Criteo S.A. (CRTO) Earnings Call Transcript & Summary

Hi, everyone, and welcome to the Deutsche Bank Technology Conference. I'm Lloyd Walmsley, the Internet analyst here at Deutsche Bank covering Criteo. We're excited for the next 2 days here at our Virtual tech conference. And I'm excited to kick it off with Megan Clarken, the CEO of Criteo joining us. Megan, thanks for being here.

Thank you for inviting me, Lloyd.

Great. Well, to start off, you're now approaching your first anniversary running Criteo. Can you kind of reflect back on where we sit today versus when you joined and how you think the company is positioned to manage the changing landscape in ad tech?

Yes. It's been an interesting year. I'm sure a lot of people are smiling. It's just unprecedented. But I came in, as you said, at the end of last year. And the first earnings call, I kind of called out what the challenge was, which was the declining core business. And the new assets needing to grow faster and of course, need to turn the core business around. And so I said about doing that transformation. Of course, we got hit by the Chrome announcement in January but sent us focused on trying to overcome the effects of that. And then, of course, in March, we were hit by the COVID issues. Regardless of that, we sent -- we set a really good course for ourselves and a really good strategic plan. And we've been executing against it ever since. So while those challenges have come up, we haven't blinked at all through the time. So transformation as well in progress. And what we've done is set the strategy, set the mission, set the new values for the company, reset the company's structure and get everybody moving in the right direction. And then since then, we've continued to service our clients and make sure that we respond to their needs. Positive momentum from them or challenges from them and making sure that we can help them along the way. And continue to execute against the strategy. So it's -- it is all systems go from here, which I expect to coming into the role, given the strength of the people -- the resilience of the people and the strength of the technology. So we're feeling good about the momentum that we're building.

As we think about your effort to leverage your core retargeting strength from kind of a technology and customer relationship point of view to becoming a more full stack DSP, what do you think are the key assets at Criteo for making this migration successful?

Well, it's a good question. The first thing is that we're not starting from scratch. So if I was -- if the company were having to build from the bottom-up or move to the bottom of the funnel from the top, those things would be real challenges. So we have sort of best-of-breed tech in place. We have, most importantly, the talent in place from running the company at a technology perspective. So we're certainly not starting from scratch. The other thing we have, of course, is we have the relationships and the data. So we have retailer relationships. We have strong relationships with our large clients, our growing mid-market clients. We have access to shopper data inside of our shopper graph that I've never seen anywhere in my career, and I come out of Nielsen. So saying that I hope that adds credibility to the tower of the shopper data that we have. And we have relationships and working relationships with the publisher side of the industry plus the supply side of the industry. And those relationships are -- they're privileged relationships, we're part of their workflow. We're important for them to get their jobs done and produce a really strong ROI. So we're certainly not starting from scratch. So for us, it's about that. It's about the technology that we have. Plus we have a business here in retail media, which I call behind the scenes that Trojan horse, just given that it utilizes everything that we've already got, the carbon niche in the open Internet, particularly around e-commerce and is growing at a clip of sort of 40% month-on-month. We had a great quarter last quarter, of course. And we have a healthy financial position. So once again, I keep pointing to, I'd rather -- much rather be in our position, growing out a full-stack DSP in this environment than anywhere else. So there's a lot of strength there in many different places.

You've recruited a number of new people into the company. What has been your key pitch to get new people to join in such uncertain times?

It's a -- that's a good question, too. And we have brought in really good people from sort of our Head of our Retail Media business in the U.S. who came out of WPP to our Chief Commercial Officer, to our Chief Product Officer, to most recently, our CFO. And then further through the company, there's been some really critical talent come in. So what is it? It's the culture at Criteo is something that I've never seen before, and everybody enters the company as well by sort of the welcome that they get, the enthusiasm that they get. So through the interview process, I think the candidates see that and feel that. Secondly, the strategy is really strong. There's a really strong vision there. There's a pretty clear opening in the marketplace for us and a great story of turnaround. And everybody wants to be part of a winning team. So when they see that story and they see the enthusiasm from the associates, they see the transformation in place. Somebody described that as they can see the Lego blocks starting to fall into place. And when that is actually happening, the execution is happening, then folks want to be part of a winning team. And I think that's really what gets people across the line.

I want to shift gears a little bit into the business a bit more in detail. And before diving specifically in the kind of some of these issues with Chrome and IDFA. Can you just give us a sense of what are the key data sources for the Criteo identity graph and how durable some of those are through these shifting sands?

Yes. That's also a good question because it is -- it is sort of our best-kept secret, I guess, in terms of the identity graph and the data sets that are in there and how we get them. And how they're not. You know there's a large number of data in there that is not third-party reliant. Now, remember that the graph is about over 2 billion unique users in there and their shopper data. So it's not, by any means, small and continues to grow and continues to be refined through the AI and the tech. But to answer your question, inside of there is hash e-mail addresses that we get directly. There are cookie IDs in there, whether they'd be third-party or these days, most importantly, first party. There's IDFAs in there. Of course, that's not new to us and not something that only us use in terms of identifiers. There's Android's IDs in there. There's CRM data. And as much data as we can get to try to pull together a picture of the individual and be able to link those data sets together to get to the same user. And that's what we use in order to get a one-to-one relationship or in a cohort manner, a one to a few relationship and in an addressable advertising world targeting database to get to one-to-many in terms of the look-alike behavior. Now it's kind of -- it's really pretty simple here. The [ demines ] of identify is available. So the third-party cookies, clearly, the IDFAs or the questions around IDFAs means they're just data that we don't have access to, but we have the rest, and we'll continue to use the rest. So it's not -- it's -- and we can -- I could talk about it all day in terms of what's going on around IDFAs and third-party cookies. But for us to be resilient is to use resilient and persistent data, and it comes from all manner resources.

Yes, you guys have talked about the redundancy in identifiers and the deep relationships on both the publisher and advertiser side and that all makes sense and yet Apple's ITP rollout was pretty disruptive. So has something changed since then to make the identity graph more resilient? Or have you been able to perhaps find ways to identify people in Safari despite ITP? Help us kind of understand that.

Yes. I think sort of -- I alluded to it earlier in terms of our best-kept secret inside of the -- what we call a shopper graph. And the shopper graph is deep and wide. So let me sort of give you a couple of maybe interesting steps into what makes up that shopper graph. It collects e-commerce or commerce transactions, and that data is sitting in there. So I'll try and build up the data sets. And that's over 900 billion annual commerce transactions across 100 countries. So think of that in terms of one of the pieces inside of that data set for how we identify identity -- sorry, how we identify context and what people are doing. And that's actually 3x more transactions that Amazon sees in a year. So that is a massive, robust data set inside of there. The identity graph itself is over 2.5 billion Criteo IDs that represent users across the open internet. So if you look at IDFA or you look at third-party cookies, and you think that and you know that Criteo has access to 2.5 billion Criteo IDs across the open internet plus transactions of 900 billion transactions per year. That is a deep and, again, a unique data set that is for us. It comes from over 20,000 online retailers. And what else is inside of there is about 10 billion data on about 10 billion products and services. So I go deep into the strength of the shopper graph, the persistent IDs that I called out before, and then sheer sort of volume of data associated with them. It's like a massive panel if you like. And again, coming from Nielsen, I look at the strength of that and what we can do with that regardless of where pieces of that data might go in terms of third-party or IDFA. That is a really solid graph that I've never seen before. So we'll continue to keep using that. And as identifiers may go away, we -- and we tweak out technology to continue to provide the service for our clients using the huge other enormous data sets that we have coming from various different places.

So let's dive into, I guess, first, the Chrome issue. Google said they'll eliminate the cookie by, I think, 1Q 2022. We've seen this back and forth between TURTLEDOVE and your SPARROW. I guess just at a high level, what is the framework for coming to a solution? And what milestones should we be looking for as this evolves and kind of how prepared do you think you and the industry can be by the time something gets implemented?

Yes. Look, TURTLEDOVE is what's coming out of the Chrome sandbox. And it's basically a framework or a structure of how they're going to pass data out, and also how they're going to process data and what they're going to process on the browser. And so SPARROW has been a very strong attempt at influencing that because it represents the rest of the industry in terms of what's needed to continue to operate the ecosystem as it operates today. And we have, I guess, formed a really strong relationship with Chrome to sort of influence what they will present as TURTLEDOVE at some point in time. This is always a negotiation backwards and forwards and a proof-of-concept, and a way of saying, actually, guys, it would be better to do it this way than the other way and give and take on both sides. So we are the only ones at the table, actually trying to, and I guess, to some extent, successfully influencing Chrome around TURTLEDOVE. And TURTLEDOVE is really about passing cohorts out to the ecosystem so that they're not giving one-to-one data. And so for us, it's about how do we get the cohorts, what's the right data that comes with the cohorts, et cetera, et cetera, just trying to again make sure that Chrome sort of don't go down a route where what they present is not usable to the industry. It also influences reporting. And reporting is critical. The marketers need to know straight away what's happening with their campaigns. And the TURTLEDOVE framework has not been so kind to that. And so again, we know what we need to present to the marketers so they can run effective campaigns and get strong ROI. And so we've been influencing the way in which reporting is run, the frequency in which it's run and the types of -- and the quality of the data that comes out of Chrome. Now on top of that, so that's our influence on TURTLEDOVE, which again, is around this notion of cohorts or small groups of IDs, if you like, coming off the browser. On top of that is an industry effort to create a new way of collecting ID from consumers. And again, it is an industry effort. So there's multiple players that are working on that to sort of get in front of the time that third-party cookies go away. And there's a new way of actually getting consent from consumers directly, and we're involved in that as well. So tons of work going on.

So yes. So what -- so in terms of the process vis-a-vis Chrome, I mean it sounds like Google is working with the industry. Are they being open? Do you feel like you're making progress or the industry is making progress influencing how this ultimately will come about?

I do actually. I do think Google are listening. Google have their own objectives, which are basically understandable in an environment where they need to protect themselves from being exposed to challenges around identity. They own their browser. And so they've said all along that they want to make sure that the changes that they make, don't disrupt the ecosystem. And so they put out a strong model, and the idea is the ecosystem responds to that. And we have responded to it pass-through SPARROW and continue to respond to it. But again, there's a broader effort that says, look, if this is a growing pattern, be it IDFA or third-party cookies, that says we need to protect the identity and privacy of the consumer, then the industry has to step-up and say, what is the next wave of what is the next way in which we're going to collect identity, continue to do our jobs, but protect the privacy of the consumer. And so this is a way that you're seeing starting to evolve that is making progress and it's a positive wave. I think everybody wants to move away from what we have today to some where the consumer feels like they had control of their own identity.

So a good segue, you're working on an open-source revocable IT system. Can you elaborate on what you're building and what would need to happen for this to get widely adopted?