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

Good morning and good afternoon, everyone, and welcome to our video chat with Todd Parsons. We're delighted to have you join us today. My name is Edouard Lassalle, Senior Vice President of Market Relations and Capital Markets here at Criteo, and I'm really pleased to be hosting Todd Parsons, our Chief Product Officer, today. During the next 30 minutes or so, Todd and I will do a fireside chat to talk about Criteo's product vision and road map. And of course, we'll be happy to open up for questions later. First of all, let me thank you for participating today. It's certainly very much appreciated by the audience. With that, and with no further ado, I suggest that we go on with our discussion.

So first off, we'd like to hear a little bit about yourself, Todd, who you are, what you did prior to joining Criteo, and of course, why you decided to come on board last summer?

Thanks, Edouard, and thanks, everyone, for joining us today. Yes, so a little bit about me. I've been in the marketing and advertising space for about 15 years, starting as an entrepreneur, building a couple of companies, the most recent of which called Aditive was bought by Acxiom in 2014; did a couple few years of work at Acxiom, leading product for the Marketing Services division, which was eventually bought by Interpublic Group. I learned a lot about first-party data, consumer privacy and third-party data, as you can imagine, working at Acxiom. And while I was there, we brought LiveRamp to market. We bought LiveRamp about the same time my company was being absorbed. And so I had a front-row seat at building the world of onboarding, as it's come to be known. From Acxiom, I went over and ran product for a company called SocialCode owned by the Graham Holdings Corporation, the media company that Don Graham, who was on Facebook's Board runs. We were, at SocialCode, Facebook's largest media partner, largest third-party marketing partner, if you will, running about $700 million of advertising through Facebook. I learned a lot about the API ecosystem platform eccentricities, also the convenience of working with platforms versus the somewhat cobbled together world of data and programmatic while I was at SocialCode. And right before I came to Criteo, I ran product for OpenX, the supply side, the exchange company, and learned a lot about the plight of publishers. And so the combination of that -- those 3 career stops really got me to Criteo. I just -- I would say was irresistibly seduced by Criteo's unique assets. Our network of first-party data and integrations, direct integrations, both to marketers and publishers, retail -- retailers as well, in the case of Retail Media, the company has been known for the longest time for performance. And performance means delivering advertising, value to advertisers through advertising that works, that is more engaging to consumers and actually performs against whatever the key performance indicator is. So the company has a very, very rich history in performance in the DNA and amazing assets at a huge amount of scale. And that is what landed me here at Criteo and with you today.

Perfect. So lots of great ad tech experience prior to joining us. So now if we -- maybe if we move on with a bigger picture, can you tell us about your vision for the future of advertising? And specifically, obviously, what digital advertising will look like down the road, a few years down the road, and what's Criteo identity road map against that future thought?

Yes. Okay. That's cool. So I mean, first of all, I think it goes without saying to this group, just about anyone that, that digital is just getting larger and larger, the scale of it, all the time. The number of surfaces and just opportunities for advertisers to engage consumers is only going up. We're still just dabbling, I would say, with things like audio and digital out-of-home. We haven't really even touched the connected car, the connected household, and I expect just a massive amount of scale for digital advertising opportunities. I think with that, to your question about identity, personalization and -- is a key part of making advertising relevant, whether that personalization is signaled by identifying a user or identifying the content of user traffics, or perhaps by inferring an interest that, that user might have demonstrated in behaviors prior, the idea of personalization being fundamentally important to making a good advertising experience, good being subjective on behalf of consumers, will be a part of the landscape, that growing landscape I described, whether there are universal IDs available by platforms or OSs or not. So I think what we're giving way to, as third-party cookies disappear and as we experience IDFA and some ad blocking, automatic blocking by Firefox and so on and so forth, I think what we're seeing that's different is the variety of levels of precision of identification that are going to need to be applied to make marketers and advertisers successful has to change, too. So the slide that you just punched up here, Ed, really is evidence of the way we think about helping navigate our customers through this tricky time of platform identifiers disappearing. And there are really 3 parts to it. One is making sure that we take the best of our first-party data connections, with publishers, with retailers and with marketers and provide an easy way to make those first-party data sets interoperable to replace a third-party cookie. That's a very key piece of product work for us and -- but it's just one. The second, in the middle there, is a little shot of what we're doing with Google. And we're just starting to experiment. I actually think there's quite a bit of promise with what we can do with Google around the privacy sandbox, and they've been very engaging with us to innovate, as evidenced by their consideration of our SPARROW proposal, and now as we go into test design to preserve some of the better aspects of interest group targeting optimization measurement, I'm encouraged that we're going to have a partnership that bears fruit in that middle box there with interest groups and cohorts. And on the right, something I'm personally really excited about is what I like to call next contextual. The -- my first business was a contextual business, and I think not a lot has changed. Natural language processing, semantic analysis has gotten a lot better since 2007. But not a lot of innovation was really done otherwise around meeting extraction from either pages or now a variety of multimedia formats, video and so forth. What we're doing at Criteo is particularly exciting to me because of our first-party data assets, we are able to append a commerce signature to every one of those pages or every one of those interactive formats. And that comes from us being able to look at a couple of billion dollars worth of transactions and on a daily basis. And it comes from us looking at a very large universal product catalog and being able to attach meaning from those assets to every page on the Internet, which is a very different way to think about contextual than what we're doing today. So all of which is to say like our game is navigating marketers through this changing landscape and providing tangible, touchable experiences now that broker on our assets.

Thanks, Todd. That was a long answer to that. That sounds great to hear that identity gets you so excited. So just going beyond the multifaceted of privacy, what would you say are the key trends that are shaping Criteo's market today and even more importantly, tomorrow?

Well, yes. I mean, for us, e-commerce and just time spent online, are pretty simple. We're all sitting here in our homes glued to whatever surface we're on. And I expect that a lot of that is here to stay, as much as I think like others, I want things to return to normal, we're going to spend a lot more time in front of screens. And we're going to spend a lot more time because we've developed the comfort with e-commerce transacting online, which is super important for us. It does a couple of things. One is it plays to our strength of having served retailers who broker in e-commerce and other advertisers that do the same, travel and so on and so forth as it returns. But the other thing it does is it means that a lot of advertiser dollars, trade dollars, co-op dollars are flowing in to the space where we've been expert and we have relationships and we have our network. So those are a couple of really important things that are kind of dovetailing the e-commerce and time spent mention that I just made. And I think above all, advertisers, more than ever, this is a complex landscape to navigate. We just talked about identity, like you said, long answer because it's not an easy topic to cover, and you can imagine what most marketers and advertisers think, they really would rather not think about it. They would rather have the help of companies like Criteo.

So maybe moving to more recent discussions we've had. As you recall, on our Q3 earnings call, we talked about Criteo's Commerce Media platform. Can you please spend a bit of time on this, Todd, and help the audience understand, first, what Commerce Media is and what our platform brings, and maybe be even more specific around how our program map will articulate both our marketing solutions and retail media offerings.

Yes. So I think -- I mean, Commerce Media, I put about at least one way, the way I might describe it is the complete mix of touch points in a buyer experience, where a consumer can be identified in some way, shape or form anonymously, pseudo-anonymously or directly and addressed so that as they discover and consider products right to a point of purpose that, that's a more efficient path. So for us, what that means is the Commerce Media platform is really drawing on the whole of our business and some shared dependencies between them. If you think about what we're doing with Retail Media on the right of the slide that you just pulled up, Ed, we have a Retail Media platform. We're shipping a tremendous amount of insights back to retailers as they find revenue from our on-site advertising. We're helping them with audiences. We're helping them soon with personalization, possibly with merchandising and the like. And we're helping those same retailers find the same audiences offsite so they can capture more of those trade dollars, they can capture more of those co-op -- that co-op spend, and they can have a relationship with direct advertisers to their own inventory. On the other side, the conventional marketing solutions business at Criteo, which has been all about marketing outcomes, is now kind of going towards APIs. We're exposing the power of what we're doing with APIs in this business. I'll talk about that in a second. And the first-party media network that I described is stitching together a first-party data is incredibly powerful as an enabler for identity, and it reaches across to Retail Media. So in the middle there of this slide, you basically see where the 2 businesses have great synergy and some dependency. I mean they work together better because of off-site supply and audience, because measurement is made possible through identity, and because in both businesses, we're engaging the walled gardens and data that comes out of those environments, which can be a powerful way to stitch together a consumers' buying experience.

Thanks, Todd. That was really useful. So if we look at maybe today, can you help the audience understand where we are on that path of building the Commerce Media platform? And maybe what are the key priorities for Commerce Media Platform in '21 and beyond?

Yes. Okay. That's great. I mean we're -- well, first, I think just from a sheer product perspective, we have a lot of work to do to move the business forward. We're very ambitious. But I would note that unlike most other companies that I've been exposed to as partners in ad tech or that I've worked for, we have assets that others don't, that they can't -- they just can't broker on to do what we're doing here, this slide that you just punched up. So if you think about where we've been, we've been all about remarketing use cases in a display modality, and that was a great business for the company. Last few years, we've been innovating with Retail Media. I think everyone on the call is familiar with that, the success of that emerging business, and it really is growing pretty rapidly. At the same time, we've been getting away from the display modality into more interactive formats video, native and early work in CTV going into this very year that we're sitting in at the end of. What we're doing for '21 starts getting pretty ambitious, as I mentioned. One is, on the right-hand side there, We're going to be working to securitize our data assets, the rights that attach to them in such a way that we're 100% privacy-first, that we are anticipating a purpose under GDPR. We are anticipating use under CRPA. We are helping each of our marketers, our retailers and our publishers to share first-party data in a way that they can track, that they can assign and control purpose to and they can delete whenever they want. I think it's probably the first time anyone has tried to do this across as vast a network as we have. And as I mentioned, if I think about other companies I've worked for that have their own really unique advantages, Criteo probably sits in the most special position between each of these constituents with the trust to touch all of that data and to manage it in a way that everyone has come to appreciate is needed. So that's -- '21 is going to be 3 things, really getting that first-party media network, the mechanics of that in place; two, beginning to open up our assets meaning audience, I mentioned measurement, the way that our bidder performs is very tuned to performance, and personalization in APIs that other partners can enjoy. That's a second very important piece of what we're doing. So in the past, where we've really serviced our own business, our own marketing solutions business, we've always had a lot of demand from partners that want to get at those things that make us unique, and we're beginning to open those up to select partners, much in the way that Facebook opens up its capabilities. And then third, I mean, it's all about the formats that I talked about in the prior page, the advertising techniques, adding contextual, upping our game with CTV and OTT and video. And that's not just a hand wave. It's not just launching the format, it's making sure that commerce outcomes can be drawn from the formats. So you can imagine the way we think is, if the video isn't somehow informing a purchase decision in a trackable way to a commerce outcome, it might not be the right thing to subscribe our client to. And if you imagine it that way, as it relates to any formats that we're rolling out this next year, our third kind of initiative, then you'll kind of get what we're doing philosophically with product. It gives us our purpose.

Thanks, Todd. So if I heard you correctly, 3 priorities at the moment, at least for '21: privacy-first media network; audience API, many of those; and new formats, including contextual and CTV. So talking about CTV, we all know it's a very hot topic in the industry. But can you walk us through Criteo's strategy there, given it's a bit further away from the core business that people sometimes know us too much for?

Yes. Yes, that's -- it's interesting because it is -- it's like a whole story. CTV as a whole story in of itself in other parts of the market, if you look at platforms like Roku or you look at The Trade Desk, these businesses are showcasing and banking a lot of their growth on the -- on CTV for obvious reasons because it is becoming the -- where most of our attention as human beings is spent. So it makes a lot of sense. Unfortunately, or maybe fortunately, CTV has been mostly an upper funnel phenomenon of kind of a replacement of linear, if you will, my opinion. And so what we want to do is attach that attention that goes to CTV with commerce. As I mentioned before, we want to stay true to our knitting product-wise. And so one way to do that is to provide a way for our partners outside of the company or our brands that we service to do better measurement of their CTV initiatives as they relate to commerce. Let me make a tangible example because that's a little bit of a tough one. So CTV, I think about all the advertisers I've worked with the last 10 years, a good number of them want to own the household, so to speak. They want to know that regardless of whether which person is the purchaser that they're saturating the household with what it is that they have to sell. And CTV is a really good way to do that. It's a good way to get to a variety of devices and reach everyone who lives in a home. The problem is actually attaching those exposures, those ad exposures, to a downstream commerce event is really hard. So one of the things that's very central to our CTV strategy is addressing that problem, making it possible to measure a household of CTV consumers down to the purchase, which is made as part of that $2 billion a day that we watch.

Sure. So thanks for that. And moving on to another of our priorities about contextual. Can you help us understand what our contextual offering is going to look like, Todd? And what kind of performance can we expect when it's rolled out?

Well, I touched on it before. We have this kind of a hidden strength in machine learning and data science. Most of our ability to drive performances using prediction models, I don't think that's a secret to anyone. And we're really, really good at it. So those prediction models are tuned based on running trillions and trillions of events and impressions. And we're going to have to do the same thing with contextual. The difference being now is there are billions of dollars of contextual spent, I think, some $35 million or so in the U.S. depending on who you follow, in addressable markets. And most of that contextual is based again on extracting meaning from a page and assigning it to a category, a category of yard care people. Whereas for us, we're going to be looking at, did the page actually sell a lawnmower in the last 30 days. And if it did, how many people traversed that page were of a household income that would be able to or want to be in the market for the ride-on mower or versus the push mower and be able to help people find the right product that they want to buy. So the way that it's going to perform, I believe, will be entirely attached to how well we attach that commerce meeting to each page. But I think in the spirit of Criteo, it will only get better as we run impressions through it. We're counting on that. And so it would be early to say will it perform as well as, for instance, remarketing. We have some pretty bold claims being made by other companies out there on that topic. And I feel like we'd love to give them the Pepsi challenge next year some time, see how we do driving purchases versus them.

It sounds like even in the -- even in contextual Criteo, strong commerce data can definitely add lots of value compared to other players. So maybe switching gear and going back to the identity topic, because I know some of our people here on the call are interested. We've recently partnered with The Trade Desk on UID2.0. Can you explain what this partnership is about and what the expected outcome might be?

Yes, I get that a lot. And there's a good reason for it. Trade Desk, I think a lot of folks have thought is competitive to us. And in some ways, that's not untrue. The Trade Desk partnership was really predicated on one1 fundamentally important thing, and that is that in a world where there is no third-party cookie, there is a loss of interoperability of data between buyer and seller, between marketer and publisher and the consumers that they both service. So The Trade Desk partnership was about remedying that problem -- is about. It's very active, I might add. I just had to stand up with them yesterday, and we are actually working to not only co-innovate but to build product together. There are 3 things that we're building, what we are working on together. Well, first of all, what they've done is they've built the UID2 service, an endpoint for us to hit and to take down an ID. What we're doing together is 2 other things. One is we're building a lightweight log-in capability for consumers, either to say, yes, I'd like a personalized ad experience, or to say no, much like they sort of do with CMPs of today, but in a way that doesn't use third-party cookie and still is able to work across domains. So we are building the -- we are designing and building that consumer experience that where people actually interact on a publisher site or on a retailer site with that personalization choice. And then there's a third component, which is what ties the UID service and that log-in capability together, which we call the transparency portal. That is basically just a place to, hopefully, in the simplest way possible for consumers to say, this is how I want to be identified for personalization or not. So the concept is to go a long ways beyond opt-out, a long ways beyond the idea that any of us as a consumer have to understand that Criteo is an ad tech firm that is targeting them. This is much more about, "Do I want to have a clean web experience where my ads are frequency capped, where I'm not being overexposed to the wrong things or maybe that I don't want to play at all." So those are the 3 components, Ed, and we're building them and actually getting ready to test them this quarter, which is pretty cool.

Just to follow up quickly on the transparency portal, Todd. Can you maybe elaborate on the compatibility of that -- of the IT at the end of the day with regulatory frameworks, whether it's GDPR or CCPA? Can basically the portal replace or act as a constant banner, as we typically know them, for example, here in Europe?

That's how we're designing it. One of the great things about Criteo I've come to appreciate is we operate under the banner of GDPR first. Okay. So in a way, that was a pleasant surprise to me because I've been operating under CCPA, and now CRPA. And that's a different standard, if you will. So yes, the -- we're certainly designing with that in mind, Ed, and we're designing for great scale. Part of the reason that we got together with The Trade Desk was because it's a big bet to address consumers on this privacy and personalization issue across many, many, many websites. And our thesis was, it is going to take the collection of folks that have a fair amount of influence that are bound by the same principles and that want to comply with all of the myriad of regulatory frameworks. And that -- and Trade Desk, we saw, was doing that. And so we got together, we got our CEOs together, and we said, "Let's give this a whirl. Let's try to do something that preserves the Internet and advertising in the process." And I think we've got a good chance of doing what you just described.

Thanks, again, Todd. So maybe a last question to allow for Q&As from the audience. But closing on a high-level note, what would you say makes Criteo so different in the ad tech space today and obviously, tomorrow, Todd? I'd like to think of ourselves as the Amazon advertising for the open Internet. But what's your take on what makes us so different, Todd?

Yes. I mean -- first of all, I love that comparison. We have -- I have a very large group of analysts at my disposal who are always looking at the traffic across our network, the amount of shopper data we have versus Amazon. And I'll tell you, the numbers are eye-opening. We are well positioned to be the Amazon advertising for the open Internet. We have the assets to do that. And I think the things that really are setting us apart now that actualized this vision or enable us to is we -- is that we've taken an open approach, just like we did with TTD on the Universal ID 2 (sic) [ Unified ID 2 ] initiative, we are opening our assets to an ecosystem of demand, an ecosystem of applications that work with other marketers that we don't -- that can bring dollars into our ecosystem through APIs and where we can feed our ecosystem and strengthen our stewardship of first-party data, which is what stitches all of this opportunity together. What makes Amazon possible is that they don't have to do that stitching. They control the platform. For us and anyone who wants to compete, it has to be built. And so for Criteo, we have a unique position to have most of the tools, and it's more about pulling together those tools in a way that can be touched. So that's what those partnerships are all about. And you should -- everyone on the call should look for us to do a lot more with partnerships this year, and I'm very excited about what's ahead in that vein.

Thanks again a zillion time, Todd. So I mean, I'm trying to be conscious of time. That was really awesome to have you, but I think we want to leave some time for our friends asking question here. So please raise your hand. I mean, Clemence, you can walk us through the process, please.

[Operator Instructions] I'll unmute you, Tim Nollen, from Macquarie with the first question.

Okay. Can you hear me?

Yes.

Yes, we got you.

I had to click unmute myself. After [ 9 months ], I'm still getting used to this stuff. I really appreciate you taking the time, Todd. Very informative. My question is to dig a little deeper, if I can, on the measurement issue. I wonder -- now I'd like you to get as technical as you can that we can still handle. But my question is, there's so many different ways to measure. And obviously, your CEO comes from the big measurement company. And I'm just wondering how much are you using Nielsen type of measurement, whether it's DAR or any of the new Nielsen ONE they're talking about? But I'm sure it goes much beyond Nielsen because even cookies and now first-party identifiers, Unified ID 2.0, that allows you to track and measure viewership as well. Is there a way you can just help us understand with a little bit more technical detail how you're doing this?

Yes. I appreciate the acknowledgment of like the measurement got a DNA that Megan brings to the company. So the way that, that's kind of impacted our thinking about product, I mean, it includes working with DAR, but I want to go beyond that for a second, when you asked, I'll tell you a couple of different things. One is, we feel like we are -- should be the great measurement enabler, not the great measurement solution. So working with Nielsen kind of flips the script a little bit from us looking at DAR and more providing them a combination of events and logs and possibly aggregates that they could play back in their own products. So -- and I think they're more like Visual IQ as opposed to us ingesting or working with DAR. I think of an output to Nielsen, which would give them the opportunity to do multitouch in a way that their clients appreciate already. So we have a couple of things just to get more specific again that I think are important to convey. At the top of the funnel, if you will, to use the Nielsen comparison, we have a very large and active panel of ad response from consumers. I don't think people know that we're getting hundreds of thousands of hits on the ads experience from being exposed to an ad, did you like that ad? Did you recall that it was [ Jasper's ] and that sort of thing. So the idea that we can build kind of our own panel across -- that is commerce specific across our network, at the very top of the funnel, is something that we're looking at productizing more. In the past, we've merely used this as a way to advise our customers about their creative and how they might think about media planning. So you can imagine we want to do more with that upper funnel piece of feedback loop, if you will, a very large panel. In the middle, it's all about incrementality. And I mean, you saw on an earlier slide, I said we want to automate incrementality. And that's ambitious. For now, we'll settle for measuring it. And again, there are 2 problems with that. One is, and you mentioned it, having enough identity to be able to infer the outcome better than like a mix model or a panel would. We've been held to an incredibly high standard. Everything is conversion. And then it's for the market to say to us, will we agree or disagree with the way that you're giving yourself credit for conversion. You know the drill there. What we're doing now is we're saying, okay, look, all bets are off on a lot of the funnel. So instead of taking aim at our measurement practice or our framework, we will build an A/B capability, and we will give you the log data. And if you're P&G and you're doing that upper funnel to mid-funnel thing, great, here's the data. And if you are interested in the whole buyer journey, we'll stitch that together and package it up in a way that you can do MTA. So that's the way we're thinking about it right now. And this is a big initiative for us. We have some of our brightest people on it. I'd go as far as to say, we have to figure out the measurement problem, Tim. If we don't -- and so I'm not saying it's done, but I am saying that we have an amazing head start and it will mainly broker on providing events and logs and package data to partners that do a great job of conveying the value of the advertising that's being delivered. Hopefully, that helps. There are a lot of other questions in the log.

The chat, yes.

Yes.

Do you want to take one of these? Or I can unmute someone else.

Go ahead. Take your pick. You're driving.

Okay. Nick Jones from Citi, I'll let you talk right now.

Great. I guess I want to ask a little bit around ID. We've seen Unified ID 2.0, Criteo is partnering with The Trade Desk. Can you talk about what this is going to look like in kind of a post-cookie, post-IDFA world? Is this largely going to be first-party cookies with hashed IDs? Because I think Apple kind of quietly responded that they're going to not allow kind of hashed e-mails as an ID source on the iPhone. I guess I'm just trying to understand what the puts and takes are as you see it for identification in a post third-party cookie, post-IDFA world. And then I guess the follow-up would be, are there other examples in other industries where you're trying to replicate this type of coordination between this many publishers and this many advertisers to kind of share use -- sharing use data?

Yes. Thanks, Nick. So a couple of things. I think you kind of nailed it. The bull's eye right now for stitching is brokering on first-party cookies because that's the most viable way to get to scale. So of course, we're doing a variety of that ourselves in terms of portability, but we're also looking at different ways to write to local storage and -- in a variety of synthetic IDs which could anchor to what's written there locally. So the advantage that we have is we're on page everywhere. We're interacting with our customers pretty deeply on page. But what we're not doing yet is we haven't gone -- I mean I think, it might be bold to say, but the future is likely a server-to-server world, okay? And we have to make it efficient. But the last thing we want to do is to be on every call talking about Apple's next move or Google delayed and hurrah for everyone. That is not controlling our destiny nor is it leaning into the relationships we have, the trust relationships we have with folks. So I can't talk about everything that we're doing beyond first-party cookie because that wouldn't be prudent here. But you will hear from us on other ways for us to do the syncing that you're inferring might be blocked, at least in part -- fortunately, a small part, I might add, by web kit, if that's where you were going. And I think server-to-server will play a very large role in that looking forward. And you asked one1 other question, now it's slipped my mind.

It was just the coordination, right? Like you're trying to get a bunch of publishers and advertisers to kind of share data or pool data to help everyone out. I mean how hard will this be to do? And other examples in other industries where something like this has been done that can be replicated.

No, thanks. Yes, I mean there are a lot of other examples. Most supply chain management companies are -- enables our good corollaries. One that's a lot closer to home, and this kind of takes me back to the direct marketing roots that I laid down at Acxiom is Abacus. I mean -- or any off-line co-op, which I think most digital people are like, "Oh, those household co-ops and pretty easy to pop the direct response or the direct marketing folks." Reality is now that the need is there, these models actually look a lot more tempting than they did before. And what I'm seeing that's different from when I was at Acxiom, where we were doing this stuff one-off, when we lit up a partnership between Citibank and -- if Citibank was underwriting Marriott Rewards credit card, which they did at the time, it was just horrendously painful to get their data combined. Just those 2, never mind onboarding it and matching it and getting ads sent out into the digital ecosystem, that was only 4 or 5 years ago. And now all those same parties, even in really regulated spaces, are saying, including your -- I used Citibank, I'm sorry, I had to. But the -- what I see now is compliance hasn't changed, but motivation has changed a lot. So people are looking to test things, try things. As long as they know that they're working with a trusted partner who's not going to fumble with their most treasured assets, they want to do it, they want to do it a lot faster because these are kind of existential threats to advertising. So I gave you the Abacus example. It's not the best, but supply chain management and EDI is like a better one. You look at it and as those systems have become real time, they look a lot more like the ad tech ecosystem. And so the models, the patterns are there. You don't have to be that creative. What you do have to do is anticipate the moves of forces that are greater than you and see beyond those so that you don't end up with a solution that's dead before you ship it.

I would suggest we'd maybe take a last question.

Matthew from SunTrust.

Yes. Todd, greatly appreciate it. Maybe 2 quick ones, if I could. The first one on the portal that you talked about a little bit that you're working with The Trade Desk and probably others as well. Could you just maybe talk a little bit about what a time line could look like there and just what kind of buy-in you need? What constituencies do you need, buy-ins or whether it's regulatory or web browser? I'm just kind of curious how that works. And then just secondly, if we take a step back and think about Criteo's, the number of impressions that they put out there, where do you think we'll end up in terms of the percentage of impressions that will ultimately be 1-to-1 targeted as opposed to maybe contextual or batch cohort base targeted in the future? I'm just curious maybe what the before and after will be in terms of that mix, again, before the cookie goes away at IDFA and all this. I'm kind of curious what the before and after looks like, any color there.

Yes, you got it. That's a burning question, that second one. I'll take the easier one first. Just the time line for testing transparency portal and so forth. What we're doing right now, I mean, just to take it real time is we're going into user testing, not live testing, but we're going into user testing because we actually don't want to slap a product into the ecosystem. We want to build something that is going to be simple to understand. If it doesn't delight consumers, at least it won't be poorly understood. So that means that the log-in experience and the options and the transparency portal will go into user testing. When I say live testing, I mean on publisher sites and so forth, we have no shortage of publishers and retailers that want to test with us for the aforementioned motivation that I pointed to earlier. I don't think we're going to be there until in Q1, but we will definitely be looking to push towards live tests. It's a big push, okay? Like I said, it's weekly stand up so a lot of work going on, user testing between now and year-end, like we indicated to market we'd be doing. And then once we have a pretty good idea of -- that we have the design patterns right, we'll go to live testing. I expect we'll do our first live test sometime in Q1. And then after that, I think it really just depends on how we do. Like most start-up businesses, we're going to have some cycles of feedback and change. And I think we'll probably have those -- if I had to guess, we'll probably have those at least understood by midyear, and we'll be into something that's more production mode like in the back half of next year, but that is a complete -- that is not grounded in fact at this point. It's more of a road map of sorts. Hopefully, that helps. In terms of logged-in day, I hear a lot of numbers. I hear 20%. There's just a bashing that goes on, which is, you'll never have all the logged-in users. And I mean it's really a preposterous conversation. I think, overall, the ad dollars have to go somewhere, right? Advertisers need to acquire customers at a price and they need to delight them by bringing them back. That's not going to change. So that would be the first thing I would say. I think the 20% number that you've heard people talk about is quite low. I do think there's going to be a lot of nonreal time processing needed, a lot of mapping, a lot of server-to-server stuff. And it's hard for me to predict how much more can be stitched together. But look, when we started getting LiveRamp out there in 2000 - I guess, this was 2015 really in earnest, we were looking at a 30% match rate back then. No one was talking a whole lot of smack about that, okay, because there just wasn't another way. So the way we're thinking about it, and of course, today, depending on who you believe, a file might get 70% or more. So my point is this is not a, we're going to get into the game and find ourselves at 80%. What we aim to do is to tell the market how we're doing, to explain the approach and to expose the amount of logged in -- pardon me, of targetable known users that are not relying on a third-party cookie. So we're going to baseline that. We're going to come back to the community and make sure that, to the extent we can disclose it, we will. And I wish I could do better than that for you. But I think it's going to be better than 20%, if I were to bet against the pundits.

No, that's helpful. Todd, maybe just one quick follow-up. The constituents that need to buy into the portal again, do browsers need to buy in? Does regulatory need to buy in? I'm just kind of curious, I mean who do you need the thumbs up?

Yes. We -- I mean we want everyone to be happy. Whether they're completely bought in as a dependency is a different thing. Regulatory, for sure. Everything privacy by design, everything really with an opt-in perspective as opposed to opt-out. So kind of going the further mile there, at least in the current design. They're important to say that there are still design considerations we're working through and as such things are subject to change. But we want the browsing experience to be a great experience. As such, we'll use the browser control surfaces to make sure that someone can easily get to the transparency portal without having to go away from their usual business. So -- but in terms of needing permission, this is about consumer permission and making sure that this is collected. And the power of that is that everyone wants that to work. Browser want that to work. Publishers want that to work. Our retailers, marketers, everyone wants that to work. The regulators want it to work. So the way I look at it is, unlike ad tech in the past, we're kind of going down this road where we have to be very consumer-ish with what we're doing because then it's less about the buy-in of the kind of the ad tech crowd and more about what are 2 million consumers saying they have spoken. And so that's the vision. That doesn't mean we're not working with every one of those parties because we are. And like I said, we're partnering with folks that we might be competing with ad dollars for it to prove it.

All right. I guess that closes the call. Thanks a lot again, Todd, for your time. I think that was really helpful. Hopefully, we will get good feedback. Thanks, everyone, for attending today. And the IR team is here. Should you have any follow-up questions or for those who ask questions, please send them on, and we'll make sure we can get satisfactory answer. Thanks, everyone.

Thank you. Thanks, Ed. Thanks, everyone. Have a great day.