DoubleVerify Holdings, Inc. ($DV)

All right. Hi, everyone. I am Mark Schilsky, JPMorgan's Tech sector specialist. Thanks for joining today at our TMC Conference. I'm joined today by 2 members of DoubleVerify's managed team; Mark Zagorski, CEO; and Nicola Allais, CFO. Gentlemen, thanks for joining.

So let's get to it. Mark, you often have described DoubleVerify as the trust layer of the digital ecosystem. The industry has been moving more towards walled gardens. -- away from the Open Web, how has DoubleVerify's role in that world changed over the last couple of years. How do you see it going from here?

Yes. So the -- I think there's kind of a tale of 2 different challenges as we has evolved from kind of an open web trust layer to one which now is very well embedded into the walled gardens. In the open web, it was kind of the Wild West, but you could see what the challenges were, right? It was the open web. We acted as a layer between buyers and sellers of digital media on mobile, display, video, et cetera. As that's now evolved into playing that same role with walled gardens, it's a very different story. It's a walled garden, which in its definition has opacity. You can't see into it, right? And I think that is where our value really shines, which is when an advertiser is buying digital media on TikTok or on Meta or on YouTube and there's an increasing amount of kind of obtiscation of what's going on there? Where is my ad going to show up? What kind of context will be involved, will be there? The role that we played has never kind of been more important. And as these black box optimization tools that you'll hear about you hear Google has P-MAX and that has advantage plus. As those solutions become more and more prevalent, the need for third-party objective verification becomes even greater. As a matter of fact, on meta, we see that customers who are employing their black box solutions, their Advantage Plus solutions have a 2x higher rate of TV measurement solutions than those that don't. And it just shows you that advertisers still want independent objective verification and transparency particularly in walled gardens, where they don't have as much control of the buys. They are using tools that are embedded into the systems, and they're looking for someone who can give them a stamp of approval saying, "Yes, this is what you thought you bought is what you bought.

All right. Perfect. So there are 3 legs of growth that you've talked about, social, CTV, AI, so social, obviously, one of the walled gardens, meta being the largest of them, social activation, potentially one of the next leg. Like what is driving that business? It's relatively small today but growing really rapidly. So what's driving it? How big can that get? Walk us through that.

Yes. So we've said that social is going to be the next real growth catalyst for the business. Our social measurement business with Meta alone is a $40 million plus business. Social measurement across the board is well over $100 million of revenue for us and growing. We saw our social measurement business grow 23% in Q1. So nice double-digit growth. Our social activation business grew over 90%. And the catalyst behind those -- both of those is really kind of the growth of meta. Meta, we launched an activation or a prebid filtering tool on Meta early last year. That's gone through a couple of iterations and evolutions where it's now a product that we've got a relatively strong attach rate. Over 30 of our top 100 customers are now using it and starting to scale against that. We've said in the past, we believe that, that prebid solution alone on Meta is a $40 million to $60 million business. We're already at a $12 million run rate. So we're getting where we need to go with that business. And the nice part about it is social measurement and social activation go together. You can't do one without the other measurement feeds activation. So when we look at, for example, on Meta, we're about 60% penetrated in our top line customers with meta measurement. So every one of those customers were going to upsell prebid -- we've sold about 30 of our top customers. So there is growth tab there. But we also know that we see more customers using measurement when you have an activation tool. So just for a quick definition, measurement for us is actually measuring and blocking a transaction after someone's purchased. Activation or filtering is actually keeping them from buying an impression beforehand. So the 2 work together, measurement feeds and creates intelligence for activation. In meta, we didn't have activation up until last year. So now that we have both of those, we have got this kind of virtuous cycle that we've seen in the open web play out when we had our Open Web solutions, we had -- we launched first with Measurement eventually. Our activation tools became larger, almost twice as large as our measurement tools because, a, they're priced at a premium; and b, no-brainer that advertisers rather avoid a challenge then have to block it or get a make good after they've already bought it. So we see social continuing to grow meta being a major catalyst that -- we also now have launched a revised activation tool on TikTok, our YouTube solution, which we've now kind of bundled with an optimization solution called Authentic Advantage is growing really well. So across the board, we see social as a major catalyst for our growth. We see that being driven by the bulk by bulk in meta by actual percentage in TikTok, which is growing off a smaller base. But ultimately, that meta business, we think, is a $40 million to $60 million prebid business on its own. You add the measurement business, which is $40 million already, and you're looking at a potential $100 million plus just on meta.

Perfect. And then you mentioned TikTok. So, a, how big could potentially get for you? And then b, like you also have similar products on Pinterest and Snap? I know they're meaningfully smaller, but just curious if those are available a swell.

Yes. So we do have -- so tick talk, I think TikTok is still coming off a small base. So it's probably ultimately 20% to 30% the size of a meta potential over time. With regard to Snap, Pinterest, TikTok -- Snap, Pinterest in some of the other platforms. We have measurement on all of those. We have various levels of prebid, but they're significantly smaller. I mean they're really marginal when it comes to kind of the overall -- it's really YouTube and Meta and YouTube, we put into our social bucket. YouTube and Meta 12, then TikTok, relatively distant third and then all the rest in the bundling on.

And then you said 60% of your top 10 customers are on -- you've gotten with the social products today. What is it going to take to get to 100%? And then Meta famously as 10 million-plus subscribers. I know they go way down their SMBs. But like how far into that SMB TAM, can you potentially penetrate over time?

Yes. So the 60% number was just for Meta. So just the meta penetration across our top 100 customers. varying levels of penetration of different social tools across the top 100, but lots of room there. When we think about our customer base, as you noted, someone like that has 10 million customers. We work with the 1,000 largest brands. So we really are focused on kind of enterprise-level engagements. So I think customers like Unilever and Colgate folks who spend really hundreds of millions of dollars across digital media. That being said, we do start looking at mid-market advertising agencies that we're engaged with, that provides opportunity to reach more customers. We are plugged into what we call demand-side platforms. So folks like the Trade Desk Google and others, in which our solutions can be applied basically in the self-serve motion against buys. And so we see a percentage of our customers coming from those self-serve motions, which are longer tail. But ultimately, right now, we're focused on enterprise customers. We're still only working with about half of the largest advertisers in the world. So we're not even talking to the other half. So big opportunity there. Plus we're still underpenetrated with the customers we have. So we think there's a lot of legs with current customers, a lot of legs with new customers in the enterprise side before we really have to start pushing into smaller markets.

Perfect. Let's switch gears a little bit. Let's talk CTV. So CTV, I believe it grew 20% in the most recent quarter. How much is that existing customers where their advertising dollars are growing versus new logos? Walk us through exactly what's driving the growth there? And then what are some other major initiatives you have in that space?

Yes. On the CTV side, we grew, as you know, 28% on the measurement side of our business. The opportunity there is I think we have a relatively low attach rate with current customers. And that's really based on a product or product innovation. I think we've had a product which has been good on the CTV side, but not great, the level of granularity that we've been able to verify impressions against is not equal in CTD as we have in social in the open web. That's changing. So earlier in Q1, we launched a solution called Verified streaming TV. What this allows advertisers to do is actually identify whether or not impression ran against high-quality full episode player, for example, a Hulu or Paramount player versus just ran and kind of an open web or a mobile environment. We launched a solution in ABS our ABS tool through DSPs that's called not Airless, automated do not airless that allow advertisers to exclude certain programs or genres out of their buys. What that means and why I'm talking about that is that it's allowed us to drive a higher attach rate. Our attach rate was in the single digits on the prebid side for CTV solutions. Just in the last quarter, we've seen that double -- so we've been able to double. So to give you kind of a sense of what that looks like on one of our DSP partners, our attach rate for what we got a product with all the ABS is around 60% for open web impressions. For CTV, it was only around 5%. Because of the lack of granularity. With this new solution, we've now been able to double that to 10%, which is driving a lot of prebid application drives postbid, and we see that reflected in measurement growth. So we think CTV is, again, another untapped opportunity for us where we're just scratching the surface. But as we noted kind of in our earnings call, everything we're doing right now is very much product-led innovation, right? It's not just selling more of what we have. It's actually driving new solutions into market, breaking new ground, getting folks to take on new social applications, new CTV applications and eventually new AI applications as well.

And then amongst all of the CTV applications out there and AVOD providers. So the biggest ones are YouTube, Roku, Netflix, Disney, Tubi Pluto, et cetera. Are you plugged in all those platforms? Can you see all of those? And then where are you actually seeing the most growth come from?

Yes. So we do -- we're plugged into the TV solutions. The challenge there is we get various varying levels of data. We get the basic level of brand safety data at the app level. We get impression level viewability data, we get impression level fraud data, et cetera. But our entire solution is really based in where the value prop comes to bear is on brand suitability and getting that on a program level is our next big unlock. And we've been talking about that. Lately, we just closed the deal with spectrum. We had a deal previously with NBCUniversal to get impression level brand safety and suitability data. That's a big unlock for us. And so we're going to start to see growth, I think, in CTV as we bring more and more customers into the full -- more and more platforms and default to share that data with us. on a platform-by-platform basis. I think CTV as a category. They're all growing well. I think the big catalyst occurred last year when we saw prime ads. I mean that just was a huge unlock because it flooded the market with ad impressions, which did one thing which is relatively positive for us. So we charge on a per impression basis, other platforms sometimes charge in a percentage of media. And we've never been able to take advantage of that kind of high CPM that CTV has. But with a bunch of impressions come into the market, it lowers the CPM increases the number of impressions sold. So that is actually beneficial to us. So we've seen volumes go up pretty much across the board on all CTV platforms, both because of the changes in CPMs but also because of the new products we've launched that have helped grow our attach rates in that market.

And then you mentioned you signed a deal with Spectrum on brand safety, like -- what's kind of your forecast for signing new deals there? And like what are you working on?

Yes. We're -- so Stay tuned. We've got other things in the work right now we're pretty excited about announcing over the next quarter or so. The great news is, just like we saw in social -- once you kind of break the log jam with one platform working with you, so TikTok was really our log jam breaker in social. Once they leaned in and said, we really -- we know as we grow, we need to have verification. We want third parties in here. Then you saw others follow, right? I think we're right on the edge of seeing that log jam with CTV, which is, hey, if I am competing against others in selling my media in my streaming media, and I'm not providing the same level of granularity to buyers so that they can understand what's happening in suitability, I'm at a disadvantage, right? So if my sales guy goes in and there's an advertiser buyer on the other side and they're saying, "Hey, we can use Double Verify to buy across spectrum or by NBC, but I can't use it on Paramount or Netflix or some other platform. That's a challenge for them. And I think it's just about getting enough Domino's to fall where they say, "Hey, we all need to be part of this. And I think we're close.

All right. Sounds good. Let's -- there's a variety of topics to talk about in AI. AI ads are extremely nascent -- how do you think the AI ad world is potentially going to evolve? And then what your place in that world going to potentially look like?

Yes. We look at, I think, kind of 3 -- AI is all opportunity for us, just to put very, very, very transparently because I think -- there's a narrative -- the narrative is AI eats everything and destroys every business, every software business, every media business, it's all over. Our role as a trust layer has evolved from Open Web to video, to social to streaming and will eventually play the same role in AI as we play in those platforms as well, which is wherever there's opacity with others someone selling digital media, an advertiser wants that are independent third party to be able to determine what's going on there. So we think AI is a big opportunity. There's 3 things. I think of 3 areas where we've leaned into. The first is the general environment in which AI has created chaos, right? You've got slopped -- you've got AI slop everywhere, whether it's on social or whether it's on the open web, there's AI slap that advertisers need to navigate. We launched a product called SlopStopper, which keeps advertisers away from AI slap allows them to their ads from running against it. We launched on the Open Web. We've recently launched SlopStopper on YouTube, which allows them to avoid low-quality AI-generated content. That's the opportunity for us, right? There's also a ton of AI-generated fraud, we call AI cyber fraud. As a matter of fact, in Q1, we just released a report. We saw 140% increase year-over-year in AI-driven fraud variance in the first part of this year. So it's creating a that means people are leaning on us more. The second on the AI front is really the growth of a genetic buying. Advertising was bought and sold between 2 guys who passed a piece of paper, right? Then it became a fax machine, then it became an e-mail, then it became programmatic buying through bidding. The future is a genetic buying, which is my agent is going to negotiate with your agent, and you're going to sit as a seller and my agent is a buyer agent and they're going to talk and they're going to make a trade happen there. We will play the same role in that world. We've already started building for the ADCP protocol and the other MCPs that are out there that someone needs to determine whether that agent is real or not. So we've seen a huge amount of general fraud in the marketplace. The next thing you will see is a genic fraud, which is people flooding the market with fraudulent agents trying to intercept that transaction, which if you think about it, we've got faster, we've gotten more efficient, but we've got much less trust -- when someone sat across the table from you and sold you an ad, you knew who they were, you knew where they lived, you knew where they worked, all that good stuff. Now you have an agent talking to another agent. You have no idea if that agent is legitimate or not. You have now whether that agent is lying about the inventory they're trying to sell you. Our role will be to intercept that agent and make sure that agent is legitimate. We've already started building around that opportunity. So you've got this agentic AI opportunity that I think we play the same role in it. And then finally is the AI ads themselves. -- which is you've seen large platforms like Chat, -- you've seen smaller ones like perplexity and others who have said, "Hey, advertising is we have to have a way to monetize this future of ours. Like we're not going to get it through enterprise customers. We saw this with Netflix, when Netflix said, they will never ever, ever, ever sell ads. And then they said, "We can't support $12 billion of content costs by selling $30 subscriptions in India because we're not going to get enough of them, right? So we need to sell ads against that. Same thing is happening in the LLMs right now. And the same response that we've seen happen in all walled gardens is happening there, which is our advertisers are coming to us and saying, we will test this out we've added agency holding group tell us, they've informed their customers and they've informed the ad-based LLM, -- our customers will test this out. But until you have objective third-party measurement and objective third-party verification, we will not scale beyond test campaigns. So we see the LMs again as the next catalyst of growth for this company after kind of social and TV -- we have AI opportunity down the road, and we've been building some proof of concepts for that space, and I think there's a large opportunity. The last thing I'll say there is if you think about the size of that opportunity, it's probably the 1 where we're leased furthest along, but is the largest potential for -- all of that money or a large portion of that money that's going to go towards the LLM ad models is going to come from search. And that search is a $350 billion to $400 billion market. And that's a market where we generate 0 today. So as streaming dollars came from linear, that was all new money for us, dollars from search into the LLM as a potential to be all new dollars for us too.

And then what's the pricing model look like for AI slop stopper, which is great name, by the way. And also for like agentic verification, is it still per impression?

Yes. So right now, it's bundled into what we call the authentic ad, which is an impression-based measurement tool where we get paid on each impression. .

Okay. And then Jay, you have to ask on AI. So OpenEye eventually at least according to their own internal projections are going to have a very large digital ad business.

$100 billion.

[indiscernible] 2030. I can't remember what when it was a -- but -- why doesn't that evolve like search, right? So search you're functionally shot out of because Google does everything like why would open AI or any of their partners need to use DoubleVerify? .

I think it's -- there's a couple of reasons. The first is search is a monopoly and LMs are not a monopoly. There are many of them. They're all biting for space. And you've got Google in the space. You've got ChatGPT. You got anthropic who said they'll never sell ads, but they will. You've got grok and lots of others, perplexity and other smaller ones. So you have definitely a different playing field than you have with search. You also have advertisers who are looking at how are being placed in this environment. And it's very different than search. Initially, they're banner like ads, they're contextually based. So they're -- when you're asking questions around how to plan a guard there's Home Depot ads, right, as opposed to specific search query links, which are -- the environment is very specific and very driven. There's nothing really overly challenging about that for an advertiser. They are also sold on a CPA basis, which means if you don't click, you don't pay. The LLM ads are being sold basis. There's some talk about CPA. They're in an environment that can be very challenging for an advertiser. You may start that conversation about, hey, I want to dig up my backyard, right? And the Home Depot is like, "Great, this guy is going to be looking for backyard tools and then the end of the conversation is because I'm looking to bury 27 bodies, right? That -- does Home Depot want to be -- have their ad show up there? No, right? So contextually, it creates a different level of challenges that I think search just doesn't have, right? So you've got kind of competitive situation, I think it's different. You've got environmental situation different. And you've got just the way that the ads are currently, which are some level of display and now there's even talk of potentially video and video -- short video ads that could be run during LLM. So it's very different than search.

Okay. Perfect. And can I ask how are you using AI internally both for like your average knowledge worker and also for your devs?

Yes. So the engineering team is fully embracing AI tools to do just basically do their blocking and tackling to do coding. Our goal is for every engineer to be not writing a line of code by next year to have them actually managing a team of agents -- so each one will become a manager. Some of them love it. They think it's the greatest thing ever. Others are more old school and are challenged. But those school guys will be gone and who embrace it will be the future of the business. So we're driving efficiencies. As we noted in our call, we're driving margins up because of it. We had a strong -- strong Q1 margins, which beat our expectations. We raised earlier -- or late last year, earlier this year, we raised our target for EBITDA this year. We'll have less people in the house at the end of the year than we will at the beginning. So internally, it's driving more efficiency just from basic blocking and tackling. But the other big win for us is what DD is? Is it just a big contextualization engine. We look at stuff when we say -- we say, is this suitable or not suitable? Is this viewable or not viewable. We just make decisions. And there's nothing better for AI to play a role in is kind of that scaled opportunity to classify context. So we've mentioned in the past that we're now using this on our contextual engine to label content. This was something that we used to have humans do. We had Symantec scientists and linguist content, the label at once and put it into our machine learning models, and that would learn over time. But now we have AI doing that. So we're going to be removing 100 contractors this year just based on that capability. So internally, AI is making us more efficient, it's making our product more effective and it's helping us drive better margins.

Can I ask -- I don't know exposome, but like your token costs these days? Is it -- have you had to upgrade -- update your budget over the last couple of months? I wouldn't be surprised if the answer is yes.

So we have, but it was off a low number. So it is managed. We're managing centrally or CTO. I think the benefit over time will translate into higher margins anyway, like the token costs will not offset what we're able to save, as Mark said, around people that are -- we're using even outsourced to do classification. The other part that we haven't discussed is AI is really helping us do the actual gross margin part of our business much more effectively. So we're able to maintain gross mg 80%. So we are managing -- we know the costs will go up on token, but the net effect of using those tools is that we'll be able to grow with fewer people and grow margin..

Yes, I was going to just comment on that was -- may I also assume that if I'd asked you a year ago, what your head count growth will be over the next 3 years, that number is now somewhat lower meaningfully lower.

Yes. Yes, we'll be growing with fewer head count and [indiscernible] the company for sure. .

Perfect. Competitive landscape check historically, has been a pretty competitive space for you. I think like MTF fees have been generally deflationary what has changed there, if anything, over the last year, anything you expect to change in the future? And then does AI potentially bring in any new competitors that you've seen or heard about?

Yes. So I'll start with the sort of what's happening in the market and peers and the effect that it has on their business. I mean, we've been competing with similar companies for several years right? So the peers are known, the companies that can do exactly what DB does, they're far and few in between. The -- we have a product offering that's now broader. So we generally compete on just product offering and able to more and more services to our clients. The net effect of what we do is that we're growing the relationship with our clients, our top 100 last year spent $4.5 million on average. It was $2.2 million in 2022. And the way we're growing it is able to verify more and more of their impressions in whichever sectors that they're spending their dollars in. The effect of that is that the fee, the effective fee, the output might be a lower fee as we go into verticals where the overall CPM is lower social generally drives a lower CPM, especially outside of the U.S., and that leads to a lower fee that we receive because we kind of track CPM even though we don't charge on a CPM basis. So all this to say, we're happy to see the client growth, the top line growth, and if the effect of it is for the fee to be declining, that's an okay situation for us because we want to be able to verify wherever the client is spending. What's happening right now is dollars are shifting from sort of mobile online video to social as we've been discussing. We now have the products, but we're not yet penetrated in every single sector, the way we are on mobile, online and video and display. And so as that transition happens, there will be a moment where it's not dollar for dollar. But the opportunity is clear. What we have on mobile online video is 2x the revenue on versus measurement. Half of our measurement is already social. Once we get to those kinds of ratios, you can see how we'll get to what we think we can do, which is 50% of our revenue being social in CTV by 2029. That will be a nice benchmark at which point social will have scale also on activation. And so you'll see a very different business profile.

Right. And I imagine like the bull case there would be a half of your business or more is growing at a much higher rate. So we would see accelerating revenue growth.

Yes. Yes. And so just to speak to total revenue growth. So if you look at -- we look at Q1 as a good indicator of what we think is going to play for 26, which is we had higher growth in measurement that we had in activation. The higher growth in measures because half of it is social already and social grew 23%. So within that number, that is a significant amount of dollars that 1 to grow. -- activation within activation, which grew 6%, you have social activation that grew 92%, but obviously a very small base. As we continue to make that transition, those numbers over time will become a larger part of the business and will show in the total revenue growth. We're just not going to see that in 2026. The most important for us is to continue to show strong social measurement growth and strong activation growth and the overall profile will change as those sale.

Perfect. Nicolas, could you actually touch a little bit on your capital allocation policy. So you bought back, I think, $100 million of stock year-to-date. Why is that the right use of capital for your shareholders? And lay out for us the framework for the pace of buybacks from here?

Sure. So there's 3 prongs around the capital allocation. There's investing in the business. It's really important for us to continue to do that, even though we're able to show growing margins, we're still investing in the business. There is M&A opportunities and buybacks, which is now the third year that we're doing it in a row, and it's really part of how we think about capital allocation. We said at the beginning of the year that we will do more buybacks this year than we did last year. That was $123 million last year, and we're obviously well on our way. We're going to remain sort of measured as to which 1 of those 3, we do M&As it is a bit of a disruptive space in terms of valuation. There are companies that feel they're still worth a lot that raised 5 years ago, and there are companies that have no revenue and they feel that they've really worth a lot of money. So we're being very judicious on what we look at for M&A, especially because we think we have a good in-house opportunity to develop the products that we need for AI anyway. On the pace of the buybacks, yes, I'll stick to what we said, which is it will be more than last year for sure. We're not a trading on this. We basically set up plans and we kind of just make sure that we do say we're going to do -- we keep our eyes on making sure we more than cover stock-based comp, which we definitely will. If we do more than last year, that was $132 million, and that also benefits ultimately earnings per share.

Perfect. Then Nicola one other question, which is I'm not mistaken, your Q2 guide is like a small deceleration on revenue -- if you look at -- the Street, I think it's actually the slowest quarterly growth, so there's going to be a reacceleration in the back half -- let's assume you do reaccelerate in the back half? Like what's really going to drive that? What confidence do you have that you can do that?

Yes. So first to talk about Q2, it comes off a comp last year that was 21% growth. we're lapping a strong last year number. What's going to happen in the second half is, again, if you project out what we did in Q1, if within the headline 10% growth, if you look at social measurement and social activation and you extrapolate the rest of the year, you have a different -- a very different pattern of growth in the second half. We are still winning new logos. We had a win rate in the first quarter of 77%, where we didn't displace anybody to get that business. That's also grow into the second half. And we still have opportunity to scale the clients that we won once mote shut off at the end of 2023. That takes time for those clients trips, but there's a very names. And as we're able to upsell them to our products, you also see that the benefit of that in the second half.

All right. Perfect. Mark, we're almost out of time. But if there's like 1 or 2 last thoughts you want to leave with the audience. What do you think are 1 of the 2 biggest misconceptions that investors have about Double Verify that you often get asked about in meetings? .

Yes. I think there's 2. The first is we get lumped with a broader category of ad tech -- and really 95% of those companies in ad tech do 1 of 2 things, buy or sell media. The buying and selling of media is going to be very disrupted by agentic and is pretty much controlled by the -- we don't buy or sell media, and we verify in the places where most of those companies don't even have an opportunity to play. So we're -- obviously, we've talked here a lot about I mean we're 1 of the few companies that sees across all the platforms across TikTok and YouTube and Meta and Netflix and all the open web. We see everything. So we're not concentrated in just challenge buying and selling of Open Web impression universe. I think that's one. The second is the whole story around, well, AI is just going to eat everything and you're part of their dinner. We don't think so. The role we've played for all of the other places where ads were being sold as a role we're going to play in the LMs. If anything, the AI universe has created more chaos for advertisers Navigate, which has made verification that much more important. So we see AI as a huge opportunity to save money, huge opportunity to grow top line and not a challenge to our business.

Perfect. All right. Mark, Nicolas. This was wonderful. Thank you for coming.

Thank you.

Thank you, Mark.