DoubleVerify Holdings, Inc. (DV) Earnings Call Transcript & Summary

All right. Well, good morning, everyone. My name is Matt Cost of Morgan Stanley U.S. Internet team, thrilled to be joined this morning by Mark and Nicola from DoubleVerify, the CEO and CFO. Thank you so much for being here.

Great being here.

I just need to quickly go through the disclosures. For important disclosures, please see the Morgan Stanley research disclosure website at morganstanley.com/researchdisclosures. If you have any questions, please reach out to your MS sales representative.

Okay. With that out of the way. Maybe let's just start high level, and I'll go to you first, Mark. For the people in the audience who are maybe newer to the DV story, let's start there. Talk about an overview of what you're doing, how you fit into the ecosystem? And maybe most importantly, how things have changed for your business over the past couple of years?

For sure, for sure. I just want to say to people on the dark side of the room, you can come to the light, come over to the light. Yes. So in a nutshell, what DV does is we work with advertisers to ensure that their digital ad transactions, so when they buy a social media ad or an ad on connected television or an ad in the open Internet, we ensure that, that transaction, that ad transaction is brand-safe. So the ad ends up in a brand-safe environment. The transaction itself is fraud-free, so the ads aren't fraudulent, and they're viewable and viewable by human beings. So we provide basically verification services. We ensure the integrity of that ad buy and that it aligns to that advertiser's prescribed rules and criteria around brand safety, viewability, et cetera. We do this across, as I noted, social media networks, so places like Meta and TikTok and YouTube, across basically every major connected TV platform, the open web, mobile app, mobile web. If it's a digital environment, we're either there or going to be heading there soon. So we do that pretty broadly. We do it for some of the biggest brands in the world, folks like Unilever, Mondelez, Colgate, large media brands and basically 500 of the top 1,000 brands around the world, we work with at some level. How our business has changed and how our software and how it's seen by CMOs and marketers has changed, is that we started off really protecting ad spend, ensuring that those ads were viewable, that they were in a brand-safe environment. And that was the real focus. So think of it like we were really protection. We're almost like ad integrity insurance, right, making sure that this is protected. What's evolved over the last few years is -- what advertisers saw is when they took garbage out of the system, when they took the stuff that wasn't viewable or that wasn't brand safe or was fraud, what they found what was left remarkably performed better, right? So we kind of started digging further into that path and saying, "Well, if we're removing stuff and helping drive performance, what else can we tell them about that transaction that will help them drive performance." That led to things like pre-bid solutions like ABS, which allowed us to pre-filter certain very granular targeting segments, allowed us to build tools like Attention which now measure how users are engaged to help drive better outcomes. So we really have evolved over the last few years from a protection business to a protection and performance business that is really driving better outcomes for advertisers.

That's a great overview. I want to stay high level for a little bit before we dive down into a bunch of interesting things you touched on there. I mean let's talk about what you're seeing in the ad market. I mean it's kind of a mixed environment out there. It felt like we were at a moment of really great uncertainty last fall, things have gotten a little better, but it's still -- there's areas of strength and weakness. So I guess in your conversations with advertisers, what are you hearing from them in terms of like their level of confidence? Are there pockets of strength or weakness that you're seeing? And then what are advertisers' priorities this year?

Yes. I think some of the characterization of the market you just mentioned there is true, which is it's a little bit all over the place. Last year, for sure, people entered the year very concerned, right? There was concerns about interest rates. War in Ukraine was still top of mind. There was lots of challenges going on. I think a lot of those have been cleared out. The interest rate thing kind of -- everyone just kind of takes it for what it is in the ad world now. And I think a lot of those macro issues still aren't weighing on advertisers. However, it doesn't mean that they have full scale started blowing out ad spend again. We're not looking at 2021 post-COVID, when everyone just started spending. I think that from a -- it's really advertiser by advertiser. You're seeing some companies doing very well, leaning very heavily in. You've got areas like retail media networks really starting to explode. There's still a decent amount after the writer strikes and the actor strikes last year, you've got theatrical advertising starting to come back. So there are pockets that are showing bright spots. But all in all, I wouldn't say it's like a super rosy environment for advertisers, but much better than it was last year.

So let's talk about your outlook for '24. You talked on the earnings call about like a double-digit deceleration in topline growth from 4Q to 1Q, followed by your expectation of a reacceleration thereafter. Walk us through what's happening in 1Q to drive the slowdown in your expectations and why you're confident in a reacceleration?

I'll take it. So first of all, there is some seasonality to our business. So looking at Q4 to Q1, there always is a deceleration. What we're seeing in Q1 is slow spend from a few of our large advertisers. So on average, our top 100 customers are spending $3.7 million with us. The top of that funnel was spending a whole lot more with us than that. And there are some specific issues with some of our large advertisers that has led them to spend less. They haven't turned off our service. They just have issues endemic to their companies that is leading them to spend less dollars, [ mature ] dollars in the market. What's important for us is that they didn't turn off our service. So when the advertising spend comes back, we will be able to see it straight away. These issues are endemic to the companies. One -- as an example, one of our large clients is in a situation where they have to close 100-plus retail stores, and it creates issues within the company that means that they're spending less. That has lowered Q1. What we're seeing and what we're hearing from our customers is that the spend as a cohort will come back, right? So we've assumed that the spend will come back in Q2, Q3, Q4 based on the conversation that we're having with our customers. It is only a handful of customers. We have a lot of customers. The overall outlook is positive, as Mark was mentioning. We don't see this as a macro issue at all. We just see it as a moment in time for a few of our customers.

Got it. Maybe let's break down the business a little bit into measurements, activation and supply side. And I'll maybe stick with you, Nicola, on this. Like which segment has the most runway of those 3 that you would call out? And what are the drivers between the 3 businesses?

Yes. So let's just take a bit of a step back to understand where we are today based on where we were. We started with measurement, right? So 90% of our business is measurement and activation where the advertiser pays us directly. We have a side business, which is a supply side business, which is a lesser part of our business, where our data is used by platforms to preclear kind of basic level brand safety, fraud and viewability before that inventory is then sold to the advertisers. So ultimately, the advertiser is really our customer. Within that 90% between measurement and activation, our business started with measurement. So measurement is -- our data is used once an ad is run, we can tell the advertiser how it did. That is -- that's where the business started. A few years ago, what started then is pre-bid, which is, can our data be used even before an advertiser bids on an ad. That is obviously a very exciting area of the market. This is where most of the dollars are going now for us. It's about 60% of our business that is our data used ahead of actually bidding on an impression. The -- so with that in mind, the drivers of the growth that we see right now, social is only on the measurement side right now, right? So they have walled gardens where the walled gardens are providing a level of brand safety that we can then measure for our advertisers. So social is in measurement. That is the area that we see the most growth in the short term. I'm sure we'll talk about Meta and brand safety, brand suitability. That's the driver. Pre-bid, we have premium-priced products there that are unique in the market, that are continuing to create a lot of growth for us. So ABS, which Mark already mentioned, is premium-priced product, 2/3 of the growth that we saw last quarter was from existing customers continuing to use our service on more and more impressions. That's a 5-year-old product. So we have a lot of drivers. I think short term, social is really where we see a huge growth.

Maybe let's stick with social. The partnership with Meta, I think is a major focus. It's an exciting announcement. The expansion of that partnership that you announced earlier this year. How are you quantifying that opportunity for investors and what results are you seeing so far?

Yes. So the way to think about the opportunity, it's actually pretty straightforward. So brand safety and brand suitability is what was not available on Meta until a few weeks ago, and it has been available on YouTube and on TikTok for longer. So if you think about the top 100 customer and adoption of the product, YouTube where brand safety, brand suitability has been available, along with fraud and [ anti brand draft ] and viewability, 90 of our top 100 customers use our product on YouTube. TikTok, which basically just started within a year, 40 of our top 100 are already using the product. On Meta, 50 of our top 100 use the product. The other 50 were basically waiting from brand safety and brand suitability to launch, to turn on the service overall in Meta. So the opportunity is basically that other 50 of the top 100 to turn on the product. It's going to take some time. It's not an obvious -- it's not an easy thing. The customer needs to test, and then we need to get into contracting. But that's kind of the opportunity. It's basically doubling the revenue that we're seeing today.

Great. Let's talk about Scibids. Really exciting, important business for recent acquisitions. How do you see that helping you differentiate your offering from others in the marketplace? And where does -- just to break it down for investors, where does Scibids add value for your customers that you weren't adding it before?

Scibids is a really exciting addition to our differentiated bundle of goods and I'll kind of dig into that for a quick second because you mentioned it. A core part of how we're going to market now, as I talked about earlier, has really changed. We saw it as a protection business now going to performance. And the performance solutions that we have includes Scibids, include ABS, include our prescreened social solutions that we use on YouTube, include things like optimized, made for advertising analysis and attention -- pre-bid attention filtering. These are all differentiated versus our competitors, which allows us to drive a different performance narrative. And where Scibids fits into that is -- I mentioned this idea of pre-bid filtering, right? The ability for us to actually -- before an advertiser even buys an ad or bids an ad to say, don't do that. It's not going to be viewable. Don't do that. We think this is not brand safe. Don't do that. This smells like fraud. Those were very binary discussions, right? We have a tool that goes into DSPs that allows us to say yes or no. When we launched ABS, which as a product as Nicola noted, grew 45% last quarter, that business allowed us to be more dynamic. So for each customer, we could come up with a brand suitability criteria, customize it for each campaign, learn from the measurement and learn over time of what suitability criteria actually performed better. Now enter Scibids. Scibids allows us to take data sets like viewability, like attention, things that were usually seen as binary, yes or no. It's either viewable or not viewable, bid on it or don't bid on it based on it's viewable or not viewable. Scibids allows us to say, maybe -- maybe you should bid on this, even though it doesn't meet viewability criteria, but it delivers a KPI on the other end at a rate that makes sense for you. So think of it as moving from static analysis with our basic tools to dynamic analysis with tools like ABS to totally algorithmic and fluid analysis with Scibids. Scibids takes multi variables. It's a multivariate AI tool. Bangs them all up against a bidding process and says, "Let me look at all these variables," things like attention metrics that DV provides, things like viewability that DV provides, to see which ones are the biggest influencers, and let me make a decision on what I bid and how I bid on that impression based on this KPI on the other end. It's a total game changer, but it's a normal evolution of what we're doing on the pre-bid side. It's totally differentiated in the marketplace. There's no competitor that has anything similar to this and it allows us to employ our data on many more impressions. Because if you think about where we sat before, we get paid when we send a signal back that says, "Yes, buy. It's okay, it's viewable." That limited the number of impressions that we would buy because if we said something is bad, someone wouldn't buy that impression and we don't get paid. Now we're able to say, "Hey, if it drives a KPI on the other end, this is actually going to drive -- then buy it, this will actually drive volume for us. It also allows to apply more data sets in a more dynamic way in that pre-bid fashion. We're super excited about it. We just closed the deal to buy the company in August. We've been bringing sales teams together. But so far this year, and we mentioned this on our call, 40% of our new deals that we've closed for our broader bundle of goods are testing Scibids. That's a huge number of tests. Now the actual conversion rate, let's say, we convert 25% of those folks, right? You're looking at a 10% attach rate for a brand-new product, which is huge. So we're really excited about Scibids, big differentiator, able to apply data across many different places and drive volumes up for us.

So we're talking about social and Scibids and you're closing on that point about the attach rate, makes me think about or want to ask you about your customer life cycle because you're doing a lot on the product side, and it's expanding at a pretty rapid clip, both organically and non. So I guess talk to us about the process of customer acquisition. What it takes to scale a relationship, what the roadblocks are with investors, with the customers that you have to work through to get them on to these new products. And what are they the most excited about? It sounds like Scibids is probably one of them?

Yes. Yes. So I think it's interesting because Nicola mentioned this a little bit before about kind of measurement is being where we started. And for the most part, that customer life cycle still and many times starts with measurement. They want to understand how their ads are performing. They want to be able to block them because the one thing we didn't mention is we measure what happens, but we also have the ability to actually block the delivery of the ad, right? If it's an environment that's not brand safe or it's not viewable. They may have already paid for that ad. So they'll get to make good at some time, but we'll block the ad there. So the life cycle usually starts with measurement. It evolves then to adding activation tools, things like ABS, tools like pre-bid attention filters, and then as those advertisers become more sophisticated and we dig in deeper, we upsell tools like Scibids as well. So there is kind of a land and expand motion that we have with customers. Nicola noted that our top 100 customers now average $3.7 million of spend with us. A lot of that is driven by the fact that they are expanding the number of solutions that they use with us, and we charge them more for every one of those solutions. So if you think about a core business driver for us, is we want to get in with a customer, we want to verify every transaction that they do, but we also want to upsell them because on that same transaction, we can sell them pre-bid filtering, we can sell them post-bid measurement. We can sell them attention metrics and we can optimize using Scibids. We have a metric we call MTF, which is measured transaction fees. That's what we make on every transaction. Last year, we measured -- we monetized 7 trillion transactions. Our fees for that went up 3% for the year and 5% for the quarter in Q4. So we're actually selling more to those customers. And as Nicola also noted, things like our social growth in Q4 or ABS growth -- I'm sorry, ABS growth in Q4, 66% of that came from current clients, 90% of our social growth in Q4 came from current clients. It's upselling them new solutions, upselling them new coverage, and that's what we try to drive, which is this land and expand model. If you look at our average customer life cycle, our average customer tenure right now, our top 50 customers have been with us for over 7 years. So it's incredibly sticky. We drive more revenue per customer over time, and we had an NRR last year of 124%.

You talked about MTF, it's major topic of investors coming out of the quarter because of what we heard from a competitor. Obviously, you're in the duopoly space or at least that's the way many of us view it, maybe you would disagree. And I guess how would you help investors think through the puts and takes of your trajectory and strategy in that duopoly market right now? And then how does pricing factor into your discussion with customers?

Yes. So the interesting thing, this is a market that's had 2 to 3 players for the last 10 years. So the actual dynamics of the market haven't changed much. Pricing has always been one component of many, many components that we engage clients on. The reality is no one makes decisions on price. They make decisions on functionality. They make decisions on engagement and service, coverage, global coverage, all those things are super important. But most important, and as I talked about today earlier in this discussion, they make a decision based on ROI and performance. And if you have a differentiated bundle of goods, like we're coming to market with that drive real ROI, that's the most important factor. So for us, we never lead with price. We haven't. It's -- we've been in the same kind of market dynamics for years with multiple competitors. And as shown by our MTF, which increased last year, we're actually making more per transaction, not less. So we see pricing not as a major dynamic we focus on. We're selling -- it's an apples-to-apples comparison with any of our competitors. We have a differentiated bundle of goods that we're selling, totally separate solutions with a very different go-to-market around.

So I think would it be fair then to summarize that by saying no impact on pricing and nothing that you would expect, just based on your sales motion and the way that you interact with your customers?

The only change that we've seen in the last 12 months is that our differentiated set of goods is becoming more and more of a factor in deals that we closed. I mean you mentioned the duopoly. I think let's cut to the chase here. These were 2 companies that were similar sizes 3 years ago. Last year, we did $100 million more in revenue. This year, we'll do $160 million more in revenue. We're growing twice as fast on every line item as our competitors do. And in areas like social, where we grew over 60% last quarter, they grew around half as fast. We're just growing faster. We're selling more products. We're selling a differentiated bundled good that drives performance. We're winning more deals and we're charging more per impression than our competitors because we're driving more value, period. I've tried to put that to rest, Matt, can you tell?

No, I think it's clear. And I heard keyboards clacking at a higher rate when we were talking there, so I think it worked. CTV, very, very important topic for the past couple of years, has not slacked off at all in terms of the investor focus even at the conference this week. Talk about the latest trends you're seeing in that market, and what do you think will need to change for CTV to be a larger go-to place for, not just ad dollars, but frankly, your customers ad dollars and their usage of your products going forward?

Yes. CTV is the really exciting and intriguing space for anybody in the ad business right now. It's exciting because the CPMs are really high, right? And they're coming mostly from a non-digital source, which is linear television. So it's like dollars that us in the digital world never could get our hands on. Now it's all coming over here. So that makes it really exciting. The challenges are some that we're kind of digging into, which is there's definitely a lack of transparency around buying process, right? Unlike normal linear television, where you bought American Idol on Thursday nights, right? You knew what you were getting, you knew the schedule. There was a finite number of shows. Now you buy a genre. I buy documentaries across Netflix. I buy sport shows across Peacock. You don't usually get the show anymore. And that gives buyers pause and creates a lack of -- some confusion around them. Where we've stepped in so far is around the areas of invalid traffic, which there's still a considerable amount of fraud in the CTV universe. Last week, we announced in partnership with Roku that we discovered a fraud scheme called CycloneBot. CycloneBot was stealing upwards of $7 million to $8 million a month from advertisers based on fake CTV signatures out there. So we've been in the market with a fraud solution that's helped advertisers kind of avoid CTV fraud. We also lean on viewability. I know it's hard to believe, but there are viewability issues in connected television. So let's say, how can a CTV ad not be viewable. It's on a huge screen in my living room. Well, what we found is signals were not being sent by the apps like in 25% of the engagements, that the actual screen was off. So think about what you're watching. When you're watching CTV, you're just watching a computer that happens to deliver it to a screen. That screen doesn't have to be on for that computer to be running. So in 25% of the engagements, we saw that the ads were running when the screen was off from certain applications. So that's -- like that's not an unviewable ad. So we sell something called Fully-On Screen. That being said, those 2 factors are not what tip the equation when advertisers are looking to kind of really engage with us. What really changes the map is brand safety and suitability. And what they want going back to that whole transparency discussion is they want suitability and brand safety on a program level. They want program level data. So that's been a challenge because the major platforms and major apps haven't been interested in providing that data because they've been able to sell based on the genre. We believe that when that opens up, when -- when the platforms are willing to say, yes, we'll provide program-level data to you as a third party as long as it can keep advertisers from cherry picking buying, I think that opens up our ability to do brand safety at a program level. I think it opens up the CTV market at a much greater level. And right now, CTV is only about 5% of our impression. It's pretty small. But I think that there's an opportunity there when that program level data becomes available. And I think it's pretty close to happening.

Interesting. Nicola, I didn't forget about you. Maybe let's talk about capital allocation. You've been active in the M&A space. Obviously, we talked about Scibids, but how are you thinking about capital allocation more broadly? What are the priorities that you would highlight for investors?

Yes. So we've been successful with M&A in ways to accelerate our product road map or accelerate our geographic expansion, right? Those are kind of the 2 major areas that we look at. And the results are that we're growing faster than our peers in the industry. And so our focus remains to kind of find those companies that are going to help us differentiate our product even further, right? Scibids is a perfect example of that, right? There's nothing in the market that's the same. So when we lead with products to our clients, we can say, "Hey, we have Scibids, you can't get it anywhere else." So priority one will remain to look for M&A that doesn't distract us from the organic growth, right? But it helps us kind of accelerate the gap that we're building between ourselves and our competitor. That's priority one. We are investing a lot in the business. I mean, we're a profitable business. Our gross margins are very high. They're higher than our peers as well. And the investments we're making, we're now 3 years into our life as a public company. G&A is pretty normalized by now. Sales and marketing, we made a lot of investments a few years ago because we were a little behind in terms of penetration of the rest of the world market. The next layer of investments on sales and marketing is going to be just opportunistic. Like is there a market where we're not on the ground where by just putting a few people on the ground, and we can really go at the local market advertisers. So the most -- the largest investment is in R&D. And that's data scientists, that's ways for us to do our classification more efficiently, especially around video content, which is now the vast majority of what we do on social, and you can see it in the margin profile of the company, right? So those are the 2 main areas of investments. And I think '24 is -- it's reasonable to expect that that's what we'll do in '24.

Got it. And so when we tie that back to the prospect of operating leverage going forward, how should we think about the potential for margins to expand over time? And how does that tie into those line item level data points you were just going through?

It's an inherently very profitable business model. So we are maintaining -- we're investing into maintaining the margins that we have because really what's driving our performance right now is superior revenue growth, right? We grew 29% in Q4, 27% in full year '23. If that equation continues and we're able to continue to invest into a margin that's still 30%, we'll continue to do that. Of course, there is flexibility around that percentage on the margin side. But right now, that's not what we're focused on.

Yes. And if I can just add on that, too. One thing to be really clear on is that this is a wide-open market. Like this is not -- people talk about a duopoly. This is not a 2-horse race where we're just like fighting each other for market share. 60% of our wins last year were greenfield. That means the advertiser was not using any product of that type that we went after. We look at the global market as being incredibly underpenetrated. Only less than 30% of our measurement revenue came outside the U.S. last year. Yet 60% of digital ad spend is outside the U.S. So we still have like a lot of beachheads to establish. We've got a lot of advertisers to go after. We talked about 500 of the top 1,000. We could be working with one of those companies in one division and one part of the world. So like we're leaning into growth, market growth, product expansion. Like this is still a market that is wildly undertapped. We're scratching the surface on the size of growth we can have. And I think one data point just to refer to, when we look at our current customers today and what they spend, we take 0.5% of their media transaction, 0.5%. And there are other companies in the space that -- ad tech companies that take 20% of the media transaction, right, on average. So we've got room to grow the number of products we're selling. We've got new markets to go into. Like this is a long-term play for us for expansion, right? This is not, "Hey, let's beat each other up." There's 2 companies who beat each other up for like the last customers out there, like we've got tons of growth, new markets, new products and current expansion with current customers to go after.

When I think about those companies in this space that are getting take rates -- I know you're not a take rate model, but that are getting take rates better 20% or thereabouts. A common question from investors is how do they convince advertisers to pay this? How do they demonstrate the value? So when you say, okay, we're at 0.5%. We can grow that. What is that conversation like? How do you demonstrate the value to them to say, like, we're worth more than what we're getting right now.

Yes. So there's kind of the fundamentals of what we do, and we have a UI or a system, it's called Pinnacle. Pinnacle is where advertisers basically set up their campaigns, they implement their safety or suitability filters, they kind of manage all of our tool sets. So it's the software they do it through. In there, there's a ticker -- a running ticker that shows what -- for any campaign, how many impressions that we flagged as fraudulent, how many impressions we flagged were nonviewable. So if you're running a $10 million campaign and we flagged 5% of them as fraud, we just saved you $500,000. There's an immediate ROI calculator within the system that shows you what we're calling out. If you didn't use us, if you didn't use DV solutions, you -- 5% of your impressions you bought would have been what we call non-authentic, right? So there's a direct ROI that we're able to show every advertiser in that system to say, "Hey, we know you're paying us." And as Nicola noted, average size of our customers are getting big. This is no longer a rounding error for some of these companies. They're paying us $5 million, $10 million, $15 million a year. You're paying us, but we're going to show you the direct ROI of not using us, right? And here it is. That's part 1 for our measurement stuff and our blocking capabilities. When it comes to tools like Scibids or even ABS, where we're actually directing bids, we're showing a return. There's a real ROI there because, for example, in Scibids, there's a KPI on the other end. And that KPI can be client acquisitions. It could be brand lift, it could be attention engagement. It could be anything. But we're showing basically using our tool, we're able to drive the ROI or the KPI that you want on the other end. So there's real ways that advertisers can determine whether or not it's worth it to work with us. And that's why when we talk about advertiser tenure and growth and scale with each of them, we don't lose customers, and they pay us more over time. The [ ad ] -- like we mentioned $3.7 million. That was $2.3 million a couple of years ago on average for our top 100 customers. So they're getting bigger, they're spending more and they're buying more products.

Let's talk about AI, hot topic, certainly for the past year, remains a very hot one at the conference this year. I guess I'll leave it a little open-ended. What do you see as the biggest opportunities in your business in AI? And I'm also curious because we get this question a lot, particularly in ad tech, are there any threats that you're solving around?

I'll start with the latter part of the question around threats. As far as threats to the general advertising universe is really -- a lot of it's surround around generative AI. And whether those threats are things like MFA or made for advertising content, which we've seen explode over the last 12 months, which is -- think of these websites you go to where it's like, what do The Brady Kids look like now? Click through the slide show. Like all of that kind of stuff. Well, there are companies that have focused on that for years and maybe they manage 50 sites and would generate 5 new stories a day. Those sites now have blown up to be -- they manage thousands of sites that can actually produce thousands of new articles a day. So think of the world that the advertiser needs to wade through out there has now gone from a little bit of garbage to a massive amount of garbage that gets perpetuated, that getting -- growing every day. And that's the benign stuff. You also have MFA -- you also have generative AI being used to amplify misinformation and disinformation and actually create misinformation. There was a study that came out last year that showed that tweets that were made by AI, that were generated by AI were more believable than human tweets on the same subject. So basically, disinformation created by AI is actually more believable than disinformation created by real humans. So this is the kind of stuff that advertisers have to wade through. So it's created more challenges, but more opportunities. Now fast forward how we're using AI. So -- and then you also have deepfakes and stuff like that. So we -- like there's all kinds of mess out there. We're using AI to actually identify that stuff, right? So our made for advertising tool was built based on AI models to allow us to determine whether or not this stuff is safe or not safe, it's made for advertising. We're using AI in language translation. So think about what we do. We're just a big classification engine. We say right or wrong, good or bad, viewable, nonviewable, A lot of that has to do with the text on the page or the language that we hear in an audio clip. That language, we have linguists, we have semantic scientists that were doing that. We're now moving a lot of that into AI-based translators. So before we put it into our machine learning models, we have AI translate it first. Video analysis. We talk a lot about social video and 70% of social engagement now is video based. So our ability to analyze massive amounts of video is really important to do it really cheap. We're now using -- over the last year, we've built an AI tool that allows us to analyze video at massive scale, very cheaply, so 1/3 of the cost of what we've seen competitive solutions at and very quickly. So if the future of social is short-form video, whether it's TikTok or reels or even in the news feed, and the future of our business is tied to that, our ability to do that very cheaply and very efficiently is critically important. AI is driving that tool set.

Got it. Another big topic in the ad industry right now is cookie deprecation, probably one of the most common questions I'm getting year-to-date. So talk to us about your position with regard to cookies going away, your use of 1P versus 3P data? And then how do you partner across the ecosystem to help your customers avoid some of the negative effects of cookies going away?

Yes. So the good news there is that because we're looking to verify an ad transaction, we've never really been engaged with PII or first-party data. The idea of looking at a user, which is really what the whole cookie concept is about is how do I track a user or user engagement from place to place. So our entire business has always been based on the context of a page, the veracity of a transaction whether or not this was -- this is a bot that's engaged based on patterns that we've seen. So net-net, we've not been challenged from a privacy perspective, and we've not been challenged from a cookie perspective. That being said, we think that there's potential upside for our business based on cookie deprecation because if you can't target an individual or track a user, what becomes more important? Things like context, things like attention. These metrics that we've built based on understanding the page or understanding the ad impression become more valuable. So we think that some of those things that we're doing will become more interesting and more valuable to advertisers once they lose the ability to target users. But we're also very engaged with the Google Privacy Sandbox. We're not beholden to any specific ID. So whether it's UID or LiveRamp's ID, if it needs to be employed in any of our solutions, we're very flexible on that front. So I think net-net, we've always looked at this as neutral to positive for our business. Our customers, I think, are still not ready. I think advertisers still aren't fully aware of what they're going to do when this happens. I think there'll be some knee-jerk reactions. I think you'll see things like retail media networks get stronger. I think you'll see social -- you'll see social get stronger. But it's still -- I mean, look, I think the jury's out. I still think there's a chance this thing is going to get kicked -- the can is getting kicked down the road again. You have the U.K. policy. The U.K. policy initiatives right now, they're actually looking into the Google Privacy Sandbox saying it's anticompetitive. So you may have maybe legal stops to this.

Maybe we'll close with one minute just on international. Talk to us about your progress building out the international footprint, where you're planning to really focus in '24. And then really differences you're seeing between the different markets you're trying to roll out in right now because I think it's not monolithic [ with it ] outside the U.S?

Yes. So we've invested -- this is an investment that we made 24, 12 months ago, right, where we actually went where we were not. And the technology works even without people on the ground. But what we learned is the customers do want people on the ground, right, to address issues that are very local. And we've also learned that by being on the ground, you can actually go after clients that are local clients, right? They're not necessarily part of a U.S. entity. We are basically where we now have a footprint that feels right. So as I said earlier, the additional opportunities will be within specific countries where the sales team will say, "Hey, by putting a few people on the ground, we know we have the opportunity to go after a very large local advertiser." So it will be opportunistic from now on. The mobile is very important in APAC, right? Agencies play a different kind of role outside of the U.S. So we're also learning that there's an opportunity that's a little bit different than the U.S. But overall, the share of our revenue from outside of the U.S. is still much smaller than the percent of dollars that is spent outside of the U.S., right? So the opportunity for us is just to close that gap.

Got it. That's great place to close. Thank you both so much for being here.

Absolutely. Thank you for having us.