Okay, watch this space. I'm going to leave it at that since I don't want to sound stupid on my own podcast. This is all science fiction right now.
- [43:04]
C
This is all science fiction right now, I think. But it's good that we're putting the guardrails in place. Right. For things to happen in the future.
- [43:11]
B
And very good that this happened to be the week that you were on, Matthew.
- [43:15]
A
Yeah, exactly.
- [43:16]
C
Always happy to come back to help you decrypt the complicated. Any topics in the world.
- [43:23]
B
Absolutely. All right, let's move on from identity to creative in the realm of AI. So two interesting things this week. The first was an article that Sora is trying to license ip. So there's a new feature called the Character Cameo, which is a way for individual users who are already able to use their likeness to insert themselves into AI generated videos to now incorporate characters they have created or downloaded. So either of you guys playing with Sora yet?
- [43:57]
A
I am not. I'm not a big video guy, but I've seen it online.
- [44:02]
C
Matthew. Nope, Nope. I've been a bit busy over the past few weeks, as you've heard and haven't had a chance to dig into this.
- [44:09]
B
Sorry, you get a pass. Sora puts you as the user, the star of your videos. So it records your face, it records your video, and then you can start running wild. One of the things they did that was very smart from the beginning is they did a couple of creator deals to Allow you to, you know, make videos with a creator. And the consent was always on. So one of the examples is Jake Paul was, you know, one of these first creators I was messing around with creating like fighting videos of me and Jake Paul, like fighting each other. So it's, it's fun. So now extrapolate that onto like every possible creator or brand or thing, right? So Ronald McDonald, Disney characters, all of that stuff. I personally think there's real upside to brands who, you know, have like incredible likenesses of characters to give ChatGPT, give OpenAI the consent to do this in the right way. Because the ability to like have this run wild with marketing is like so interesting. The other side of this is brands might, you know, not want to have any control at all, lose control at all and say, absolutely not. This is a binary decision.
- [45:24]
C
It's going to be fun.
- [45:25]
A
I think the, the rollout though is interesting because they roll out Sora like literally only like four weeks ago, and it has no controls whatsoever. And they, they kind of thumb their nose at every intellectual property law in the world saying, oh, you can opt out of having your IP in our system. You could call us or email us and list every single piece of IP you do not want to want in the system. Right? Which is totally unrealistic and illegal. And then, you know, just days later they back off of that and say, no, actually it'll be opt in. And they start, you know, hammering down the obvious, like the family guys and the things like that, which are obviously. And now, now they're positioning it like this is their idea all along. They're creating a marketplace for content. This was the business plan for day one. I just wonder, is this, you know, more sort of Silicon Valley, you know, break it, then fix it strategy or if they didn't even think about it four weeks ago and this is all a surprise to them.
- [46:21]
B
I think it's a little bit of both. But ultimately I think this use case is real and it could be a really interesting brand new channel, frankly for brands to engender loyalty with users. So I think it's something to keep an eye on. You guys got to use sorted to think fully, grock it.
- [46:36]
A
Do you, can we draw a line or connect the dots between this and our interview with Omar from Rembrandt? Is this like, because it's a similar tech, it's like a little bit of a different use case. Is he, is he like, is he on this or is this a competitor to him? I know I'm asking you to speak for him. But what do you think? How does this relate to your investment in Rembrandt?
- [47:01]
C
Sure.
- [47:02]
B
So Rembrandt puts brand images or products into creator or professionally produced videos.
- [47:14]
C
Right.
- [47:14]
B
So they're very focused on the applications in CTV because it creates kind of like net new product placement inventory for them. This is absolutely another application of that type of technology or that type of buying. So yeah, I think if it scales it ends up being just another surface for Revbrand to put products in.
- [47:33]
A
It gets kind of complicated, let's say. Okay, so let's say you have a deal that allows you to put Sonic the Hedgehog in your, in your videos. Right. So Sega's licensed it to you and that's cool. And then separately you have an ad model. So Tropicana is paying you to put their Tropicana in the videos. But then some creator has Sonic the Hedgehog drinking Tropicana and Sonic the Hedgehog has an exclusive with Minute Maid and suddenly the world's ended. Right. This is pretty complicated.
- [48:04]
B
Yes, yes. The way Rembrandt is going about it right now is very simple. The creators or the content producers get paid. Their brands pay for it. Once you start creating these like three part multi party transactions who all have to like approve this stuff, figure out rev share. I agree. It's going to get pretty complicated. It might be too complicated frankly.
- [48:30]
A
Right. Well nothing's too complicated to make money. I guess we figured it out in ad tech sort of. It'll be very interesting also to see if we hear from Meta and TikTok on anything similar because I don't. Because obviously TikTok was innovative in licensing the music. That's kind of through, through the predecessor company. That's what really was their breakthrough. And maybe this is the next thing and we'll, we'll hear a lot about.
- [48:58]
B
Is so clear that the future is video. The future is short form video and AI is going to like unleash all sorts of capabilities. It's just so clear this is the future. Just like right now it's like, like very much this emergent weird stage.
- [49:12]
A
Yeah.
- [49:12]
C
I think it's very clear that the tracks are being laid in front of the train right now. And like we don't really know where we're going with this but like yeah it's, there's a lot of it is being invented as we speak I think. And the fact that they go back on like oh yeah, but actually that was a business model or are we going to do this now? But like you know Actually stealing people's content is not good or IPs or you know, I don't think that, I don't think there's a five year plan being developed that has been developed on those things yet. Right. So. So yeah, it's interesting to see how the future unrolls, but also not to jump too much on the latest pad because it's probably going to change 27 times before it becomes stable and mainstream.
- [49:50]
A
Yeah. The economics are also interesting because with music, the platform is paying for the music and the users get it for free. And so that's sort of the cost of doing business with these characters. You can imagine it being a little more like ringtones where the creators might actually be charged like hey, you want Sonic the Hedgehog in your, in your video? You know, give us some tokens or something like that, you know. So that, that's another pretty interesting question here.
- [50:19]
B
Absolutely.
- [50:20]
C
Related as, as a customer I'd watch Eric fight Jake. Well, or, or Sonic the Hedgehog for that matter. Like I'd ve what's my cpm.
- [50:33]
B
And I have not approved the distribution of those videos. They're not ready yet. I'll send them to you in a group chat. One more on this. So Coca Cola is doing another AI generated ad for the holidays. Talk about that one.
- [50:48]
C
Ari.
- [50:49]
A
Yeah, so last year Coca Cola did their holiday ad using AI and it was widely panned like the wheels were not spinning like real wheels and things like that. Yeah, there's all kinds of. It was just, you know, AI a year ago. That was the dark ages. We barely knew how to create electricity. Now to their credit and also to the credit of, to the ability to have an apples to apples comparison, they've done a new ad for this holiday season and it's much better. And they said publicly that there it costs a lot less than building this the old fashioned way with, with you know, CGI from scratch. And they're pleased with it. But I think for us we could just look at it and say, and we'll put the link in the, in the newsletter. It's better than last year. And that's telling you something about the quality of video creation tools in AI.
- [51:39]
B
And the fact that it's one of the largest brands in the world being so leaned in I think is like really, really interesting.
- [51:45]
A
Yeah, I think it's a little bit of like look at us like the same way super bowl ads have the halo effect of all the pr. But it's, it gets us talking about Coca Cola and their winter ads and that's probably part of the point.
- [51:59]
B
Yeah.
- [52:01]
C
To be honest, it's not an ad they could have done for real because there's so many animals in that ad that like, it would have been impossible to, you know, so if that was a brief in the first place. Right. The squirrels and the panda and everything, like, there was no other way to do this than with AI Anyway.
- [52:18]
A
So, Matt, I don't want to harsh on. I don't want to like, overly focus on the fact that you're French, but I've been told or I've seen TikTok videos that say that the word squirrel is like the hardest English word for French people to say, and you just rolled it off your tongue like it's not a problem when do. Have you. Do you have a history the words. Is squirrel, you know, particularly difficult?
- [52:40]
C
Squirrel is. Okay. The. The. The French traditionally struggle with the th. Right. That sound doesn't exist in. In the French language. And that's a real struggle. But I've lived in. In the UK and in New York long enough now that I can say squirrel without. Without flinching.
- [52:57]
A
Do they have squirrels in the uk?
- [52:59]
C
They do, actually, yes. Even in France particular or.
- [53:01]
A
Not really. Okay.
- [53:03]
C
I saw Chipmunk Go the other weekend and. And I was. I was. I was stunned. Never seen a chimpanzee before.
- [53:09]
A
They're very cute.
- [53:10]
C
Still. Still learner 49. There you go.
- [53:12]
A
Yeah, they're very. They're very cute.
- [53:15]
C
Squirrel in French is ecurate. That's what it is. Okay.
- [53:19]
A
I cannot pronounce that.
- [53:21]
C
Which Americans can't pronounce. There you go.
- [53:25]
B
I'll pass on that. Hey, we have more news than time. Is there anything here?
- [53:30]
A
I want to talk about the meta thing. I'm really worked up about the meta.
- [53:33]
B
Thing, and I'm worked up about something else.
- [53:35]
C
Please go ahead.
- [53:36]
A
All right, so report came out today that meta internal documents. So meta knows this. That yet 10% of their revenue is fraud. That is an astounding revelation. $16 billion of fraudulent ads or ads for fraudulent products or which is worse.
- [54:01]
C
For fraudulent products because remedy fraud. We've seen that movie before, right? They over report metrics. They over report delivery, over report performance. But like, you know, I guess when you're grading your own homework, like, you know, you got to have a, you know, take everything they see with a pinch of salt. This is worse. This is them selling ads to people who make like the thing that they're. They advertise on their platform. Don't exist.
- [54:22]
A
Don't exist.
- [54:23]
C
All of Those shit, they're never going to come to your house. Like, this is actual scam.
- [54:27]
A
Yeah. I think this also includes ads that are deceptive. So this also includes those ads that show like a celebrity who's dead, but they're not dead, stuff like that. It's bad though. And we had Rob Leathern on the, on the show and he actually tipped me off to this because he just started a company called Collective Metrics that's focused on this. But what he was telling us on the show when he was on was that they get millions, literally millions of new ads every day and they just don't review most of them and they take them down if there are bad things happening and the result is just massive unstopped fraud on the second largest advertising system in the world.
- [55:03]
C
World.
- [55:04]
B
To give you an idea of the scale, it's not millions. I'm taking this as a quote from Reuters. The social media giant internally estimates that its platform show users 15 billion scam ads a day.
- [55:20]
A
Unbelievable. I mean, this is, this is. Someone should go to jail. Like I'm, you know, we. I spend all my time on antitrust. I could give a shit about antitrust. I care about grandma signing up for, you know, something that's going be to charge their credit card every month or buying a product that doesn't exist. Like. And they know about this. This is awful.
- [55:37]
C
Right?
- [55:38]
A
This is awful.
- [55:40]
C
I'm. I'm surprised that we're still surprised, but. Yes, it is awful, but I'm surprised that we're still surprised. All right, we don't have to cover.
- [55:48]
B
Off on my thing. My thing was Brian o' Kelly versus versus Jeff Green.
- [55:53]
A
Come on, come on. Five minutes on Brian o' Kelly versus Jeff Green. Give us, give us the timeline on this.
- [55:59]
C
Okay, that's not on the show. Notes. Hold on, hold on. We didn't prepare for this.
- [56:04]
B
We didn't prepare for it. Matthew, it's okay. I'll walk you through. I'm very invested. So Brian posted a video response to some of Jeff's posts about the ecosystem and was just a complete troll. It was on LinkedIn and it was basically one of these videos where I think we talked about this. Ari, he was doing. He was writing Jeff's content and like sort of kind of explaining what he was thinking at the time. It was a complete troll.
- [56:38]
A
It was the funniest thing I've ever seen on LinkedIn.
- [56:41]
B
It was the funniest ad tech content I've ever seen on LinkedIn. I've seen some funny things. He recently Just yesterday or the day before posted a follow up which was a bit of an open letter to Jeff. And at the end of it there was a Screenshot of Jeff's LinkedIn and it appears that Jeff unfriended or unconnected with Brian on LinkedIn. They're no longer connected or that was the implication.
- [57:08]
A
Yes.
- [57:10]
C
So.
- [57:12]
B
I will say I'll let you guys go first. What do you think?
- [57:16]
A
I think it's hysterical. I love a good beef. It's a friendly beef. They're not like really being mean to each other. And the comments on Okelley's post are hysterical. There's trade desk PR is involved there, there's some, some, there's some woman who's like, women can't beef like this. We would be thrown out of the industry and various people laughing and, you know, it's pretty funny. Right? You know, no harm, no one's getting injured here. No stock prices tanking. It's just like a little beef between, between people, frankly. They've known each other for 20 years. Because what a lot of people don't appreciate is that Jeff was a senior executive at Microsoft after the ECN got bought thought and he was instrumental in the decision from Microsoft to outsource all their programmatic to app Nexus, which was the thing that made AppNexus's business. So they've, they have a long history.
- [58:07]
B
And Fast forward to 2025.
- [58:09]
C
They're beefing. Everybody loves a good celebrity death match, right? They're, they're, they're at a celebrity. They, they are allowed to do that.
- [58:18]
A
Yeah. What I would love to, I'd love to hear them talk for real about AD CP and AI because to a large extent o' Kelly is positioning AD CP as a replacement for rtb. And that probably doesn't sit that well with the trade desk and other DSPs that bid entirely on RTB.
- [58:39]
B
Brian, Jeff, come on.
- [58:41]
C
A pod.
- [58:42]
B
Let's do this. We'll have a longer special episode. You could air it all out anytime. You know how to get in touch with us.
- [58:48]
A
You do.
- [58:49]
B
Let's end it there.
- [58:50]
A
All right. Well, it's a great conversation, Matthew. And to the both the ID5 team and the TrueData team, congratulations. Definitely. We're going to be watching this space, hopefully and see some exciting things going on.
- [59:04]
C
Absolutely. Thanks. Thanks for having me on, guys. Congrats.
- [59:07]
B
We'll see you next week, everybody.
- [59:09]
C
Bye bye.
- [59:09]
A
See you next week.
- [59:10]
C
Thank you for subscribing to marketecture.
- [59:13]
A
New interviews are added every week at marketecture. TV and your favorite podcasting app. Thank you for listening to the marketecture podcast. New episodes come out every Friday and an insightful vendor interview is published each Monday. You can subscribe to our library of hundreds of executive interviews at marketecture tv. You can also sign up for free for our weekly newsletter with my original strategic insights on the weekend week's news at News Market tv. And if you're feeling social, we operate a vibrant Slack community that you can apply to join@adtechgod.com.