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

Okay. Great. So I'm Laura Martin, I'm the senior media and Internet Analyst over here at Needham & Company. And I'm here on stage with the CEO and the CFO of DoubleVerify, Mark Zagorski and Nicola Allais. And Mr. Zagorski, I've spelled your name like 5 time in the last week, and I keep selling like the crystals. And I must say, no, he's not a crystal guy, no he's not jeweler. I wish you were a jeweler.

I used to get [ Sikorsky ], the helicopter people, you know that one falls in there too. I'll take helicopters or jewels. Those are much easier than ad tech these days.

So true. So true. Good point. But anyway, I had to laugh, I'm like, oh, my god, do you think I would have learned this after all these years. Okay. So let's level set. Let's start with -- our first question should be, let's explain what DoubleVerify does and what are your primary drivers of growth? You do what it does, what are the primary growth drivers over the next '25 and '26, let's keep it sort of short.

Yes. So DoubleVerify is an ad tech platform that ensures that ad spend is between a buyer and seller. Those ads are viewable that they're fraud-free, that the ads are delivered in context that is suitable for that advertiser and they're in the right geography. That was the core of our business and where we started. We've now expanded beyond media quality into optimizing the cost of that quality through our Scibids AI tool and recently acquired a company called Rockerbox, which now shows if that quality or that ad impression actually works. So it does attribution. So think of everything from media quality to price compression for that media quality to determining whether or not that quality move the needle. We are a complete independent platform for advertisers.

And the growth driver for '25 and '26 and '27 is going to be -- to replicate what we've done in the open web for the walled gardens. Our solution works across the entire digital ecosystem. We now have access to the walled gardens in a way that we can replicate the success we've had in the open web on the walled gardens.

Okay. And let's just talk to product specifically. So you had -- you started with a post-bid product where you'd measure after the facts, like, almost like a measurement, Nielsen kind of idea. Here's what happened after that. Then you very cleverly -- you introduced a pre-bid product. And most cleverly, you said you couldn't buy the pre-bid product without the post-bid product. So you bundled the 2 together, which took the -- they get paid as a flat fee like $0.08 per 1,000. But by bundling the pre, which I think was $0.14, it turned it into a $0.22 ad unit. So now you have pre tied to post and you were doing that in the open web and now you're doing it with the walled gardens. And specifically, I think the importance of the walled gardens is huge, specifically TikTok and Facebook because you're already doing it with YouTube before, right?

We were. Correct. Yes.

Correct.

What's sort of new is the Facebook. Let's call it meta because it's not just Facebook.com. The meta and the TikTok are like what's new when you say replicate on the walled gardens.

Correct.

Okay. Great. Fantastic. So one of the things, Mark, I wanted to talk about is you guys have seen some of the biggest margin -- not margin, valuation multiple compression. And you're down at $2 billion used to be a $6 billion company. So what -- tell me what you think that the market is missing about DoubleVerify, that it's getting wrong, that it's underappreciating about DoubleVerify today?

Yes. During that period, the fundamental value proposition of what we deliver to our customers hasn't changed, right? So we ensure that their spend is protected, and we are increasingly helping them perform and ensure that it performs. I think what -- the Street is missing is a couple of things. Number one, what Nicola just mentioned, which is the opportunity in social for us is great and broad. We are not just an open web company. Now a good part of our growth in the past has been in our activation line in things like ABS, Authentic Brand Suitability, which is a tool that actually creates custom filters for advertisers when they're buying on the Trade Desk or they're buying on Google DV360 but buying open web inventory. That same model is now being transported into the social universe, which arguably attracts 70% of ad spend right now. So number one, we're not just an open web company. We're increasingly a social media evaluator company. Number 2, the fact that although we started our core value proposition was in protection was insurance, making sure your ads don't show up next to bad stuff, making sure they're viewable or not fraudulent. We're increasingly being seen as a performance business. So can we help the ad spend drive a result? And can we do that cheaper than anybody else? Scibids is helping us compress the cost of quality. Rockerbox is a new solution that we have that will help attribute that quality. So I think they're missing that aspect of the business as well as we kind of evolve over time. And the third is just going back to the fact that this is a highly profitable business that still has strong growth that is underpenetrated in most markets, particularly outside the U.S. So we've got growth levers, not just new products. We've got new markets that we're expanding into. We've got new platforms like we just mentioned TikTok and Meta, which are still in early days. And even if you look at our Q1 growth, it came from not just new customers, but core customers expanding with us on new products, too. So we've got probably more growth drivers and more growth levers than anybody in this space right now.

Okay. What do you say, Nicola?

I think the -- in addition to all of that, if you compare us to other players in the market, we have been gaining share in our strategy around M&A and being able to diversify our product offering. And the fact that we're not just, as Mark said, verification, but we're all the way to performance. Now it just sets us up for a much greater expansion of what we can do for our clients in a way that others cannot in the industry. So I think when people start to see the results of that, they'll understand the potential.

One of the subjects that has surprised me on this stage today is we have been really spending a lot of time on curation, right, which you guys know immediately what that means, but it's a new word for the audience, right? And that's just moving data from the DSP sort of to the SSP. I would shorthand it. How do you -- is that good for you or bad for you? Are you indifferent because you're getting paid by Trade Desk anyway, and now you're just saying it paid by Magnite and your fees aren't any different so you don't care.

Yes. I mean, our thesis has always been, let's get our data wherever and however advertisers buy, right? And as you noted, initially, it was buying and applying our data through Trade Desk or through DV360 or through Microsoft. And that's great if that's how they buy. Now they're increasingly saying, "Hey, let's attach data to the sell side. We don't -- we're indifferent, right? So we announced last quarter that we're working with Microsoft and Criteo and Index Exchange and Google, all on sell-side curation. We will talk about that at some point. But we're deeply embedded on both the buy and sell side. We've had relationships with all of them across many different solutions. So I think curation is just another way of buying, which I think the big question around curation is going to be are advertisers comfortable and willing to change the way they buy, right? I think you have to change an entire workflow in some cases, if you're going to take the DSP out of the equation, right? If I'm just going to use ClearLine and buy directly on Magnite. And I think at some point, if the cost savings are great enough, they will. But for us, again, we're such a small part of that transaction fee. On average, it's like 1%, 2%. And you're looking at other players that are taking 10%, 12%, 15% fees, not transaction fees, right? So DSP fees can be as high as 15-plus percent and SSP fees are higher than...

Your statement that other -- so one of the things that somebody on my stage, and I'm already losing track, I'm sorry, said this morning was that a lot of times, platform fees on the DSP are only 8%, but the average gets up to 20%, and that's all layering all the data. At 1% and 2% each, it's easy to get to 16% or 12% data fees on top of the 8%. So you're saying -- what I heard you say, Mark, is that you price your product the same, whether it's attached to an SSP or a DSP, so you really are indifferent. You're not pricing it higher at the SSP. Your price is the same. So you're indifferent. You don't make more money, you don't make less, wherever somebody wants to buy the data, you'll let them buy it there.

There's one difference and the fact that the DSPs take 20% from us and the SSPs don't.

That is a difference.

It's a big difference for us. It doesn't matter to the buyer, but it matters to us.

Right. So to me, that isn't priced the same, then you're not indifferent. You'd rather have it attached to the SSP.

At this stage, yes. But the scale of the SSPs on curation is very, very small right now.

It's new. It started in January Yes, like it's a brand new thing.

Yes.

Okay. But -- so would it make you -- incent you to raise your price on the DSP data so you can cover your 20% fee that you're not having to pay on the SSP side.

Yes, at this point...

Until If you cut your price marks, seriously.

We're not cutting any prices. We -- look, our driver here is to keep friction low and make buying really easy. So driving prices up or down is not part of our goal right here. That's another thing that is important to understand about DV is that we drive revenue is through volume and through product stacking. So it's not about the individual cost of a product. We want people to use attention with us in context with us and ABS with us, and do measurement and pre-bid. So it's every -- that impression we want to stack as many products as possible on top of it from a data perspective. The individual cost of each one of those products is less important than us selling through more products over time.

Okay. So one of the things that you and I talk about a lot, Mark, besides price, which I know I just give you a really hard time on, is this issue of, are you inventing the products or others? So you've just said that a big pivot of yours is moving towards performance, but you bought right, and outcomes. That was Rockerbox. And then with Scibids being more efficient, like you're pivoting from where DV started to where it's going, you've had to buy both of those products. Tell me why -- tell me buy versus build. Is it just that you saw something coming. They had already been working on it. They had decided 3 years earlier, so they've been building it and it was just going to take you too long to build, so it's cheaper to buy them.

Look, we started talking about performance well before Rockerbox or Scibids was even on our radar stream. We've always looked at ADS and some of our prebid tools as performance tools. When you pull garbage out of the system, what's left performs better, right? So we've leaned heavily into that. We started playing around with looking at bidding algos, and we found, hey, we can either build out a data science team of ourselves, but we can buy these great team that's in France right now and integrate it really quickly. Same thing with Rockerbox, which is if we want to look at closing the loop, we can start building that competency or we can do -- find a bunch of great technology folks who've been doing this for the last 6 to 8 years and put it into our strategy. So I think we've had this thesis around acquisitions, which is either they expand geographic footprint, so we are able to move into a market quickly, eliminate the competitor there and like kind of put our platform in. It accelerates our road map something that we're going to build? Or is it complementary tool set that we don't have a ton of like innate knowledge on. I think Rockerbox is one of those things that is totally complementary. Understanding media mix modeling, multi-touch attribution. We don't have that DNA in our business. However, it's an extension of what we're doing today, and it made total sense to acquire that business. The common link between all of these are that they're generally underscaled because they haven't invested sales. They don't have access to big customers, but they've got really good tech teams. So if you look at the last couple of acquisitions, like we've had very low turnover because we've kept those teams together. We've actually grown the resources around them, and they've given us a core base. It's kind of just -- they've almost been acqui-hires because the revenue from them has not been big. It's like [indiscernible].

Anything for you guys, right?

Right, they're all done with cash. We don't have debt. And they also already had right now earn outs. And they also already had a product in the market and the integrations with the DSPs or the other partners, which is important, right, to allow us to go faster.

Scibids is a great example. I mean, we've now sold them to 200 customers already, upsold them since we acquired them almost 2 years ago that they would never have had access to, right? But now they have some of the biggest brands out there, everyone from Diageo to huge global brands that we talk to every day, right, that they would not be able to get into those customers. Same thing with Rockerbox, a very small company, had 2 salespeople. Now they have access to 150 salespeople, right? And that's the value of bringing these assets to New York.

Okay. What I was going to say is, I cover a lot of companies that aren't based here. And I think it's so funny because outside America, people are like, if I build great tech, they will come, like that's not a thing. Like if you have great sales, they may come and maybe the tech can be secondary. But I think it's a great -- like it's a great arbitrage. You guys have great sales and great relationships. That actually matters in media a lot. And great tech you can buy. I mean it's better to buy them and put their awesome tech to someone will actually have impact. It is really a win-win for everybody.

Look, we've never been a CMO who goes, "Wow, I wish we'd have 3 more companies calling on us, right? They don't want that. They want one company to call them, that can provide them multiple solutions right. They want someone to come in with a complete platform to say, great. So you can tell me what's good. You can get me what's good cheaper, and you can tell me if what's good worked awesome. I don't need to talk to anybody else today.

Now when you think about the Scibids and Rockerbox products, and you just said stacking is a key, like important thing to you. Do you stack those on top are they stacked products from your point of view? Or are they somehow open new TAM, which isn't really a stack?

There's a little bit of both. So the -- so from a go-to-market perspective, we're starting to now bundle them more, right? So I'll give a good example. We were pitching an RFP last year.

Request for proposal.

All right. Request for proposal. Every acronym I'll try to define.

By the way, we don't have that in ad tech other than you. So it's not like a thing they should necessarily know.

Okay. Well, it's more software. It's more software like, right?

Yes. That's fair.

So we're pitching that for a large global CPG customer and they've been a verification customer for us for years, quite large, but they're like, hey, we want to put you out there and see how you go against everybody else. And the old way we did it was, we do these head-to-heads, right, put our platform against a competitor's platform, see who filters out more fraud, do all that stuff. Well, that was part of it, and we knew we were going to win that. But then we brought Scibids in and said, guess what, if you bundle these 2 together, we will guarantee that you're going to save enough money by using Scibids that you'll actually pay for our measurement solution. Because every $1 that you push through Scibids, we save you $4 in media costs, right? So now we're bringing 2 solutions together to an advertiser and saying, no one else has these 2 together. So if you're buying verification from somebody else, you're going to have to buy optimization and algo optimization from another party. We'll bring them together and show you that we'll save you enough money to pay for this other solution. So...

Feel free in a sense.

Exactly, in a sense. They are paying for it. I mean they're paying for it. They're paying for both. But we're saving them enough money. And I think that's where the value prop comes from like stacking solutions, right? We're selling you one thing, now we're selling you another thing. And now we can go in and say, we're going to sell you Rockerbox, too. And that's now one core value prop that's integrated, sold together. And if you look at the average spend for our top customers, so customer spending $200,000 with us grow 14% in Q1 in the last quarter. So our customers are spending more with us, right, on average. And it's not just because they're buying more media, it's because they're buying more products for us.

Okay. So one of the reasons I like Scibids so much was because they got paid as a percent of platform revenue, not as a fixed fee, which as you know is a fee structure. Anyway, moving on to my question is, have you kept that business model? And given that you just said just now that 200 brands that they never would have had access to have adopted where all of those brands paying for Scibids based on a percent of ad revenue that they -- ad spending that goes through Scibids.

Yes and yes. So...

The answer I wanted to hear. I'm happy. It's totally great. Okay. So in pricing power, one of the things I always ask you on any opportunity, Mark, is your prices remain the same when you move from, I'm going to call it, DV+ but that's the Magnite product. When you move from the open Internet product to connected television, even though the cost per 1,000 that you are adding value to went up 10x, right? $20 CPMs when you started, no longer because we have more supply now. But at the beginning, it was $20 CPMs versus $2 CPM, you charge that same, let's call it, 20% rate per 1,000 and I was -- I objected. And you've said for a long time, we know we're adding more value to CTV, we need to raise price. Tell me where you are. I get that you agree with me, when are we actually going to see price raise on the CTV piece?

So we've been talking about price increases on CTV for a while. I think we are right on the verge of actually providing a much different value prop than we have to date. We still -- look, CTV measurement still has a lot to do with finding nonviewable impressions, so we found almost 8% of apps that we're measuring were not sending a signal that the TV screen was off, which is crazy. There's still issues with CTV fraud. Those things are still incredibly valuable. But what advertisers really want and what they pay for as we started the discussion around kind of brand suitability, is can you give me a media quality score? Can you actually tell me something? And we are on the verge of actually kind of introducing that solution set, which has a quality aspect to it the same way we do on the open web, the same way we do with social to do it in a much more granular way for CTV. When we get that product to market, we believe that's the opportunity for us to actually get a price commensurate with the value that we're delivering on CTV.

But it sounds like you would -- the old product would stay at the same price and you get the price increase by, I'm going to say, stacking this new product onto it.

Yes.

I don't object. Because that's what you did with pre-bid when I was yelling at you at that time when you only had 1 product and I'm like, when do we get a price increase? You're like, we're just going to introduce this new price and bundle it with the old, which tripled the price point. So in a way, it was better than what I was asking. So I totally -- this is totally a fine way to raise. Look, you're basically saying everybody has to be a customer first at this $0.08 per 1,000, but then we're going to add stuff, that gets you the call. That takes the meeting, right, because you already have a relationship with them. But then we're going to upsell something that we know. And a score would be -- so would a quality score for Coke then be a different quality score for like diapers? Is the quality score related to what the ad pod is?

It could be because think of brand suitability for every brand is different, right? So it's suitability of brand quality score based on what their preferences are and based on the granularity of data we can get from the content providers.

Okay. Okay. So like a brand [indiscernible] wouldn't be like a Hulu has all these super high-quality units that still might not be the brand safe, that might not be the same brand suitability score for diapers versus GEICO versus Coca-Cola. It would be unique to the advertisers.

Right. And think of other scale. Absolutely, because it scales in ABS today. Every customer has different ABS profile. So think of the other aspects, too, of CTV, things like screen resolution right? Things like what screen was it actually seen on? How big was that screen? Was it fully delivered, but also the things that a lot of people don't talk about, which is was this really on the network? Or was it an extension? So a lot of CTV inventories sold in bundles or in PMPs, right, private marketplace packages. Those marketplace packages can include the publisher entity that you know and love but they are made Paramount Plus, but it also could include impressions on other places that are part of the Paramount+ network. There's the same way retail media networks use extension. CTV does extension too. Is this really helpful? this provides much greater transparency, yes.

And you guys have access to this data?

We are slow building relationships to give us that access, yes.

That would be really -- that is really valuable, especially to the extent they use extensions, and they're putting it on a FAST channel, yes. Very interesting. Okay. And the timing of this product rollout would be 2025?

It will be at some point in '25.

Okay. All right. So this would be -- I think this would be very exciting, mostly because of the pricing implications, but just to have the notion of a product that does quality scoring for CTV is really interesting because basically, that becomes best-in-class and everybody else has to keep up with that. And it would make better behavior of the CTV ecosystem. It seems to me.

Absolutely.

Okay. Let's talk about acquisitions. So to both of you, question. You guys have acquired Scibids, and that was just to tell you guys -- that was in the third quarter of '23, which for $125 million. Rockerbox for $85 million. So again, tiny little acquisitions, but hugely beneficial from a bundling, pricing, let me call it product offering point of view in terms of when your salesman goes in, you want them to have -- it's almost like a ride at Disney Land. You want to be able to talk about something new. So they took the meeting for a reason and not just we want a price increase on what we showed you last year. So I totally get that 1 a year is -- it makes a lot of sense. Going forward, it sounds like you're building the connected television product that allows you to raise price. What other kinds of things can you see building that would make this stacking idea more valuable?

Okay. So I'll just -- so I think from -- just to be clear, M&A is not a strategy, right? We have a strategy. And that strategy is to evolve from protection to performance and ensure that we close that whole loop. And if M&A fits into that, great. Our main driver right now is to take these assets, so these raw materials and look at ways of bringing them together to create unique value. So can we bring media quality plus Scibids together to create a unique value prop for advertisers, which gives you quality at a lower price, right? Can we bring Rockerbox together with Scibids? That allows you to take what's working and optimize against a side an attribution performance metric that sits within Rockerbox. We've done that. We've done that with clients before we even started buying these companies. So I think that's the one thing to note. All of the acquisitions that we've made over the last 2 years are companies that we've worked with. We've worked with before we bought them. We built solutions together because we saw that there was a path towards where we wanted to get to. We didn't say, "Oh, well, let's buy this company to get into this space. Let's buy this company in this space." We're like we're going into this space. Let's start working with them. And if it makes sense for us to buy them, let's buy them.

No, I would say the -- you were -- the question was what are we going to be innovating on? I think creating the full picture of verification, performance and outcomes is really what we're focused on, right? Those are -- there are 2 prongs out of the 3 there that we're the only ones that have in the space for what we do. And we did acquire the technology and the integration. So that's what we're really focused on in terms of how much more differentiated our product is.

Well, I'm a big believer in bundling. Bundling says it lowers churn by 50%. So the bigger your bundle gets, then less likely you are -- at least from a consumer point of view, I assume it also applies to enterprise. I have only done the research on consumer.

And I think the bundle here to isn't just a sales bundle. It's a tech bundle. It is actually a seamless solution. So 1 UI, 1 data set that can move seamlessly from verification to optimization to actually attribution, like those things are important. That takes a lot of work, right? And we're still in the early stages, yes.

If something doesn't work, there's somebody to call and say, fix this. And you just fix it.

Yes, yes.

And it scales, right? That's another...

That's the key.

I would assume this kind of thing is a faster growth driver than going offshore. Wouldn't it be? I mean don't you have enough growth in America that you don't have to go offshore, where they might want different things.

I mean, look, some of these products -- so specifically about Rockerbox. Rockerbox is a product -- is an acquisition that actually expands the dollars that we can go after, brand -- brand marketing, brand advertising is really what verification is about. Rockerbox is performance dollars and performance marketing, which for us, expands our funnel down towards smaller advertisers, more DR focused advertisers. And small advertisers outside of the U.S. is a more prominent kind of activity. So it does actually help us kind of expand our reach outside of the U.S. So I don't think it's -- I don't know if it narrows us to the U.S. I think it's global. Yes.

I'm just wrong, completely. That's fine. I have learned to...

We have the other things too -- when we think of expansion outside of the U.S. too, it's not like us going into market. For the most part, we sell local brands and local markets. But we're working with folks like Microsoft and Unilever and Colgate and Reckitt. They're global companies, right? And they have divisions all around the world that need support and that their ads spend is different in market A to market B. Southeast Asia is very heavily social focused, right? They spend a lot on mobile and social. So having solutions that serve Unilever in Southeast Asia is just as important as having products that service their brand business in North America as well.

Well, I will tell you, Wall Street has decided sometimes we're wrong, that the future of television is performance that because Amazon has sent this new benchmark, but to keep up in connected television you really do have to drive to purchase. The question is, does that make data irrelevant? Like you're selling a lot of data and verification. But I think one of the things you and I have talked about in the past, Mark, is that let's say, half the impressions are wasted and they're fraudulent. So long as somebody is spending $1 getting $4 of return at the purchase level doesn't matter. Like they're not going to pay to hear that half their impressions are wasted because what they're going to solve for is I spend $1, I get $4 of purchase. So can you speak to that? Would you disagree with me?

Well, first off wouldn't you rather not buy those 4 impressions that were wasted? Like why waste the money that makes your return even better to begin with, right?

It's fraudulent what you're seeing.

So that's the first thing. But the second thing is what you noted is, yes, but who determines attribution? The platform that you bought from. Amazon is going to be the only people that give you attribution? Absolutely not. That's why we have Rockerbox now to say, you need to be outside of that and say, okay. Well, the person saw a CTV ad. They also did a search. They were -- saw out-of-home ad like where -- who gets the credit. And I think who got the credit has been gamed by a lot of the big platforms for a very long time, right? So I think, again, our position and our value prop has always been, we're outside of the transaction. We're independent. We're agnostic. And in the same way, we're not going to make a call based on the quality of an ad based on what platform it's on. We're not going to make a call on giving attribution to somebody based on the platform that we're going to give attribution based on an agnostic independent way. So I think what you just said, sure, if everything goes performance, okay, who determines what that performance is. It can't just be the platform. There has to be someone independent outside of it to say that.

You're raising a really interesting question that I'm thinking -- or really interesting learning that's happening on my stage in real time. My very first speaker, which was Criteo in the morning, said that the sides are softening. The DSPs are taking on SSP capabilities. SSPs are taking on DSP capabilities and the open Internet is getting to act more like a walled garden within themselves because these walled gardens, like with the rise of Amazon and TikTok are so powerful that these walled gardens are becoming more powerful as a group in a sense. And probably, they're mostly taking share from each other, I would say. But anyway, my question is staying with this thing, this point you've just made, Mark, is are we seeing that really going forward, your solutions speaks to walled gardens, that's where you said you're going and the open Internet, both is the whole ecosystem is the demand and supply requiring more of a holistic look so that you take away the disadvantages of the walled garden and marry them with the advantages of the open Internet, which is comparability. Your solution compares walled garden self-grading homework to the open Internet.

100%, right? I mean that's -- look, again, when we verify an impression or a transaction, we're indifferent to whether that transaction was on the social or on the open web, and it creates an even playing field for quality, but it also now with Rockerbox, it creates an even playing field for attribution, right? So if you can work with the same metrics wherever you buy, have the same kind of independent view wherever you buy, then I think that makes open web on par with social across the board with regard to how advertisers evaluate the efficacy, the quality and the return on spend.

Tell me where that goes. Is that good for the open Internet?

I'm sure. It's good.

But that would imply that the open Internet returns are at least as good on a well -- like-for-like basis as walled gardens. Is that true?

I mean, look...

As you see both.

I mean -- I'm not going to -- it depends. There are so many factors. It depends on the advertiser strategy, what they're willing to pay, what their actual KPIs are. I mean the -- if advertisers weren't finding value on the open web right now, they wouldn't be spending there, and they obviously are, right? So they're finding value. They find value in walled gardens. I think a robust strategy includes all of those things. And there won't be a single advertiser, I don't believe that will sit on the stage and ever say, we spent all of our money in just one place. They don't. They spend their money where people are. And if people are on the open web, then they're going to spend money there. If people are on TikTok, they're going to spend money there. They want to engage customers. They want to sell product and they want to know that their ad spend has been efficient. And we can help them with that third thing, which is understanding how their ad spend across all those things compares.

Well, the point I would build on is incrementality, even if your walled garden was the most efficient way Amazon, just for CPG. You still have to reach everybody. And not everybody is on Amazon. Amazon tells us, they only have 120 households here, and we know there's 240. So you need to reach the other 100 if you're selling to certain products.

Yes. Linear TV still is a multibillion dollar business, right? Why?

I think Wall Street's gotten wrong in the last decade. How many linear TV subs, they're still watching?

People are still linear TV, people still are watching connected TV. Like again, there's the -- if you want to reach a mass audience, if you are a P&G or Unilever or any CPG company, yes, you want to lean into building brand through social, but you also have to reach everybody and you do that through multiple different media.

Okay. Questions for DoubleVerify. Anybody, any questions? Okay Great. We have about 5 more minutes, and then it's time to go. Okay. So let's talk about moats. So competitive advantage, pricing power, I really like the strategy are articulating not only of the pivot to performance, I like that a lot, but the bundling strategy of adding more things and adding products that drive pricing power, I'm like a huge -- it's sort of an editor of CEOs and CEO strategy, like I like all of that, like on the 10 out of 10 level in terms of adding value. Bigger, more complicated bundles, lower churn, more pricing power through more value-added. I totally hear that, that works. And by the way, if it doesn't work, they'll push back and you'll have to cut the price. But you guys are good. So far, you've been really good at pricing products at a premium and then getting them adopted. So, so far, I think you have a really good track record of introducing bundles.

Can we just stop right there, Laura. I just want to leave it right there on that will be done.

Okay. And I hear you from this thing that I've learned on this stage, which is maybe where advertising is going, is integrated walled gardens to open Internet, you guys have a pole position in that because very few people can say that. But my question is competitive advantage moats, which goes to pricing power, of course. But tell me how you think -- do you only think of your competitors as IAS? Or do you -- I mean, is one of the benefits that you have pricing there's only 2 of you or do you think of single point solutions as competitors? Like who's your competitive set? And -- or I would have come on to the stage saying you're just getting displaced because there's a drive to purchase, we don't need the data layer. But you probably disagree with that?

I mean I think we have a very robust competitive set, right? Particularly due to the fact that there are lots of point solutions that solve small problems, right? Ultimately, I think when we talk out moat though, scale matters right? And even with AI, the data that you train your AI on matters and that data that you train your AI on is advertiser data. And we've got 1,000 of the biggest brands on the planet that work with us, including every major tech company. TikTok's ad spend, Meta's ad spend, Google's ad spend.

Who knows. There's no one except them and you.

Right. So we -- so like we had an incredible data set that's unique that we can build knowledge on. We've got broad scale, and we've got agnostic ubiquity across platforms, not just Open Web, but walled gardens, CTV, et cetera. So I think that puts us in a unique position. It does not mean there aren't competitors out there, including some of the platforms themselves.

[indiscernible] They only do themselves.

They only do themselves, and you can't take a metric from platform to platform, right? So we are in a unique position, and I think that gives us strength moving forward.

Yes. I mean that's what they don't have, right? Each platform doesn't have what we have, which is scope across every platform.

I wouldn't find them as competitors actually. And so I'm going to ask my last hard question, Mark, which is when you think about Gen AI and data, how are you guys using Gen AI to lower costs? And the part that's hard is you are adding employees second only to the Trade Desk because I'm going to ask this of Jeff too, this is coming, you guys add employees like no one else in the business except for Trade Desk, which means what that says to me is you're not a tech platform. You're some kind of a service platform because scale should mean that your revenue per employee is going up, and it's not. When you add Google and Facebook, they are really tech platform and they scale. Why are we adding so many employees, please?

Look, I think you have to look at where we are in the evolution of the business and when we really started growing as a company, which is only in the last few years. We're still building infrastructure around the world from account management considerations, marketing considerations, like all of those things like feet on the ground, boots on the ground still matter. Yes, we're a tech company, but I'm going to tell you that a major CPG brand wants a person in Singapore that they can talk to for their division there. That's just the reality of it. You need to have that. However, we have incredible leverage in our model. You're going to see that we've mentioned in the last 2 calls, you're going to see head count growth start to level off as we see efficiencies of scale. Right now, 25% plus of our customer service requests are now being matched by Gen AI, right? So that's helping. We -- I think 20% to 30% of our code now is being created through AI tools. We've been able to manage and have an incredibly high gross margin of over 80% in the last 2 years because of our unified content intelligence tool, which does predictive modeling to understand context of video and short-form video, which has been blown up across shorts, TikTok and Reels. So like if you look at the volume of transactions we're measuring versus kind of how fast we're growing, we're getting incredible leverage and it's only going to become better.

Yes. And I think these are investments against innovation, right, that provides us more scale and actually a greater share of the market, right? Like we're growing faster. We're now a larger company than our closest competitor because we continue to innovate.

Okay. Well, that's okay. I spent an hour with Roku and the CFO has convinced me that revenue for employees is not the rest, right productivity metric. So we will be changing our productivity metric. I don't know how you guys come out on that, but I'm keeping a close eye on the FTEs as you know. I'll call it there. Thank you very much.

Thank you, Laura.