Good morning, and welcome to Criteo's Third Quarter 2025 Earnings Call. [Operator Instructions] Please note this event is being recorded. And I would now like to turn the conference over to Ms. Melanie Dambre, Vice President, Investor Relations. Thank you. Please go ahead.

Good morning, everyone, and welcome to Criteo's Third Quarter 2025 Earnings Call. Joining us on the call today, Chief Executive Officer, Michael Komasinski; and Chief Financial Officer, Sarah Glickman, are going to share some prepared remarks. Joining us for the Q&A session is Todd Parsons in his role as Chief Product Officer. As usual, you will find our investor presentation on our Investor Relations website now as well as our prepared remarks and transcript after the call. Before we get started, I would like to remind you that our remarks will include forward-looking statements, which reflect Criteo's judgment, assumptions and analysis only as of today. Our actual results may differ materially from current expectations based on a number of factors affecting Criteo's business. Except as required by law, we do not undertake any obligation to update any forward-looking statements discussed today. For more information, please refer to the risk factors discussed in our earnings release as well as our most recent Forms 10-K and 10-Q filed with the SEC. We will also discuss non-GAAP measures of our performance. Definitions and reconciliations to the most directly comparable GAAP metrics are included in our earnings release published today. Finally, unless otherwise stated, all growth comparisons made during this call are against the same period in the prior year. With that, let me now hand it over to Michael.

Thanks, Melanie, and good morning, everyone. Thanks for joining us today. What excites me most about this quarter isn't just the strength of our results, but how our business is growing into a platform that reaches far beyond any single channel or format. Consumer attention is more fragmented than ever across websites, apps, social feeds, connected TV and now AI-driven assistance. That fragmentation creates complexity for advertisers, but it creates opportunity for us. Criteo is built to meet shoppers wherever they are and deliver the outcomes brands care most about. Progress we're making in Performance Media, in Retail Media and in agentic AI shows our strategy is working, and our opportunity is only expanding. Since stepping into this role, I focused on getting close to our teams, our clients and our partners. And what I see gives me confidence that we are on the right path. We're building on our strengths and sharpening our focus, and we're seeing early results from our push for greater decentralization and agility. Our purpose is clear. We power shopper journeys that unlock commerce outcomes. We help brands, agencies and retailers connect with consumers wherever they are across every channel and every device. We've evolved from being an open web company to a diversified multichannel platform with about 85% of our media spend now happening outside of desktop display. Retail Media represents roughly half of our spend with strong growth across mobile, social and video, including CTV. This cross-channel diversification continues to set us apart from single-channel ad tech players and walled gardens. Criteo's diversified reach across these channels is a core strength that's often underappreciated. Our platform enables consistent, measurable returns across channels, delivering stronger performance than siloed alternatives. As advertisers focus on outcomes rather than media buying, they're looking for the ability to…

Thank you, Michael, and good morning, everyone. We delivered strong Q3 results with significant operational leverage driven by top line growth and disciplined cost management. Revenue was $470 million and contribution ex-TAC increased to $288 million. This includes a year-over-year tailwind from foreign currencies of $6 million. At constant currency, Q3 contribution ex-TAC grew by 6% year-over-year, representing 15% on a 2-year stack. Client retention remains high at close to 90%, underscoring the resilience of our model. Macro trends remained stable throughout the quarter and during the back-to-school season. We saw higher advertising spend year-over-year across several key categories, including office supplies, furniture and personal care. In Performance Media, revenue was $403 million and contribution ex-TAC was $222 million, up 5% at constant currency and 10% on a 2-year stack. This was driven by our Commerce GO solution, up 6%, which leverages our large-scale commerce data and AI-powered audience modeling technology to connect advertisers within market shoppers. We also benefited from incremental AI-driven performance enhancements on top of the strong AI improvements we implemented last year. Ad Tech Services reduced Performance Media contribution ex-TAC growth by approximately 100 basis points due to lower spend in our media trading marketplace. Overall, we benefit from a diversified client base and a global footprint. By region, we saw media spend growth in Asia Pac and EMEA and softer but improving trends in the U.S. Travel remains our fastest-growing vertical, up 24%, with classifieds up 14% and marketplaces also performing well. Retail med spending was softer, including an 11% decline in fashion. In Retail Media, revenue was $67 million and contribution ex-TAC grew 11% at constant currency to $66 million, up 34% on a 2-year stack. Growth was driven by continued strength in Retail Media onsite. We benefited from the traction of our auction-based…

[Operator Instructions] And your first question comes from the line of Justin Patterson from KeyBanc.

Great. Michael, I was hoping you could elaborate on just how your clients have responded to your agentic products. I know it's early, but would love to hear more about how you're thinking about that. And then as just a quick follow-up. You also called out CTV as a multiyear growth opportunity. What are some of the investments you need to make so that can become more material over the next few years?

Thanks for the question. Yes, we'll take both of those. On agentic, really, I'd say our opportunities are threefold, if I try to categorize them a bit. We have delivered internal agentic products around our current workflows. For example, the audience agents that advance how audiences are built and activated. We're moving towards real-time intelligence audience creation. It's less work, it's faster. And really, the goal is to enable brands and agencies to generate the right audience instantly without delays, and that's in market today. Similarly, campaign agents, right, where we build tools for agencies where they can access our growth engine directly. We can create and manage and optimize full funnel, cross-channel campaigns, just much simpler prompt-based interface really drives the modernization of that product. Second category would be how we integrate sponsored ads into retailers' own agentic chat-like experiences. So think of how we power Retail Media networks today, all retailers are going to need to beef up their agentic capabilities to create easier shopping experiences, and we can power that with a lot of the feeds and technology that we use to serve ads today. And then lastly, the one that we highlighted in the script is the potential partnerships with AI platforms that bring our product recommendation capability into their responses and make them better and more informed because of the breadth of data that we have in our engine combines nicely with the semantic capability that really is the way that the responses are generated in AI platforms. We put a blog post out on this and some social post a few weeks back, and that's the thesis that we're really working on now with one of the large platforms to test that theory and see the results. And then on CTV, I think I'll hand that one over to Todd, and he can elaborate on some of the investments required to scale that channel.

Great. Thanks a lot, Justin. Just a couple of points on CTV. We have been investing on the supply side, specifically in the United States, but also beginning to fan out globally with direct relationships with CTV providers that are scaled. That gives us a basis for bringing cross-channel, full funnel tactics into play where CTV can help with product discovery at the upper funnel, customer acquisition in the middle funnel and obviously, performance, which we're seeing show up in GO, for instance, with social. We would expect the same thing with CTV. So we're doing 2 things: supply side integrations and on the demand side, working with our tactics to make sure that CTV shows up in each of those 3 categories so that our advertisers get to net new incremental audiences that are on CTV can be acquired. And then on the performance level can be reactivated in a variety of different lower funnel tactics. Measurement underpins all of this. And of course, for us, making sure that advertisers see performance at a constant level across their entire marketing mix is a very key point in bringing these 2 worlds together. Those are the 3 key investments, I would say.

And your next question comes from the line of Ygal Arounian from Citi.

First, just a follow-up on the third opportunity like Michael or Todd or whoever, just on the partnership with the AI platforms. And I guess a ton of debate from investors on how agentic Commerce changes e-commerce product discovery and kind of where the transaction is owned. And it sounds like you're kind of integrating to be part of that. Just philosophically, how do you see that playing out? And is there any more you could share? I know you're not going to share the economics specifically, but how that model would work for you guys if more and more of the transaction is moving into that agentic AI platform?

Yes, sure. Ygal, great question. Certainly, the debate in the weekly press week-to-week. Look, we see the monetization strategies for those platforms probably skewing towards a native advertising solution. There definitely has been some headlines here recently with the commerce capability for single product checkout, which we think is interesting. But I think ultimately, those platforms probably move towards a native advertising solution, think of it as the modernization of paid search. And really, our solution works either way. It's sort of not dependent on that. What we're trying to do is bring an API data feed that allows product-oriented responses to simply be better. And so as AI platforms are competing for market share based on user experience and quality of response, data feeds like ours that are proprietary, deep, complete, really start to differentiate the types of outputs that those platforms are able to generate. And so we see them being interested in it sort of no matter what their monetization strategy is. And I'll hand it over to Todd, maybe just to kind of talk about some of the hypothetical economic models that might support that.

Yes. I would say this, we see the affiliate model as being sort of an underpinning to the trading that will occur regardless of the native ad format and how it might present itself, Ygal. I think what's not happening in the dialogue today is enough attention is being given to how a merchant is going to participate and show up in that clearing. So when you don't know what the format is or how it's going to be traded, it's easy to neglect how a merchant might play. So we're spending a lot of time with data feeds, as Michael pointed out, working with our retailers, with our marketplace partners to help them get ready to show up in answers in a way that native formats might be agnostic to. So there's a very big investment on our end to make sure that retailers are ready to play regardless of format. And we think that some affiliate type setup will be the clearing capability for economics, but we're not sure of that either. What we're really focused on now is making sure that retailers are going to show up in Answers the right way. And that's the work that we're doing on the platform side or exploring with different platform partners. So we have a real advantage in that regard.

Great. That's fascinating shift in how e-commerce is going to play out. And then on the Google partnership, that seems like a pretty big deal to me, getting integrated into Google Search. You talked about brand dollars kind of being part of that and opening up incremental budgets. Just expand on that relationship a little bit more about I wanted to [indiscernible] just like how that rolls in terms of contribution and all that into next year.

Sure. Yes, happy to expand on that. Yes. Look, the Google partnership is exciting. We announced that back in August. And it really is another example of Criteo unlocking more demand for the Retail Media category. And obviously, it allows retailers to capture brand search budgets that have traditionally been outside of Retail Media, and it's going to give advertisers true cross-channel visibility into search performance. And so the incrementality comes from net new brands and new search budgets from existing mutual brands. And back to the sizing we talked about, obviously, it's a huge addressable spend at $172 billion. A portion of that will move in. And importantly, the API connection with Google is up and running now. So we will start to see campaign volume flow into the business across Q4, and then we'll open that up to the other regions in the first half of next year. So it's a strong multiyear growth lever, really probably starting more in 2026. And if you just sort of go off a few other things, from a take rate perspective, it's directionally neutral, the same demand side fee, whether a brand goes through SA360 or through CMax. In terms of measurement, retailers can choose whether they want to share click data with Google to enable that cross-channel measurement. And so Google has an incentive to prove incrementality. And importantly, we only share click data tied to Google campaigns, not all click data or broader user behavior on retail sites. So I think a real great win-win here for retailers, for Google to get access to the fastest-growing part of the ecosystem and for Criteo to continue to benefit from unlocking demand for our retailers, which is core to our mission.

And your next question comes from the line of Mark Kelley from Stifel.

I'll stick with the agentic theme. I wanted to get your thoughts on -- I think there's a debate across the investor base, not for just Criteo, but just the Retail Media category overall, where as people start to transact more inside agents, maybe that would be a headwind to Retail Media as we know it today, people not going to that e-commerce site. So fewer eyeballs to see Retail Media ads. But conversely, I think as people transact more inside these agents, it seems like you'll be able to tap more and more into the search budget. I know we just talked about the Google partnership. But I guess, how do we balance those 2 dynamics in terms of what you gain that you didn't have access to before versus what might be a headwind for retail media, if that makes sense?

It does. It does, Mark. And look, again, like a hotly debated topic, I think, week-to-week. I mean, at Criteo, we see agentic as an additional channel. We just don't subscribe to the notion that agentic platforms will essentially swallow or cannibalize the entirety of the commerce ecosystem. So this pattern that you see today of people doing discovery, getting shopping assistance agentically and then jumping into commerce sites to complete transactions, we see that pattern continuing. Now agentic may become a bigger channel in that mix at some point, but we just don't see it consuming it in entirety. I mean that would be a bear case for like all of retail and all marketplaces. So it's just not the future that we believe in. If you did see some of that channel shift, which I don't think has to be zero sum, by the way, retail could compensate by raising CPMs because you could argue that those placements are going to be even more valuable than they would be today. I think what also gets lost is that retailers won't stand still. They'll continue to add additional formats, improve their shopping and site experience, and they're competing for really the right to conduct business on their own channel versus an agentic channel. So I mean those are some of our thoughts. I guess we'll see how it plays out. But I think Criteo has got to play in that either way, right? We'll either play directly into that channel in some of the ways that we talked about earlier on the call or through some of our current means powering retailers as they serve up world-class experiences that give them the right to continue to command traffic and serve customers.

That all makes sense. And then maybe a second question, just on Retail Media. Nice to see Activated Media accelerate in Q3. Possible to parse out, I guess, some of the new wins that you've announced, whether it's DoorDash or other ones, how that's layered in already? Or is that mostly kind of like a same-store sales number with your existing retailer footprint?

Yes. I mean I'll let Sarah comment a little bit about it because we were super excited about the 26% Activated Media Spend growth. And so you have the 2 factors that we've talked about in the past, scope reductions and then lapping the tiered fees. And then really the new things. We had a couple of new wins that just did not scale as quickly as we had hoped for in this quarter. And so that created a little bit of softness. But we think that we'll make that up as those programs get up and running for next year, Sarah?

Yes, most of Q3's activity related to our existing base. And within that, we had some more mature, I would say, customers, right, scaled base. But ultimately, some -- there's a little bit of softness in some of the beauty area. But the new wins that we have are starting to ramp up now. So we only announced, as you know, DoorDash a couple of -- really a couple of weeks ago. So that's starting to ramp up now.

And your next question comes from the line of [ Tim Nolan from SSR ].

Two numbersy questions, I guess. If you could help us understand a bit better as to how you beat so substantially on the earnings line in Q3. It looks like your CXT number was roughly in line with guidance, but you beat pretty substantially on the adjusted EBITDA line. And then as you move your domicile to Luxembourg and into the U.S., are there any financial implications we should be aware of the cost to do this, the tax implications, reporting, anything like that?

Yes. I mean just in terms of the beat for EBITDA, I mean some of that relates to the top line beat on CXT. We also benefited and continue to benefit from the operational leverage of just continuing to make smart investments. And there are a couple of, I would say, more onetime items. So our bad debt reserve are reduced. Our receivables are in really good shape, which also translates to great cash flow in Q3. And we have some shift of marketing spend from Q3 into Q4. So I would say a pretty solid list of kind of why we be, but most of it relates to, I would say, strong operational leverage. [indiscernible] redom? Sorry, I thought you said Q3. So yes, in terms of the redom, we don't see any material costs related to that, and we will isolate those within our filings so you can see what those costs are.

And your next question comes from the line of Doug Anmuth from JPMorgan.

I appreciate you highlighting the 3 major areas of the agentic opportunity. Are there any particular investments that we should be thinking about as you're heading into '26 required to build out those AI products? And then, Sarah, just on the few guidance, I know you shift with the 2 Retail Media clients. Is there anything to call out on the slower ramp in certain new clients that you mentioned? And then as we think of '26, any change to the headwind that you had called out earlier in the year kind of over the first 10 months of the year?

Thanks, Doug. Yes, I can take the AI investment one and then Sarah can follow up on the second half. I wouldn't say that there's any investments that are sort of out of the norm to continue to scale the 3 different categories that we've talked about or at least that we can see right now. Yes, I mean you can see a little bit of it in the quarter. There's some extra marketing costs that's going into launching Commerce Go, but that's probably just like kind of more of an increase on a run rate as we scale that self-service tool in the market. The things like the campaign agent and the audience agent have been developed very much inside the teams as they are today. And then we'll see how the pilot test goes with the agentic platform. I suppose that's a bit of a wildcard. We'll have to see how the test goes and then what would be required to scale that if that's a partnership that's going to take off. But obviously, it would be self-funded with some kind of a revenue model attached to it. Sarah?

Yes. I mean -- and I think just on the Retail Media, we have a strong base of clients, 235, 4,100 brands. We're continuing to see a very strong baseline of our revenue. And there's just a slower ramp-up of some of those new clients in the latter part of '25, which is not unusual. We've seen that in past years.

And same -- Sarah, just to clarify the same kind of cadence that you had talked about through '26 in terms of headwind. Just want to make sure that we're on the same page there.

Yes. Yes, we still anticipate the $75 million. And I pointed out in the prepared comments that Q1 will be the low watermark. So then we had the tiered fees for December '25 and January '26 for our largest retailer. So our expectation is that will kind of -- it will be the same impact of about $75 million that we already discussed.

And your next question comes from the line of Matthew Cost from Morgan Stanley.

Commerce Go is coming up a lot. It seems like the launch is pretty exciting, talking about kind of doubling the number of campaigns from 10% by the end of the year for small customers. I guess how should we think about the contribution to growth for this product, especially as we look into '26 and beyond? Is this something that is sort of offsetting change in behavior for customers in other parts of the product portfolio? Or is it like really incremental in the way that we think it could kind of shift the growth trajectory for Performance Media? And then I have one follow-up.

Yes. Thanks, Matthew. It's a great question. It's a product we're really excited about. You can think about it in a couple of ways. One, there is a conversion of existing clients to the GO tool or to the GO campaign workflow. And in that case, what we're seeing is higher spend, lower churn, better results. And so those flow through in a couple of different ways for us that are great for the business. And then there is an incremental client gain angle as well as we roll out full self-registration, self-service early next year, we're looking to drive client count with that and bring on real net new customers. The other thing that's exciting about GO in that context is how it really makes our cross-channel proposition come to life. I think we mentioned in the remarks, right, we've got 35% of the campaign revenue flowing through social channel. It really is like an expression of that cross-channel, full funnel self-service proposition with a lower cost to serve model. And again, the results we're seeing from clients so far are they spend more, they churn less and they get better performance. So it's really a win-win all around. Do you have anything to add?

Nothing to add to that at all. We're thrilled about it. We're excited. And it is already showing up in several ways as being significant with a lot of headroom.

And I guess, Doug, just to -- or Matt, just to put a punctuation on that, we do see this being a contributor to '26 growth. We'll do guidance on '26, obviously, at year-end results in February. But we do see this being a meaningful contributor to our top line next year.

Great. And then just on the proof of concept with the major AI assistant that you talked about in the prepared remarks. Presumably, proof of concept is sort of an internal test that's not consumer-facing. If it's deemed to be successful by both yourselves and your partner, what are the next steps that come after that?

The next steps will be entirely determined on the quality of the test. Like you said, it's a fairly limited test at this point to determine how the data that we have, the feeds that we have that Michael talked about earlier are contributing to better answers and product recommendations. So it's relatively contained. What we will see once we have that baseline is how we take that to scale to improve the way questions are answered across the larger base of that particular partner. And we don't know that yet. So we're very much in the evidence-producing phase of the trialing, and we'll have more on it, obviously. We're going to go step by step like we do with everything, produce a good result and build.

And your next question comes from the line of Mark Zgutowicz from Benchmark.

Sarah, your net dollar retention was 107% for Retail Media in the quarter. And I'm just curious what was the incremental from offsite versus display? And then just hoping you could provide a little more clarity on your comment about the Retail Media trends being softer but improving in the U.S. and sort of what that means in terms of how we should be thinking about variables over the next 12 months in terms of gross spend versus take rate? Anything outside of the large client transition that we should be thinking about there?

Yes, absolutely. I mean in terms of most of the revenue came from sponsored, and we have seen that, I would say, consistently that sponsored and then our auction-based display ramp-up are where we see most of the uptick. And that's obviously off a scaled base. So there clearly are very fast growth within that, but there's also some more stable growth. Off-site, we now have 42 retailers, and I would say it's just over 10% of our revenue. In terms of what we're seeing for next year -- or sorry, rather of our Media Spend, in terms of what we're seeing next year, we are seeing the continued trend with auction display kind of continue to ramp up really sponsored being, I would say, the kind of lifeblood, if you will, of Retail Media, that's where we're seeing most of the expectation of our growth. And with adding more and more capabilities to that offering, that's where we'll see most of the growth. Off-site will continue to ramp up, but I would say that is absolutely secondary to the incredible traction of our sponsored advertising and on-site offerings. And what we're seeing for next year is that we continue to see our media spend growth across our client base, and we continue to add new clients to that. So we're continuing to see that we're in the right place with our retailers, and there's some new opportunities as we discussed with some of the agentic AI that we're partnering with our retailers on as well. So many factors for growth and obviously, the near-term impact in '26 have very much been offset by good traction across Retail Media.

And your next question comes from the line of Richard Kramer from Arete.

A couple of things maybe you haven't touched on so much yet. One is, Michael, can you speak a little bit to Custom Audiences share and the extent to which you're getting greater social media inventory from outside of Meta properties? It would seem pretty important for the potential resumption of growth or continued resumption of growth in Performance Media. And then maybe, Sarah, you mentioned all these new initiatives, specifically self-serve, and Michael made a few comments about the API relationship with Google, but mindful of the bid switch margin experiences. Can you talk about the relative economics of self-serve versus direct sales and the trade-offs that investors might expect to see in Activated Media growth, but also take rates?

Thanks, Richard. Yes, I'm going to have Todd take the Custom Audiences one as he is the architect of that in full.

Yes, I can -- Richard, I would say this, the way we think about social right now is ensuring constant performance across Meta and open auction and going to CTV, as we talked about earlier in the Q&A. So in terms of other social, we look at global scale partnerships. I think we've talked about the fact that we're testing in TikTok as one of those. And there are a couple of other explorations that we've done. But really, we have plenty of headroom with Meta currently to optimize our setups for constant performance between social and the open auction. So that's why you're seeing the numbers show up the way that they are. So as much as we're excited about expanding the social footprint, we're doing it pretty thoughtfully with a global scale partnership set in mind and then making sure that we don't get lost in single channel setups that other competitors are emphasizing. We're all about holistic performance across social and open and Retail Media as it comes online, and we're very focused there.

Yes. And in terms of the margin, we certainly do see with self-serve, and I think it was answered well in terms of the traction we're seeing on scaling the media spend. So that's obviously beneficial. It just gives us top line leverage and it's optimized automated flows. We're using the same -- I would say, the same capability internally as well. So even if it is more a managed campaign, especially for our enterprise clients, we are starting and continuing to use more capability to optimize all of those campaign setup. So that's all good for margin. In terms of -- you referenced BidSwitch, I mean that is margin generating. So just to be clear on that. And while it's not as high margin as other parts of commerce growth, it does continue to contribute to our margin overall. But yes, we're seeing really strong traction of self-service. We continue to see optimization and what we need is that unlock of media spend, and we have many avenues that we're highly focused on to make sure we can scale that.

Thank you, Michael, Sarah and Todd. That concludes our call for today. Thanks again to everyone for joining. If you have any follow-ups, the Investor Relations team is available to assist. Have a great day.

The conference has now concluded. Thank you for attending today's presentation. You may now disconnect.