Started here?
Yeah, let's do it.
All right, great. Thank you, everybody. Let's get started. My name's Ron Josey. I lead coverage of the internet sector here at Citi, and very happy to have with us today Michael Komasinski. I butchered your last name.
No, pretty close.
No, was that good? Okay, Michael Komasinski. Yeah, I was going to say new CEO, but not new CEO. We are 18 months in.
18 months in.
Criteo, I think we all know what Criteo does, but bottom line is, has one of the largest identity graphs out there, I believe.
Yes.
And leverages this data set and shopper graphs to basically help to drive greater sales for retailers and advertisers.
Yep.
Full stop. With that, I'll let you do a better job of explaining this directly, but Michael, I just want to say thank you for joining us, and why don't we kick off specifically, just talk to us. You're 18 months in the CEO role. A lot's changing in the industry, a lot's changing at Criteo. So maybe quickly tell us about Criteo, the business. Most people know already, but then, where are you focused here as we move forward in this new world of agentic advertising?
Sure. I'd start at the highest level. Criteo is a commerce intelligence platform, right? We pick those two words very purposefully. We believe that commerce really is the organizing construct for advertising and marketing in a world where everything is shoppable and biddable. Then the intelligence part comes in, how do you allocate media investment optimally across what is an increasingly fragmented ecosystem of places where people spend time, where they shop, where they make decisions? So really the future is about how do you drive commerce outcomes in a more complicated ecosystem in a very sort of AI-driven, highly automated way? So that's kind of the set of the North Stars or the things that we believe in, I think at the highest level.
How the company actually accomplishes that, I'll try not to get too complicated with it, but we have several different data sets that we use to drive that decision-making of where to place optimal client investment. You mentioned one of them, the identity graph. It is the third-largest identity graph in the world. We have the third-largest product catalog in the world. We are incredibly good at ingesting 25 billion raw SKUs, some updated as frequently as hourly, and then processing that in and normalizing that data set into a SKU catalog of about 5 billion SKUs globally.
Michael, sorry, the third largest product catalog in the world.
Yes.
And you mentioned 25 billion, or?
25 billion is the raw SKU count that we ingest on a regular basis, yeah. What you need to do is you need to be able to normalize that down to something that's usable so that this vacuum cleaner that's got just a slightly different aspect is the same thing as that vacuum cleaner.
And you normalize those two things together, basically. When you combine identity with product catalog, with then the shopper graph, which is the events timetable of what do people browse, what do they consider, what do they do on site, and how do the learnings from that help you to determine where to place the next best, or the next investment or the impression to buy. That's basically the three data sets that combine to enable that. Without getting sort of too deep.
No, that's perfect.
into the science. Those data sets are reinforced with the client relationships that we have. Through the ability to ingest product information, availability, and landing activity on site, that becomes a self-reinforcing loop that makes our performance better and better over time. If I keep going with a little bit of the lead into the question, the focus areas for the company, at the highest level are really threefold. One, to innovate and lead the industry in agentic commerce, and the kind of two interesting use cases there would be retailers developing new agentic shopping services and how do we power those for them. As well as true pure-play agentic channels like paid advertising on OpenAI, where we're their leading ad tech partner globally. Second priority would be continuing to innovate and maintain our industry leadership in the fast-growing retail media channel.
We can come back to that in some detail. The third would be to re-energize our performance media segment by making it more cross-channel, which enables us to conquest new budgets in the funnel. We sometimes refer to that as the full funnel cross-channel strategy. I think if we execute on agentic retail media leadership and re-energizing performance media, then this is a really exciting business over the next couple of years.
I want to sort of follow up on that last one on re-energizing performance media. I think you said making it cross-channel full funnel. Understood the vision, where are we coming from? Were we not doing that prior?
No, but for good reason, right? If you need to back up a couple of years, back into maybe kind of 2023, 2024 before I arrived, the company would have been very focused on addressability and essentially Google's Privacy Sandbox effort.
That would have consumed a lot of attention in that timetable, timeframe, and rightfully so. There was not as much product innovation going on at that time, and so I was fortunate to arrive at Criteo at the time that I did because about two months after I joined, Google backtracked on the Privacy Sandbox proposal. We essentially have now an addressability landscape that is more or less unchanged from where it was two or three years ago. I think also the regulatory environment and other factors suggest that that is a very stable foundation to build on. I wanted to embark on a more growth-oriented agenda for that segment by pursuing the full funnel cross-channel strategy. Start to innovate channel connections, again, which enables us to conquest mid or even upper funnel budgets, which is how you grow a performance platform.
Yep. Super helpful. I have more questions on this. The big three. Before I do, I wanted to talk a little bit about the leadership team. I think Connor, who is in the audience, just announced CFO. We are in the market for a new Chief Product Officer, if I am not mistaken. So talk to us about where we are from a leadership team, and then we will go from there.
Yeah, definitely. I think first of all, it is worth reminding people, we have got an incredibly strong bench in the company, across all the different disciplines. We have got a very rigorous and methodical approach to cultivating our succession plans. Then ensuring smooth transitions when it is time to do that. Connor coming into the CFO role is a good example of that. He has been with the company for eight years, had been on the succession plan for some time, and really has the perfect background for what we need as a company right now. Connor has got a strategy background, ran our partnerships group, understands the ecosystem, I think as better than almost anyone I know in the industry.
With eight years in the company, he has got a lot of familiarity with how we work and how we operate, but also has some deep opinions on things that we can change.
A fresh perspective, I think, is always a good thing, and this was a very methodical, well-thought-out transition, and with a candidate that hits the ground running 60 mi an hour out of the gate.
Sure.
Yeah. On the Chief Product Officer side, Todd did a great job in his six years at Criteo in terms of navigating the Privacy Sandbox situation that I talked about, and then really being the architect of the full funnel, cross-channel, self-service strategy that we are going to continue on. Todd had a long-time ambition to be the CEO of a company and has a fantastic opportunity to do that. So, leaves us with a great bench of talent that he's brought in and is a very natural progression to his personal career journey, and we wish him well on that. We currently have our senior product team reporting into Diarmuid Gill.
Our Chief Technology Officer. We're in the hunt for a new Chief Product Officer with a great bench of talent at the next level down. So, don't need to change the strategy in any big way and have a great bench and a good interim setup. I'm very comfortable with where we are.
Perfect. That dovetails into the three key focus areas. Don't need to change the strategy, have the tools for agentic, retail media.
Correct. We're always reacting and pivoting and-
Sure.
improving how we scale those strategies. But agentic, retail, performance, nothing changes with that strategy, and I think the major elements within each of those.
Perfect. I want to get into the dataset and agentic commerce here, particularly around your partnership. Before we do, just to level set to us, post 2Q, there was a little bit of a budget reset, I feel, around some of your largest clients in performance media. Just talk to us about the visibility into the business. I think we saw qualified pipelines were actually up quite nicely.
Yeah.
I think we saw U.S. new business revenue up quite nicely. But then, we also have some of your largest clients pulling back. Is that behind us, or how do we think about these two?
Yeah. Coming into the year, we started to notice some softness in budgets with a handful of large clients. Then that trend continued across Q2.
That's what got us to the downgrade that we announced in August at earnings. We wanted to give ourselves a prudent outlook, given that we had to do that twice. Since then, we saw trends stabilize in the July timeframe. Also the commercial teams, I think, really on top of each of those situations. In some cases, that is just competing more effectively for budget shift. A budget moves from one tactic to another, and as Criteo, we have enough products to satisfy both needs. We need to make sure that that's a left pocket, right pocket shift for us. Some things were maybe macro that are out of our control, but that's going to happen, and that's why it's important to have a new business machine that can bring in new revenue that offsets the inevitable fluctuations that clients will have.
Those were some of the stats that we shared at the August call, right? New business performance in the U.S. had been up 24% year-on-year, which is a really good signal. I would say attributable to some of the talent that Ed has brought into the business since he joined in January. We noticed that qualified pipeline was up significantly, I think in the mid-30s. Overall pipeline has now come in over 50% from agencies. A lot of the things that we've talked about come into bear. I think, to the point, that situation has now stabilized. Each account is a little bit of a different situation, it's just game on t o now get some of that share of budget back into the business and make sure that we follow through on new business efforts.
What changed in July that you said trends have stabilized?
Yeah. I think it could be attributable probably to some of the commercial efforts in some cases, or maybe some macro factors abating. Probably a mix of the root causes that I went through there.
Okay.
Yeah.
That's great. One of the questions that we often get, I want to get into the data set and everything else because clearly that that's the asset of the company. But the questions we're getting is, the Criteo and OpenAI's tech partnership, and I think you mentioned if we can focus on the three things, agentic commerce was one of the first ones.
Yeah.
Tell us a little bit more, I think about your partnership with OpenAI. Specifically, I think you now have, what, 2,000+ clients on. There were a slew, there was a release not too long ago of continued strength. Let me just make it a high level, how did this partnership come about? Why did they choose you? Talk to us about the evolution of how that's going.
Sure. I think the reason that they chose us is multifold here, but basically it comes down to, we look at advertising in very similar ways. As Criteo, we don't start with reach and frequency as our starting point. We start with, what product do you want to advertise, and what is the context and the relevancy that triggers that? How do I drive to a commerce outcome? That is the way that the OpenAI ad platform is architected, so it gets into a lot of the value add in the way that our partnership fits together. I like to say we optimize the inputs. What SKUs am I going to advertise? Is the data associated with those SKUs up to date? Is the metadata correct? Is it deep enough and rich enough to drive the relevancy signals?
When it comes to prompts, how do we engineer prompt hints that are insightful and comprehensive to maximize all relevancy display coming out of their engine? On the back end, how do I give insights into cross-channel measurement about how their paid referral traffic is performing relative to other paid referral traffic? So I call it optimizing the inputs, and explaining the outputs, and that's a lot of the value that we provide. Now, we do have a product-to-product or tech-to-tech relationship as well. That is largely about us giving them market feedback on feature function that needs to be developed, helping them with their roadmap. Ultimately, it's them building their platform, but it is quite a technical relationship-
Yeah. Okay.
week to week. They announced a couple of weeks ago that they had gotten onto a billion-dollar a year run rate, which is exciting. Opened up now in 51 markets globally, the European expansion, then the Middle East expansion, so that will start to drive client count, I believe, through the rest of the year in a pretty strong way. Really what we want, what they want, is to move really from a lot of test budgets to always-on budgets and to make them bigger. But with the efficacy we see in both mid-funnel and lower-funnel, I am pretty confident that those budgets will scale and that they will become part of always-on plans.
Just to pay that off a little bit, in the mid-funnel, there is an amazing statistic that we keep sharing because it keeps showing up in the data, and that is that 80% of paid referral traffic is new to brand, which is an amazing incrementality metric when you think about mid-funnel objectives. Then in the lower funnel, it converts 1.5x- 2x better than traditional search. So those are great conversion rates. Lots of reasons to believe that those with that kind of performance, that kind of insight, that that will move from test budgets to always on, and that it will pull investment from some other competing products in the mid or lower funnel.
Where do you think budgets come from? Because n ew to brand is interesting. Greater conversion rates is even more interesting.
I think new to brand will put pressure on CTV budgets. It will put pressure on some social budgets. Those are the places that you derive new customers, primarily. Then in the lower funnel, I think it will put pressure on certain tactics within search. Certainly, product listing ads or PLAs, maybe even, depending on the CPCs you are seeing, could be even branded search. But not all of traditional search, but certainly some tactics within. Like economically, you will just get better performance on OpenAI. Look, if they are driving incremental outcomes, there certainly could be some incrementality as well, just in terms of overall ad spend as a percent of GDP.
Open to 51 markets now. Last metric we heard was 2,000 advertisers.
Yep.
Who are these advertisers, and how quickly are they able to go from test to actual line item?
Yeah. So look, Criteo has quite a varied and global client base. I think that is one of the other reasons that they chose to partner with us, right? We have 17,000 clients globally. It is particularly strong in the mid-market. So that is a lot of the 2,000 that we're working with on their behalf. I think a client typically needs about two months of testing, go through a couple of weeks per campaign to do evaluation and tweak it, run it again. So it probably takes about two months to start to get to some comfort level of how it's performing relative to other tactics. I would say two milestones we're looking forward to will be holiday season this year, so kind of come October, November. How many clients are choosing to invest in OpenAI as a mechanism to drive their holiday objectives?
I think that will be an interesting signal for all of us, everyone in the ecosystem, let alone Criteo and OpenAI. Early next year, lots of brands and advertisers tend to set budgets at the beginning of the year by platform, by tactic, for their always-on. So I think that's another signal kind of in Q1 that we'll get a read on. Not that those have to stay fixed over the full year, but that is a pretty typical client approach to set those early in the year.
I want to get into the impact, not to spend the whole time on OpenAI. Two other quick questions, because I think agentic commerce can potentially change how we shop if you have an agent helping you find what you are looking for that you may or may not know exist. Question number one is, we talked about two milestones holiday this year. Do we have enough product with the advertisers on ChatGPT to be meaningful for users? I know we talked about your product catalog, so I am bringing it back to Criteo's product catalog. That's point one. Point two, the impact to Criteo itself. When can we get on a call and you and Connor can say it's now incremental or adding more so to growth?
Yeah. So there's enough product for sure. We continue to work with them to tune the relevancy engine to improve budget completion and fill rate. That's just a work in progress. As they develop now some of the additional performance capabilities, like optimized CPC, the CAPI measurement program that they introduced about a month or so ago, even little things like exclusion lists, I think that all starts to drive volume and help command larger budgets. For us, we've continued to be conservative about guiding on performance just because it's a decent client count, but relatively small budgets per. We believe that it will be a meaningful contributor to 2027, and we are continuing to talk about how we are going to give guidance on that. We want to give guidance on that. It's an exciting part of our business.
We need to be mindful about starting to guide to some kind of a statistic then that we can update regularly with confidence. But meaningful is the word that we keep using for now. Again, everyone's interests align. We really want to guide on something for this next year, so we are going to figure that out.
Perfect. So maybe moving on to retail media. Specifically, I think some of your core offerings like Commerce Max, Sponsored Products, was it Page Intelligence, Conquesting. Just how do you see retail media evolving here?
Yeah. It's really great, actually, because when I joined, shortly after I joined, had the sort of misfortune of the pretty big news event with our largest retail client taking some, I call them legacy scopes of work in-house, because they very much continue to use our technology. But since then, the business has really embarked on being more diversified. That is both strategic on our part, but it is also what the market is sending us as a signal. So the retailer ad stack is becoming more diverse in that there is opportunities for an ad server for Sponsored Products.
There is an opportunity for an ad server to serve display and online video. There is an opportunity for a mediation layer. There is an opportunity to optimize paid and organic and what shows up on a product listing page. Conquesting, right? So there is a lot of different monetization tactics now that as these retail media networks get bigger and more sophisticated, they really have an appetite for, and so we are just building to market demand.
But the nice benefit of that is it diversifies the business so that we are not sort of so binary on if you are not the sponsored ad server, you are just in or out. And obviously, that is not always a great situation. So being more diversified, I think, allows you to evolve with clients, bring them more product, and if they ever did in-house something, that you still have a role to play in the stack. And that is just talking to supply side, right? On the demand side, with Commerce Max, we continue to introduce new insights, features, new capabilities for how cross-retail campaigns are run and measured, and that is the core value proposition there. Because you want to make running retail media as frictionless as possible, because that is really been the inhibitor on the advertiser side for scaling spend outside of the top two players.
So that is our mission with Commerce Max, and we see really good results happening here over the summer with that.
That is great. And maybe the last one on just, we talked about agentic and retail, let us talk about performance media.
We talked a little bit about it earlier, but how do you? I think we talked about the Privacy Sandbox might have taken a little bit of wind out of the sails. What products can we look forward to so that we can see that growth come back or?
In performance media?
In performance, yeah.
Yeah. The highest priority channels for us are CTV and app. Said simply, we just need to be where digital activity is happening, right? Where is the audience going? And budgets will follow. Now, each of those channels has different dynamics about how people behave and interact in those channels that then allow you to conquest additional budgets outside of traditional lower funnel remarketing tactics. That is why we always call it full funnel cross-channel. You need to establish channel capabilities in order to conquest budgets in the funnel.
They're just different sides of the same coin.
As we get into maybe the technology behind this, I think we said agencies are about 1/3 of Criteo's total media spend-
Correct.
if I've not mistaken.
Yes.
Now most are accessing Criteo's back end via MCP, which-
It's coming. The fast-growing way that they access, I wish I could say that's most of it, but it's the fastest-growing, and it's certainly the area of interest.
The question is, ultimately, if we get in this world where agents are buying from agents, and we're sort of doing things automated, we have this MCP compatibility that's happening as we speak here. How does that accelerate the business? Or, do you see greater flow of dollars come through Criteo's-
Yeah, definitely. I'd slightly reframe the premise and I would just say that, in particular, the agency segment is moving towards headless buying.
Okay.
Kind of the simplest way to say it, because I don't know that it's fully agentic. It's agentic-enabled, but this kind of headless is really the phrase, I think, to think about it. It's agencies wanting to own a workflow that allows them to standardize the way that they plan, run, and activate media, which gives them greater operating efficiency, across their headcount. The ability to build more automation into each step in that process gives them automation and operating leverage. That has an impact on who they want to partner with. Criteo, as a partner with a lot of data assets and algorithms that are geared towards turning those assets into outcomes, we become a great partner in a setup like that, where I just want to go direct into your audience builder to help me build bigger, higher-fidelity audiences.
I want to come into your supply paths to buy supply at the cheapest and most optimal way. Or I want to get to a commercial agreement where if you can deliver me a ROAS of 10, I can pass that on to my client in a way that's good for them and still allows the agency some economics in that setup. There's a variety of, I think, ways that it plays out, but agencies want to own sort of a wide end-to-end stack, but they want to partner sort of at the things below that help scale each module in that setup.
Got it. And one last question on the tech side. Then we will get into maybe some other, the profitability and re-domiciliation, but the GO and self-service platform. So, talk to us about what it could mean to the business, and I believe it is more of a 2027 event. But we have this self-service platform. It's called GO. Talk to us about what this could be. As we just got finished talking about agencies and MCP, we just talked about the potential with ChatGPT and [inaudible]
Well, they're like flavors of the same thing, right? So with agencies, we use a fancier word like headless buying. GO and self-service is geared towards the SMB market, which is just a different flavor of the same trend, which is automated buying of outcomes. The ecosystem has gotten very complicated. Really, as an advertiser, you just really want an outcome, and you want it to be easy, and you want to work with a platform that you can trust will allocate your dollars optimally across the channels or impressions that are relevant to driving your products. That is really the philosophy that GO and self-service is built on.
Now, that is both an acquisition vehicle for us for new clients in that segment, on a self-serve basis, not dissimilar from some other self-serve platforms that are in the market. We also use it internally, so you could think of it as more efficient managed service, that makes us a more efficient organization in how we use our own tools. That continues to help us get operating leverage and maintain healthy margins in the business.
That is perfect. Well, there is about four minutes left. I do not know if there is any questions in the audience. You can think about questions, and then, we can. Oh, there is one. That is-
[inaudible]
There's a mic behind you if you want.
OpenAI's international rollout, have you seen effect on your international customer account?
You mean in terms of non-AI? Or, sorry, non-OpenAI clients?
No, on OpenAI.
Oh. We're not sharing updates on that just yet, but we should, in conjunction with them, hopefully give an update to that 2,000 advertiser figure sometime in the fall, I hope. Yeah.
When is the CTV product going to ship? Is that end of year?
It's interesting. It's already shipping on some level right now. I think the right question would be, when does it scale and become a more meaningful part of the revenue stack? I think that will happen in 2027. So we are already driving a managed service CTV offering today, and commercial activity on that has accelerated over the course of the summer. We need to get to a place with that later in the year, early next, where that's something that we would guide on or give some indication of how it factors into the stack next year. But we're absolutely going to get there with that channel. It's not dependent on some big tech build that we have to do. That potentially would help scale and accelerate something. But building the book of business on managed service is actually something we're doing already.
Thank you. Where are we with streamer partnerships out of the big 10? Are we seven out of 10, eight out of 10?
I'd have to get back to that. It's not bad. We'll come back to you on that. I know we have several of them. I'll have to come back to you on the specific number.
Maybe as we wrap up here. I would love an update on the re-domiciliation, if I said that correctly.
You did. You did.
I think we've completed the Luxembourg step. What's next? Again, remind us the benefits here.
Yeah, sure. We did complete the shift to be a Luxembourg-based company at the beginning of August. As a Luxembourg-based company now, that does two things for us. One, it removes some of the capital controls that we had on share buyback when we were domiciled in France, and it makes us eligible for actually a couple of passive indices now, most notably the MSCI. We are on track to complete the transition to a U.S.-based domicile on January 1st of 2027. There will be a shareholder vote to confirm that in December. All signals are that that will complete as planned, and then we would be a U.S.-domiciled company effective at the beginning of next year.
Does that impact anything from maybe an M&A perspective or?
No, not necessarily. You continue to progress into fewer and fewer capital controls as you move through that. There still are some limitations in Luxembourg, though nothing to the extent that we had in France. In the U.S., you would know that regime well. I think in the U.S. conversion, it accelerates the rest of the passive indices that we want to be included in.
Perfect. Any other questions from the audience? With that, maybe we will wrap it up. We have a lot to think about with the key focus areas. Michael, thank you very much for your time.
Thank you, Ron. Thanks for coming, everyone.