Good morning, everybody. Thanks for finding your way to Stockholm today or joining online. My name is Ingo Middelmenne, and as the Head of European Investor Relations of Verve. I'd like to cordially welcome you to this year's Capital Markets Day of our company. After intense weeks of planning, our whole team and I are really thrilled to finally have you here as our guests today. As usual, we've prepared some sophisticated deep dives into the most recent developments at Verve from an operational and financial perspective, as well as valuable industry insight and sessions with high-level tech experts from our group. First up, let me present today's speakers to you. From our company's executive team, you will meet today Remco Westermann, our CEO, Christian Duus, our CFO, Mishel Alon, our CBO, and Paul Hayton , the CTO of our subsidiary Dataseat.
We are very proud to have an international guest speaker with deep industry insight. Eric Seufert is going to do today's keynote. He's a media strategist, quantitative marketer, and author, and I'm really looking forward to his speech. He came all the way from the U.S. today. What can you expect from Verve today on this Capital Markets Day? The day will be split into two parts. We have two sessions, basically. The first is going to be the business update that's going to be run by Remco and Christian with the commercial update and the financial update. After this, we'll have an extensive Q&A session, which can also be followed online. After the lunch break, we're coming up with our expert sessions, the keynote I just mentioned, as well as deep dives into ID-less advertising and AI in advertising.
Now I'm handing you over to Remco Westermann, the CEO of Verve Group , who will lead you through the first part of today's Capital Markets Day. Remco, the stage is all yours.
Thank you. Good morning, everybody. People here in the room, but also online. I would like to take you through the business update and starting with our equity story. Most here in the room and also many online know the company, so for some, it will be a bit repetitive, but I think it's good always to realize what are we aiming for, where are we, and what's our story. Starting with the market. Advertising is a very diverse, very crowded, I would even say, market. It's becoming more and more complex. Where advertising started with a few newspapers, at a certain point, radio, then a few TV stations, the diversity has become enormous. We still have radio, we still have television, but there's a lot of social, different channels that have come up. Here you see basically the life cycle of advertising. A lot of those channels go down.
Traditional television, why should you send one ad to everybody if you can send it on a digital TV, connected TV, individually per household? You see that, let's say, some channels are going down, some channels are still in a growth phase, and we have the emerging channels. Emerging channels, is that what we are focusing on? The advantage of the emerging channels is market positions are not yet taken, or at least are pretty young, and the markets are growing. Working in a growth market, of course, is much more exciting than working in a market that's declining. The channels that we're talking about: mobile, the personal device, connected TV, so the old TV set, but now digitalized, retail media coming up a lot.
Amazon came up, making losses over many years and now being, let's say, the gigantic online retailer, but also traditional retailers, Walmart, Circle K, and many, many others are really getting into this space. Why not make money from the customer contacts that you have? Why not help the CPGs, so the consumer goods producers, to bring people into the store or to decide or to help them make a different decision on their product choice? Digital out of home, the screens outside. It was also previously, it was somebody going around and sticking those things, and now it is really digital, getting more digital, and additional audio, podcasts, for example, also getting in. Those are the channels that we're concentrating on. A bit about the market. We talk about a huge market, advertising, $1.1 trillion spent worldwide per year.
In the U.S., over $1,200 spent per ad per year on advertising. $600 million of that is digital, so there is still a part that is traditional, television especially. In the digital part, you see a very big part is mobile. Mobile, as you see, is a growth sector. It's growing nicely. The most personal device has already sat, and it's 96% of our revenues. Mobile is really where we are focusing on. Within mobile, it's mostly in-app, but we also do mobile app. Connected TV, also a large segment, still smaller. If you compare to linear, it's still small compared to linear. That means also in the future, there will be a lot more connected TV happening. Desktop, that's already more in the, what we say, the mature segments. That's also what we're not concentrating on.
We can do it because sometimes you need it to fulfill a campaign for an agency, but it's a small part of our revenues. That's where Google and many others have taken strong positions. Talking about our main market, the U.S. U.S. is almost 80% of our revenues, 79% to be exact. Important market. Why are we so big in the U.S.? It has a bit to do with how we started via acquisitions in the advertising part, but has also to do in the U.S. Its scaling is much easier. In Europe, you really have to go country by country. Language, rules, agencies, everything is much more local. The same counts for Asia, where often the CPM, so the prices per ad are lower, and also for LATAM, where we have the same, where the countries are maybe bigger, but let's say the revenues overall per country are smaller.
The U.S. is a super interesting market for us. It's growing. You see here mobile expectation growing 10% per year. I'll get to market growth later in the presentation because there's always some hiccups also in growth, even though there is a continuous growth there. In the U.S., we're super proud that we have been able to build a strong position. iOS, Apple's ecosystem, in-app, we're seen as the number one market share-wise. That doesn't say that we have 20% or 30% market share. There's still a lot of players there, and we have this, let's say, overall market shares are not huge, but we are number one. That helps us to open a lot of doors. On the trust index, last presentation, we were one, now we're number three.
Measuring is not always easy in those things, but I'm super happy that we are seen as one of the top recognized trusted parties in the market. That helps a lot. Advertising is about trust. There's a lot of fraud, a lot of, let's say, intransparency. In that sense, it makes a lot of sense to be trusted and to be helped because of that. To our business model, we are a tech company. We are running a tech platform. We match the advertiser with the publisher. I talk about digital channels, so all the matching is digital.
It's much more complex than in this picture, but we want to show basically what we try to do is get the best result for the advertiser, often working with his agency, and then the best result for the publisher, which means highest price per ad, which means good fill rate, and for the advertiser, it's really reaching the right target group in the best efficient way. A bit more complex. What's happening? If you open a page, there is an ad on that page, or there's an ad spot on that page, and while you open the page, we get the information that this ad is for sale. We technically connect it to the publisher. We have a contract. This information comes to us, the ad request comes to us, we enrich it with data, and we make it available to the demand side.
That's our supply side platform is doing that. On the demand side, you have a demand side platform where you have several advertisers or the agencies putting in, I want to reach a certain target group, a certain price per ad, certain, I would say, other targets, viewability, attention rate, those kinds of things. That's what they put in the system, and then there's a bidding between the two. The highest bidder wins, the ad of the highest bidder is being rendered. In this case, it would be Adidas, and the Adidas ad is winning. Talking about prices per thousand ads, CPM is the kind of term. This all has to happen in less than 100 milliseconds because during the time that the page is loading, we have to do this auction process and to make sure that the ad is shown before the page is fully loaded.
Otherwise, you have an empty ad spot, and that's a waste of money for everybody. It's not only fast, we also do a big volume. We serve 2 billion- 5 billion ads per day. That's what's all happening there in the heart, in the technology. I'll come to our migration project a bit later on the slides, but we have really started with buying some companies, we brought them together, we're integrating those to a single platform, and that single platform is doing all of this. Supply side platform and demand side platform, to just specify. Value chain. We say we want to do everything between advertiser and publisher, which means that we have the full technology stack. We have the DSP, we have an SSP, and we have the marketplace and the bidding part in the middle.
If you look at the financial part of it, it's the advertiser that pays, the agency that pays the money then for the ads, and sometimes they do a markup towards the advertiser, of course, mostly. From $100 that's paid, roughly 20% goes to the DSP, roughly 20% goes to the SSP, and roughly 10% for data, bidding, matching, etc. 50% is taken by the chain, and 50% goes to the publisher. That's not a lot, you would say. If you then look, for example, at the Department of Justice files for their Google investigation, you see that Google takes 80% of margin on external volume. This sector is still not very margin efficient, I would say. There's a lot of point solutions, parties that only do DSP or only do SSP.
If you do the whole chain, you can become more efficient, you can make more margin, but you can also give to the publisher and to the advertiser. That's an important point. It's a market that is not cost sensitive at the moment. The market is really looking at best matches, at selling the ads. There's still a lot of unsold ad inventory, and there's still a lot of bad targeting. That's more important even than the margins, but also margins, of course, play a role. That's what we're doing. When we entered this market, we were looking, hey, there's many parties here, but there's also many inefficiencies. What's really changing in this market? What's changing is privacy. Privacy is becoming more and more important. Being a European-based company, Stockholm-based, Sweden-based, we very early became already aware of GDPR.
GDPR in Europe made a lot of, how to say it, emotions. If you ask a consumer, do you give consent that your data are being used? 80% says no. There's differences between countries, but that's a rough number. The majority, by far the majority, says no. That means that the advertising market is being disrupted, is changing. The advertising market was, the digital advertising market was used to targeting with cookies. If I have a cookie, I can build a profile because I see this cookie all the time coming back. I can say this person is interested in sports, this person has a certain profile. I know when an ad from a phone, for example, of this person comes, that I have to send this ad to maybe Adidas or somebody else who is going for sports goods. That changes.
If there's no consent, if there's no identifier, I need to do targeting differently. We saw that as a disruption of the market, and when we started getting into advertising, we said that's really the big opportunity to go for. Later today, Mishel and Paul will go more in depth, but ID-less targeting, so targeting without IDs, is something that we're really focused on, and that has given us a strong market position. It differentiates us. You see here, iOS, Apple, we have really been able, you saw the market share before, but we have also been able to really grow our revenues in this part. We've developed ATOM, a solution for this, it is one of our ID-less solutions. We also see that Android is moving in that direction, and also mobile web, etc., we see already movements in that direction.
So ID-less is becoming more important, and that's where we really focus on and differentiate. A bit about the mix, what are we doing? I said mobile is very strong, 65,000 apps where we integrated. Mobile is 96% of our revenues. Connected TV, we have a lot of direct supply. We reached 200 million connected TV screens. There's not a lot of differentiation in connected TV. That's not only a problem for us, but for most parties. We are working on that because we also want to extend our mobile share. It's now 4%, but we want to get it bigger, which means we need to work more on competitive advantages there, especially our strong data position that we get for mobile. Based on IP addresses, we can use those same data towards the households.
That's what we're working on to also grow our CTV, our connected TV. We reach over 2.5 billion consumers. We have 954 software clients by the end of Q2, large software clients by the end of Q2. Christian will cover that later in his presentation, customer numbers. We do 1 trillion ad impressions in, we did 1 trillion ad impressions in the last 12 months. Demand supply, we started more supply-focused. Supply is where the data are, supply is where the publisher is, where you get the ad. We would like, as I showed before, to have equality between the two, demand and supply, because if you really get every dollar that comes on the demand side or your own demand side or your own supply, you're most efficient. I showed the efficiencies before. We're not there yet.
Last year with Jun Group, we did a big move forward on the demand side. We also organically are growing fast there. It's 25% demand side, 75% supply. Ideally, we would get to 50-50. U.S., I mentioned already, 79% U.S. market, 9% Europe, 12% rest of the world, and mobile, I mentioned already. The other channels, they're there, you don't see them, they're still very small, digital out of home, retail media, and audio, digital audio, but they're also growing and also we're building positions there. A bit of history. We started as a publisher, a games publisher, started this almost 13 years ago. We have been growing every year. We have been growing our revenues. We have been growing our EBITDA. We transitioned from a publisher, which is a very, very difficult position to grow organically, towards an advertising company.
With that, we renamed the company from Gamigo to Media and Games Invest and to Verve. With Verve now, we are an advertising company, even though we have still a bit of games, but it's under 10%, so we are really focused on an advertising company. Last year, 36% revenue growth, 30% EBITDA margin, and we're here to further grow. That was a bit of our equity story, how we go. Now the commercial update. Coming to the commercial update for Q2. I'll cover the numbers and headlines, and then Christian will go more in detail. Normally at Capital Markets Day, we, let's say, come here with all the numbers as the surprise, or let's say here show them first time. We all already went out on Thursday night with new guidance and on Friday morning with our Q2 numbers.
Coming to that a bit in a minute, we had more issues on our platform unification than we expected, and it lasted longer, it cost a bit more revenue, and that also brought us to the conclusion that we have to change our guidance for this year. Not happy with that, it's running a company, and it's not always a straight line, but the company is stronger than before. Unification is now done. I'll come to that later in the presentation. That's the reason that the numbers were already out. Nevertheless, we did 10% revenue growth, 1% adjusted EBITDA growth. We reached a 28% EBITDA margin, leveraged at 2.5%, so on the high side of where we want to be. 22% software clients' growth, that's overall growth. The large clients didn't grow that fast. 98% retention rate. Our customers did less revenues, but they stayed, and that's very important.
Very good retention rate. Christian, as mentioned, will cover that later in the presentation. What are the main highlights? I mentioned it already before. I would start with number three, platform unification. We are a tech company, and a tech company has a tech platform. Actually, when we started this, we did acquisitions, and we had several tech platforms. Our biggest part of business, as I showed before, is mobile, it's in-app, and it's our supply side. We have been integrating supply side platforms into two platforms before of the acquisitions that we did, and those two big platforms were being integrated in Q2. That was not without pain, as you see on the slide. It lasted longer, and the issues were a bit more. I'll go further into detail later in the presentation, but it hurt us.
That means that we had less revenues and that we had more cost, and that is what you see in the numbers. That is basically one point. The good thing there is that that unification is now done, and that we are growing with a single platform, single in-app platform, much more efficient than we did before. We have always, if you do acquisitions, and we already said that when we did gaming acquisitions, you need to integrate those companies. If you keep them independently, if you keep the platforms and technology independently, it's very inefficient, and it becomes more and more difficult to manage at a certain point. I'm super happy that we did it. I'm super happy that we're out of it, but it really hurt. Liberation Day impact, number one. The market is a bit softer. The market is not like, wow, it's all great.
If you look at the guidances or also at the reports from our competitors, from people in the sector, you see that the market is not strong. I'm not here to blame the market for our Q2. We had a one-time effect, and in this market, even though it's weaker, we can grow. We had some weaker customers. On the other hand, you can compensate it by new customers, by other customers. The market is a bit weaker. It's not wow, but it's still okay. As mentioned already, growing our customer base, 22% increase in overall number of customers, 10% organic. That's maybe also important to mention. Our demand side grew, so we had the issues on the supply side. 98% retention rate, I mentioned that already before.
We have customers that are super happy, that stay, and a company that has a very, let's say, stable customer base, of course, is much easier to grow than if you also lose a lot of customers. We have a solid base to further grow. The integration of June, yeah, Mishel is here, who will present later. We acquired Jun Group in July last year, or let's say actually 1st of August . The team is now integrated. We see the costs and the revenue synergies, and we're preparing now also the rebranding. The Verve brand is strong, and it's stronger if everything is under the Verve brand. Also, the June brand, of course, it was, yeah, it is known to all the agencies that they work with.
This is a process that also takes some time, that needs to be well prepared, and we are now working on doing the rebranding and finishing it also this year. Sales team expansion. We have a good product. It's B2B. There are 4,000 agencies, over 4,000 agencies in the U.S. If you want to sell it, you need to go to those agencies. We have 35 sellers at the moment in the U.S. With 35 sellers, or for 35 hunters, we also have account managers and other people behind that. With 35 hunters, you don't have enough people to cover 4,000 agencies. We also have the old coast, the big agencies, the WPs, etc. They have also hundreds, thousands of people. You need to have a bigger staff or a bigger number of sellers to really address those. We have started in Q2 with hiring.
That's painful because it means a lot of interviews. It also means that if you hire a seller, that the seller is not always good, but you normally don't find out the first week. It takes half a year, and then the sellers should really start making money. By the end of the year, you know it for sure. That means also that some hires are not good, and you have to redo that. It's a painful process, but it has to happen. We want to add 100 sellers, or we want to go to 150 sellers actually in the U.S. by the end of next year from the 35 that we are now. Regional adaption of sales strategy. If you have more people, you can go more regional. The U.S. is very regional. You need an office in Chicago.
You need an office in the West Coast or in San Francisco, in LA, etc. There, you're closer to your customers, so that makes life also easier. Segment focus, another important point. I'll cover that a bit later. You have different industries, different industries have different experts, and also for advertising, there's differences. Capital market focus. I'll quickly go through, and Christian will cover it in more detail. Our uplisting strengthens IR. Happy to have Ingo on board. Bond placement and our capital increase, but Christian will go more in detail. Those were our highlights and challenges. Normally, I put highlights first on the slides and challenges second, but in this case, because of unification, we put the challenges first and then the highlights second. A bit about the market, because there's a lot of talking about the market is weak, the market is not weak, etc.
What you see here is the price index. The price index goes with demand and supply. If it goes down, there is less demand. If it goes up, there's more demand. What we see here is that advertising is not a straight line from a development point. We have seen different, let's say, down phases. The economy is not working so well. Inflation is going up. Uncertainty is a factor there. There's some named here. What we see now is that it's not going up so fast as we've mostly seen before. Still, as I said before, this is not a reason to not grow. There are market dynamics. Programmatic advertising is growing faster than the other parts of the market. We have the ability to further gain market share, and we're doing that. We have a high level of customer satisfaction, 98% retention rate.
We also had 94%, 96% in the past, but we have a very high retention rate. Technology. We see, and we further believe, that our ID-less targeting is really very strong and giving us more market share and giving us more opportunity. What are our growth drivers? Why are we growing? First of all, market growth. Yes, there's hiccups, but there is market growth. Having acting in a market with tailwinds is, of course, much easier than if the market goes down. Emerging channels, as I started with in the presentation. Customer expansion, B2B, mentioned that before. Add more customers, add more agencies, add more brands, add more publishers, add more verticals, add more GOs. Have enough account management and a good product to really skill those customers. Typical B2B. Differentiation. There's a ton of companies in this market. New products, innovation, ID-less, I mentioned that a few times.
Also new ad formats. Fullscreen video ads, those things. Partly things that our platforms didn't have, partly things where we really innovate, where we give new formats. Having an SDK, so software integrated in an app, gives us the possibility to really show new formats, to do things. New channels, audio, podcasts, is really something where we're now focusing on and where we see a lot of things. A lot of podcasters don't get enough viewers on their podcast, viewers, sorry, listeners. Listeners on their podcast. We helped them with that. We helped them to generate traffic for that. That's an interesting market. Platform synergies. It's technology. Scaling makes sense. The more you scale, the more efficient you get. The bigger you get, the more efficient. That's also the reason for unifying the platforms because otherwise that doesn't work like that.
More size gives you more data and the possibility to invest more in AI. The combination of AI and data, and Paul will especially go into that later, is super important. That's where you can also differentiate a company. That's where you can gain more, become more efficient, be more relevant. Those are the four growth drivers: market growth, customer expansion, new products, platform synergies. That drives us. What's our strategic rationale? Verticalization by industry. We really go more deep in the industry, and I'll cover these points in more detail. Multi-channel approach. We want to do a full customer journey. Yes, we are very strong in in-app, but it makes a lot of sense to also do CTV, digital audio, digital out of home, retail media.
We want to follow the customer, the consumer, where he is, and also at the right time show the right ad to get the most effective outcome. Very important, mentioned already, privacy-first, high-quality standards. There's a lot of fraud still in this market, made-for-advertising pages. Pages that are really there to just lure people in to show them ads. A lot of, let's say, bad traffic. That's what people don't like. If you are a brand, you want to show your ad in a good environment. The privacy thing, you don't want to upset the consumer also. It's not only that there's less data available, but also as an advertiser, as a brand, you want to work with the preferences of a consumer and not go against it. Verticalization here, some examples, and the examples are coming back later in the presentation.
One of our segments where we're strong and which we're really strongly focusing on is digital brands. Those can be brands like Apple and different other ones, but we have Otto here as an example. It's the third largest German digital retailer where we showed good results. If we really understand the segment, we can use our standard tools. We can tailor them. We can really bring to the sector. We can really bring sector expertise in there. That makes a lot of sense. CPG, very important, fast-moving consumer goods. CPG is more the U.S. term. Also, they are driving incremental sales, making sure that people sell their products. Add to cart, for example, is something that we do. Medical and health, another example. Really making sure that we help Pfizer in this effect to sell their product, reducing the cost to reach customers, patients, sorry, and providers.
That's really important. With size, we have the possibility to specialize. We have the possibility to hire experts of those segments. If you have the expert, it's easier to talk to the agency that sometimes doesn't even have the expertise, and it's easier to talk to the customer who comes from that part. Very important. With scaling our salesforce, with scaling our team, we have the possibility to go into segments. Channels. Mobile, I mentioned already before, that's our strongest, very strong direct supply base. There's a lot of intermediating in this business where there's indirect supply. We have direct, we are integrated in apps. We have direct integrations to serve the ads. Premium supply, you want to be in a good environment if you show your ad. Also very important. That's what we've been focusing on, and that's the biggest part of our business now.
Connected TV, we have a strong position on the supply side. We reach already 60% of the U.S. households with direct CTV. As mentioned before, we are working on further getting our sales story, getting our USPs in that part. It's still a very fragmented market. If you look at the other players, nobody here really has a competitive advantage. There's a big opportunity that we see and that we're working on. Retail media, a lot of our customers are already from the CPG area or the CPG part or even also on the retail side. We didn't very much specialize on it. We are now. Retail media basically means that you want to do the full chain. You want to reach the customer outside of the store or in the digital store, but you also want to be able to even transplant that into the store.
More and more stores are getting media in the store. Also making those available, making sure that you target the right people and that you influence the choice in the store because that's where a lot of decisions are taken. Somebody drinks normally Pepsi Coke. Coca-Cola would like to get this person to drink Coca-Cola. That's something that you can easily do in the store. Digital out of home, mentioned before, more and more screens. Not only the big ones that you all think about, but also, for example, in Ubers now, you get Ubers get in the U.S., you have all the screens that you see with ads. There's more and more spaces with ads on airports, everywhere. They're digital. Also, there it's important to really reach people in the right way. Audio podcasts, another channel, very important, growing, still totally undermonetized, not a lot of ads yet.
A lot of podcasters, for example, that also don't have enough reach on their channels. A lot of opportunity there to also go in. Also here, because we are becoming bigger, we have the possibility to specialize in those things and to really go in. We don't do everything at the same time. We're talking here about our three to five years and beyond. This is the channels that we really want to serve. The opportunity is now, and we need to start working where we see really the opportunities. Privacy-first, quality standards. A lot is happening in that respect. Apple introduced in 2021, ATT, which was really banning or let's say making targeting much more difficult. No cookies, no ID, the identifier from Apple anymore. The IAB, META, we have a lot of, let's say, things that are happening in the privacy space.
The trend, the tendency is towards more and more privacy, less and less signals that an advertiser can rely on. So ID-less targeting is becoming standard. It takes the time. We see that with agencies. They're sometimes very slow adopters. You rather stick to what you know, you rather stick to what you have. At a certain point, if the targeting is less, you have to change. We see that happening now. First-party data, we also see that it's more and more important to be close to the publisher, to be where the data are, or to really build a direct bridge to your consumers, loyalty apps, for example. Also, there's a lot of data, first-party data. Then AI-powered programmatic optimization, you cannot do this manually. This is all too complex. You need AI to do this. AI is super important. We're covering that later.
The good thing is if you do it properly, you have a much better result. People like to see ads that are relevant for them. What we have, it's the cookie apocalypse. Cookies are not great. If you have a mobile phone and you use it and the cookie is building a certain profile, and you have a kid that borrows your phone and is doing totally different things, your cookie is totally, yeah, abused, misused, whatever. Sorry. I was looking for the right word. There's a first-mover attention. We have a first-mover advantage in ID-less targeting. ID-less targeting is really changing this market. Of course, contextual targeting, you can show an ad in the car ad on a car app. That's easy, but you don't have enough reach. It's much more complex, as I just showed. Privacy compliance is a competitive advantage that we're working on.
Products, we get to cover that later. ATOM, we've talked about it for many years. It's at scale now. We really see the advantage of it. It's helping us, but we have many more things, solutions on the ID-less side. Strategic roadmap, platform unification for efficiency and scaling. We wanted to really show you what we are doing in that front, to go more into detail, because it's easy to say, hey, we had issues with platform unification, but we want to go more in detail here. We acquired 16 companies, one six, to start our ad tech business. Actually, 15 we acquired in the beginning, and then Jun Group last year was the 16th. All those companies, almost all those companies had technology. They all had a team. We acquired 16 teams.
We acquired five supply-side platforms, three demand-side platforms, five SDKs, so SDKs are the ones that are integrated in the apps, five data lakes, and five plus infrastructures. Some had a bit more, some a bit less. This is what we had to unify. The target, one team, one supply-side platform, one demand-side platform, one SDK, one data lake, one infrastructure. All of those connected to each other and communicating with each other. We have done a lot of that already. From the five supply-side platforms, we were back to two. Those two big ones we integrated, and that's where the majority of our business is, in-app supply-side platform. Examples of hiccups that we had: a load balancer didn't function well. We do almost a trillion biddings per day on that platform, on the combined platform.
It's so huge, and those were things that we were not able to test at scale before we did the unification. There was a lot of preparation. More than a year, our engineers and product people had been preparing this migration. There are some things that you cannot test. Such a load you cannot test before. If you have a hiccup in the platform, or if you have an issue in the platform, even if it's a small one, on the other side, there's an AI system. If that AI system, the bid is not accepted, or let's say their request is not accepted, that means that it stops and only starts slowly testing at a certain point again. With all this electronics that we're working with, all the software, there are things that just take longer to ramp up again. Issues were more than we expected.
They took a bit longer. The good thing is this in-app unification is done now. We still have to do CTV and web. Those are being prepared, but there, the volumes are a lot lower. We have seen also before with other platform migrations, we didn't have these kinds of issues. This is a one-time, and we have solved it. We still have to do the SDKs further. You see the, how to say it, the accomplished points there. What's still open on the unification, we still have to further unify our SDKs. That can be done step by step. That's not one big thing where we have to do everything at the same time. As mentioned, web and the CTV platform. Smaller things where we don't expect issues. The big one was this in-app unification. The good thing now, with a unified platform, we're more efficient.
We can further scale it. We can concentrate on building features for one platform instead of for several platforms. The company is really fit for the future with this. Coming to the next strategic roadmap point, investment into sales and geo-expansion. I mentioned it already before. U.S., our main market, 97% of our revenues. We have 35 people, hunters, that go to agencies to get new customers. As mentioned before, that's not enough. We want to go to 150. I showed it. U.S. is our biggest market, and we want to really grow strongly there. We also want to internationalize. With all the products we have, with all the sector expertise, it makes sense to go into other markets. This year, we're going into the U.K., Scandinavia, Brazil, and Mexico, which means also we had already some things there, but we're really focusing on it now.
It means on the supply and on the demand side and really building it. For next year, we want to add a few markets to that. That will be done in the next years. Each year, we want to add some markets and, of course, to scale the other ones. M&A. We have done M&A in the past. The market is consolidating. We see a lot of potential. There are a lot of companies that want to sell. If you look at the ad tech market, there's over 100 SSPs, over 100 DSPs, over 200 data partners, over 4,000 agencies. It's a very crowded market. Customer demands companies that do more channels, one-stop shops, basically. As such, there's a natural tendency to this. Also, the ability to scale a larger company can invest more in AI, in technology, all the things. It makes sense that companies grow.
We can grow organically. That's our main focus, and to make that very clear. There might be, and that's the reason we have it here, opportunities to really scale faster, which can be really more acquire, getting extra sales teams. I mean, Jun Group is a very nice example of really getting a lot of agency relations, salespeople on board, which we did, and revenues, of course. Also for regional access, building up a new market is not always easy. Sometimes it makes sense to rather do a small acquisition. We will be very disciplined. I think that's very important to say because M&A is only the add-on and should be the exception. Then our team, strengthening the core team. We're continuously working on getting a better team, making sure that we have a super motivated team, that we have a very high-performing team.
We did also change in the core team. Christian, CFO, and myself, and Sameer, and Alex were already in, but we added Mishel, Brezana, and David to really get a few more shoulders to carry the business on and to also get a bit more focus in the teams. Our overall size, so 97% of the revenues in the U.S., 350 out of our 850 employees are there. We have a quite large employee base in Europe, and we are building a larger base in Asia. We have a big development center in Bangalore in India. Just as an idea, a developer, and that's a rough number, costs maybe $300,000 salary per year in the U.S., $150,000 in Europe, and $75,000 in India. We also look at keeping our personnel cost down, or let's say efficient, very important. Evolving from publisher to ad tech platform.
I showed the slide already before, the summary slide. We're coming a long way, and we have a lot of potential for the future. That's this slide. We stick to our midterm guidance, 25%- 30% revenue CAGR. This company can grow. This company is very strong, equipped to grow. I hope that I was able to show that. With a strong EBITDA margin, 30%- 35%. Yes, there's some seasonality, and yes, in Q2 we were with 28%, a bit under it. This company will be doing 30%- 35% EBITDA, 20%- 25% EBIT margin, and our net leverage between 1.5% and 2.5%. We're on the higher side now. We want to bring that down, and with more cash flow generation, we will do that. The target is to get the next, how to say it, point in the future. We want to get over a billion revenues.
We want to get over $330 million of EBITDA, and the growth should majority-wise be organic. That brings me to the end of my part, and I would ask Christian to continue with the financial part. Thanks.
Thank you very much, Remco. Good to meet you all here today. My name is Christian Duus, and for those I haven't had the pleasure of meeting before, I joined as CFO from January 1 this year. I'll go through the performance of Q2, but before I dig into the numbers, allow me just to summarize what I think has been a very productive first half of the year in terms of capital market transactions, or capital market milestones and achievements. Starting in chronological order, you'll note that we had a bond refinancing in April.
We placed a €500 million bond, unstructured, refinancing our existing bonds, and we did that at, I would believe, quite attractive terms in the sense of three months' Euribor plus 4% credit margin. That means today, we, with a roughly two percentage points on Euribor plus the 4%, we pay 6% on our bond debt. This is going to lower our cash interest expenses through the year as we move forward with an estimated impact of $12.5 million and will help our free cash flows. It obviously also gives us stability in terms of our financing and our capital structure for the coming years. It's a four-year duration and a big achievement. Secondly, we uplisted to the General Standard in Frankfurt from the Scale segment to the regulated part.
This is a key achievement we're very proud of, especially also the inclusion in the SDAX, which has a lot of benefits, both in terms of credibility and awareness in the broader European investor spectrum, but certainly also because of being in the SDAX, you're then automatically included in various funds that mirror the index. That was a major achievement for the year. It was a lot of preparation, with a very successful outcome. Thirdly, we had a directed capital raise here in June, SEK 360 million , so €32 million, well subscribed amongst institutional investors. This gives us a good structure between debt and equity for the time going forward, and also to have proceeds to fund growth, whether they be more organic investments or on more inorganic investments.
All in all, a very successful and very active first half of the year for me coming on board, also with the closing of the 2024 annual report. I'm very happy to be here today and can show these milestones. What is not on this slide is we are also, with the addition of Ingo Middelmenne, expanding and bringing a very seasoned person in our Investor Relations team, and I think it will be evident over the coming six months, probably especially for the analysts that are here today, that we are seeking to lift our Investor Relations communication and practices. Now I'll move into the actual performance of Q2. Just focusing here on the highlights, we grew as a group 10% on top line, 10% revenue growth versus last year.
This is, however, a mix of developments in that we had a supply side of the business, the SSP part of the business, which declined 3%. This is because of the unification technology issues, and I'll come back to go more into that. On the other hand, we had 82% growth in our demand side activities. It's a mix of different things that are going on in the business, and we had a very strong growth contribution from our DSP side. I just want to clarify, this means that the revenue challenges that we had in Q2 are purely from the SSP side and specifically from the marketplace activities. The performance of the SSP side, and especially the marketplace activities, was obviously not up to our expectations. It's very clear, and I'll come back to what we look at for the rest of the year.
Despite the revenue gap that emerged out of this compared to our expectations, we delivered an adjusted EBITDA, which is marginally higher compared to the same period last year at $29.5 million. This is, of course, also helped by the consolidation effect of Jun Group compared to this period last year. We continue to invest. We continue to keep our plans to invest in the platform and in our sales capacity. Remco already referenced that, and that's also potentially why you don't see a higher lift in our adjusted EBITDA. Let me then dig down a little bit in the business KPIs and how our business KPIs are developing. There's a lot on this slide, so let me try to break it down and just say, overall, we continue to grow the ad tech business year- on- year.
This is evidenced here in the right-hand corner with a 15% uptake on ad impressions. That's how much volume flows through our systems. It's up 15% versus the same period last year. Also, in the total number of customers that you see here on the top-hand line, we grew 22%. We actually added 561 customers versus the same point last year on total number of customers. The business is growing from that perspective. We, however, also observed that we were handling less budget from our existing customers. You can see that in the net dollar expansion rate here on the bottom corner, which was 92%. That essentially means that we handle 8% less volume from our existing customers if you compare one year back in time.
The big effect there comes actually from us being able to not scale our customers sufficiently and also from not onboarding customers due to some of these technical issues that we had. We simply do not want to onboard new customers when the platform is not stable. You make a very bad impression. The silver lining and the positive news is we did not lose customers. We actually had a client retention rate of 98% for the customers, the software clients that are above $100,000. You can see here that they spent less. For the customers that are above $100,000, some of them spent less to the degree where they went under the threshold of $100,000 growth per year, but they didn't leave us. We have customer retention of 98%, which is 2% churn. That's pretty solid, even for a tech company, by my standards. Net-net, we're growing.
Our customers used us less in Q2 because of the situation of the technical setbacks, but we haven't lost them. It just fell under the threshold of how we define a large software client, which is $100,000 growth. I think the KPIs reflect and pretty much reflect the situation that we've gone through in Q2. Nothing surprising in that front. We do expect the KPIs to improve, and this being a temporary impact on the KPIs. The KPIs will improve through Q3 and Q4. You may then ask me after the session, by how much and how quickly. This is difficult to assess. We have to take into account that we did have a sustained impact into July and just the beginning of August. We expect improvements later in the quarter, but you also have to weigh it with seasonality because not every month is the same weight.
Therefore, it's actually a difficult question to assess. I'm confident that we will see an improvement in these business KPIs as we move through the year. That's a little bit about how the business is moving under the motor of the business. If I then go to actual performance and the key highlights of financial performance for Q2, we see that total revenues grew by 10%, as referenced by Remco. It's up from $97 million to $106 million. We had organic growth adjusted for acquisition, Jun Group mainly, and also for currency effects, which was -4%. We did have currency headwinds, dollar to euro for this quarter. All those adjustments altogether mean that our organic growth was -4%. Net-net, while our revenues disappointed, we see personnel costs in line with our plan, actually below what we intended to deploy for the year. Those are in check.
We did have an effect, and I'll come back to that, of extra technology costs for Q2. Overall, we achieved a 28% adjusted EBITDA margin, slightly up from a comparable period last year. We achieved a 21% EBIT margin. When we look at the technology costs that hit us in Q2, it's roughly $4 million. I'll come back to you. That translates into a 3.7 percentage point effect for Q2. You can do the math yourself. Still very decent profitability for a company, even for a tech company. We, of course, expect the operational margin, as it was impacted in Q2 by these effects, to improve over the year. Cash flow generation was $5 million positive in totality. Before networking effects, it was actually $15 million, but then there was a negative effect from cash flow for networking capital. All in total, it is + $5 million.
Again, impacted by lower revenues for the quarter. We continue to invest. We had investments of $9.4 million in total. If I take a step back, Q2, and put Q2 into the context of a longer time series, just to put it into perspective. Here, and we illustrate this with the LTM basis of June 2025 on both revenue growth and EBITDA. I think we can see that we continue a track of revenue growth and EBITDA growth despite this lower uptick in performance in Q2 than we had anticipated. I think by all means you can see we have a strong track record of converting investments, organic and inorganic, into revenue growth and attractive margins. I see no reason why we cannot overcome this temporary setback for Q2 and prove that again.
It might just take a little bit more time, but I think the track record of this company is there and proves it. Moving into operating cash flow and CapEx development. I'll just pre-phrase here that now comes a very dense slide, but I try to break it up into its different components. If we look at operating cash flow development here on an LTM basis, June 2025, you can see the operating cash flow generation in the dark blue bars here on the chart. We moved to $115 million down from $137 million on a full-year basis for 2024. This is again because of the effects of the top line in Q2, and therefore it's not following quite the normal seasonality effect of cash flows through the quarters, as you normally see. Normally, Q1 is our lowest quarter, and Q4 is our strongest quarter, and it increases over time.
Therefore, we also expect an uplift in cash flow generation for the year as a whole, picking up in Q4 in particular, which is the biggest quarter for this type of business. You'll note a first positive effect on cash interest expenses here in gray. We moved from $45 million to $43 million on an LTM basis. You see the first effects of lower interest rate costs. Of course, as we move out the LTM number, you will see that your old levels of interest flush out, and you achieve the new levels. It's very positive. We see that we achieved net-net $64 million in free cash flow after interest levels, which is still at a healthy level. Note when you look at the LTM number, also we did have one-off expenses, one-off payouts in Q1 and Q2. We had fees around the bond.
We had a prepayment on one earnout. We had several things that also kind of throw off the normal run rate of the business. Moving to CapEx development, we've shown this slide a number of times. If you focus here on the blue and the dark blue, which is really our CapEx investments for maintenance and expansion development, we are at LTM basis, respectively $35 million and $8 million. We have communicated that we would end the, we expect this year to invest somewhere between &40 million- $45 million as CapEx. We are tracking closer to the $40 million than to the $45 million, but otherwise unchanged views on what this business needs of investments going forward. That was operating cash flow and CapEx development. I now move to net leverage ratio, interest coverage ratio. We largely maintained our ratios into Q2.
You see a marginal worsening from 2024 to 2.5x net leverage ratio to 2.5x now on an LTM basis. This is the operating cash flows that affect. On the other hand, you saw a marginal improvement in the interest rate coverage here from 3.3%- 3.4% if you compare again LTM with where we ended the year. This is the positive effect of the interest savings on or the reduced levels of interest rates. Deleverage remains a key point for us, a key focus point for us. Albeit Q2 probably slowed a little bit us in that regard. Our mid-term target is to achieve a net leverage ratio of 2x
I just want to make sure that this remains a focus for us and it's clearly understood. That kind of concludes my financial part as it concerns Q2 in particular. Now I'll move to guidance because this is probably also what you've been interested in hearing about. You will have noted we changed our guidance Thursday. We lowered and revised our guidance for the full year. Let me try to break down the components and contributing factors here. Here, we use a bridge to bridge from our midpoint guidance, our initial midpoint guidance of €165 million to the midpoint guidance of our revised guidance given the 15th of August of €132 million and just try to break it down in components. I think in the big picture, unification challenges, technical challenges under unification accounts for two thirds. This is illustrated by the gray area.
There was a direct impact during Q2, but also in July and just the first days of trading in August, which altogether, when you calculate it through, is €19 million. We then also had additional costs during that time on two fronts. One, of course, we used all our powers and we had mitigation costs to address the technical issues. As Remco was explaining, we also had some problems with the roll balancers, which meant we had extra hosting costs. Altogether, that is €4 million. That all relates as one-time effects. Another thing that has happened is that our expectations, and we can see where we are on the dollar to euro translation, is outside the range that we expected when we gave the initial guidance.
We communicated when we gave our initial guidance that we would expect changes in euro to dollar to have an effect of ± 2 percentage points on the total impact. We see now that we are outside that guidance and we have revised the planning assumptions to be on a constant currency basis using a U.S. dollar to euro rate of 0.855. Altogether, that would have an impact when you look at it for the full year of €9 million. That's basically the components. Two thirds from the unification challenges, one third from translation, pure currency translation effects for the full year. We acknowledge that Q2 fell short of expectations, certainly also our expectations. Trading levels are now at the same level as stabilized, same level as a comparable period last year.
For the marketplace activities specifically, I'm talking about here, and that makes us confident in a sustained pickup for the remainder of the year. When you pull all that together in numbers, I show here our revised guidance compared to the guidance we gave initially on May 28. Our new guidance is $485 million- $515 million. That represents a 10% downward revision. At the same time, we're narrowing our guidance by $5 million. Respectively, on adjusted EBITDA, we are guiding $125 million- $140 million, representing a 20% revision downwards of our initial guidance and also there narrowing the guidance by $5 million. We have good confidence in our revised guidance given the pickup we're seeing in trading and experiencing here in August. We have roughly four and a half months to go. I remind you that Q4 is the biggest quarter for this type of business.
We have some things to do here for the remainder of the year, but we are fully comfortable with this guidance. Important for those that look at currency translation effects, we are giving this guidance on a constant effects basis. We continue to be exposed, we are quite exposed on the top line towards U.S. dollars with roughly 90%, actually 89% exposure to the U.S. dollar. Please keep that in mind as we see for the remainder of the year how the U.S. dollar translates into euros. That basically concludes the update on the financial performance background for our revised guidance. I hope also it provides some insights on what exactly has impacted the business KPIs for this period. With that, I think I hand back to Ingo.
Thank you so much, Christian, for the color on financials.
You enjoyed it.
I enjoyed it.
That's good.
I enjoyed it.
That was good.
You can stay right up here because now it's time for our first Q&A session. How we're going to do this here in the room, if you have a question, please raise your hand so a team member of us can bring a microphone over. I think until we have the first microphone transferred to the first question, I'll start with some questions from the online chat. To sum it up, we still see uncertainty in the market on our release from Thursday night, which is very understandable. The platform unification has caused a share price decline of roughly 22%. Yesterday, we increased by 6%. This morning, I saw we're + 10%. We're regaining confidence in the market.
To sum this up and a couple of questions here, Christian and Remco, we highlighted that the effects from the unification will be around $34 million on revenues and $19 million on EBITDA. The question we're having here is now, is that really it for the rest of the year? Is everything in Q2 and Q3 now in these numbers? Are there going to be any further unification steps in the further course of the year?
Okay. Maybe I can take the first part and then you can take the unification part. Yes, the total effects of what our expectation for the year is included in this guidance. We have obviously tried to model various scenarios for the year, and the $34 million on top and $19 million is included in those numbers.
On the integration steps, Remco, maybe you?
Yeah, I had it in my presentation. Integration is done. That's important. We finished it by the end of July. Of course, the effects are, let's say, a little bit going into August, but we see that we are back on track. Platform runs, numbers coming up, and onboarding continue, or let's say starting again. In that sense, in time for the very important Q4 and also end of Q3 because from basically September onwards, advertising is the strongest period, or it's the strongest period for advertising. Happy that we finished this before, even though it took longer.
Thank you. Great. Maybe the first question from the room here, from the audience.
Oh, Vincent Diodon from Pareto Securities First off, I would like to ask you a question based on the updated guidance. You showcased a bridge here explaining the financial impact of the issues that you faced here in Q2 and Q3. How are you taking into account the soft market that you're also experiencing and mentioning here in the presentation?
Yeah, should I?
Yeah.
I discussed, or let's say I showed that the market is a bit softer, at least if you look at the statistics in the market. As also mentioned, we don't see that as a reason to not perform or not to perform. Yes, a bit of softness is there, but we have a strong product with ID-Less solutions. We have a strong position in the channels that grow most because that was average numbers for the whole advertising market. The channels where we are, let's say, outlying that part. It's not a market where we say, "Hey, wow, we have great tailwinds from everything is growing like hell." That's not the case. It's also not a reason to see any issues there.
Unless, of course, and that's something we don't have a crystal ball, but if something really explodes, I don't know, China goes into Taiwan or something, that's not in. For the current market conditions, which we expect to go around the same, we see the possibility further to grow and to not be bothered too much about a bit softer market.
Can I just add on the $34 million top line effect? We consider in this scenario where there is a tail end of how fast those customers that were affected will grow and scale, which is included in the $34 million. There's a tail end effect in that number, not material, but that's how we try to gauge it for a full year consideration.
Okay, thank you. That's very clear. Now we're in August 25. It's been just shy of a year since you announced the acquisition of Jun Group. How would you evaluate the performance so far of Jun Group as being part of Verve?
I would almost say Mishel, come on stage, but he's not wired up with a microphone. He can say maybe a few words later on it. Now, we're super happy with the acquisition. There were a few aspects, let's say a few risk aspects and a lot of positive things. First of all, adding a lot of salespeople, the agency relations, a lot of tools also, also the zero-party product that Mishel is going to show later. It was a very lot of accretion, a lot of synergy between the cases also bringing together. That's what we see now. The biggest risks that we saw then at that time were, let's say there were two, actually. One, the company was not growing. They were milked by the previous owner.
We had to change the mindset, but that was not so difficult actually to get them to growth, to invest, to now also hire sellers, etc. The other one was really the team, team with a tenure of between eight and 14 years, at least the core 20-30 people have such a long tenure. I've learned one thing, if you're a tight, let's say, team and you start changing things, there's of course a risk that the team doesn't like it. That was where we were really slow in integrating, doing it super careful, first getting people to get used to each other, then slowly integrating the salespeople that Verve already had into the Jun sales team. Now also with, let's say, the changes in our top management and our executive team, really also, yeah, basically getting product, technology, commerce split over the different departments.
We took really time for it a year, but it went extremely well. Later Mishel can say maybe a few words about it.
Maybe in between until the next person receives the microphone, one question from the chat. Maybe for Remco here. Remco, can you explain the future of the unified platform? How will we benefit in detail from the unification, and what are the really important drivers which we're now capable to benefit from?
More complex answer or easier answer?
I think we cannot comment on financials, of course.
I think it's about the, how to say it, the, yeah, what's the effect of a unified platform? If I have two platforms, I have two teams that need to work on features. I have two platforms where I have to run an infrastructure for, etc., etc. Two platforms are just much more inefficient than one. Platform integration has the disadvantage that you have those two teams working on each of the platforms. Plus, you need to get extra manpower to prepare the integration, which means that on one of the platforms, you basically build the features that are on the other platform and that are the condition for migrating things. You need to work on the loads of the platform, all those kinds of things. You have a phase during the integration where you actually need more manpower than you need for the two platforms.
With a unified platform, you free up that manpower. That's one advantage. With freeing up the manpower, you really can also work on the platform. You can build new features. You can really build new things on the platform. You can really focus on the market and go with that. What was not so visible, it was of course bring the platforms together. It's extra manpower, it's extra cost. Also in that period, you're very internally focused. If you have one platform, you can really externally focus. You can really build the things that your customers want that bring you further in the market. There is this double effect. It's more efficient, but it's also bringing us in a much higher speed for innovation for going forward.
Great, thank you. The next question from the audience, please, Rasmus.
Yes, hi. Hi, Rasmus here with Kepler Cheuvreux. I had two questions. Can I first start with the guidance? You had -4% organic growth in Q2. It seems that you expect organic growth to return in the fourth quarter, but what about the third quarter? Is it negative, positive during the quarter, latter half, or how should we think about that in your guidance?
Sure. I can also, yeah.
I think, obviously Q3 will be a mix. We have said that July is impacted, first day of trading for this part of the business, right? Just to remind that the 4% organic growth is for the combined group. If we talk about the market activities, some impact in July, first days of trading, and we are stable now, but not quite on the growth path. Q3 will be a mix. Q4, we expect for sure the strong organic growth. That's what I can say at this point. Obviously, we run a number of scenarios, both on top line and bottom line, and we try to pull that into what we think is a good reference point and reference range.
No, I think you described it perfectly well.
The second question, obviously, it's probably a question for your more technical people, but have you achieved what you wanted to achieve, efficiency, savings with the integration? Is it fully up and running on what you wanted to do?
Yes and no. We achieved what we wanted to, it's a platform unification, but it took longer. As I've shown on the slide, also there's still smaller things to go, so that's not ready yet. You're, of course, never ready. The yes is we have achieved it. The no is you're never ready with these kinds of things. We will work on further efficiency. I mean, AI, we get that later, will bring extra efficiency in it, will bring in the platform, but also in the team. The question is, what's the company of the future? How many people do you need to run a company, those things? There are always efficiency gains to come. You're never ready in that aspect. A company that has a standstill is going to die. You need to innovate, you need to grow, you need to get AI implemented to further get more efficient.
I hope that answers your question, but it's a difficult one to answer because it's a process and not a final thing.
Thank you. Yes, please, Ellis, continue.
Yes, good morning. Ellis Aklan with First Berlin. Thanks for the detailed presentation so far this morning. Two questions from my side. I'd like to touch on your current M&A posture. It sounds like you're becoming a bit more open-minded about that at this moment. Also, just to tie in what you were talking about with unification. Now that you have everything on a clean single platform, if you go and make another big deal, are you going to have to also integrate that? Is that going to restart in some form or fashion?
The company is a process. Thanks for the question, first of all, Alice. Yes, we have an M&A slide in there. We had also some discussion if we put it in because on one hand, the market is consolidating, so there are opportunities in the market. We need to grow and we need to grow faster to get bigger. In that sense, there is a rational, but as I said already, M&A needs to be super careful. We don't need it. We are a strong organic growth company. In that sense, M&A is a luxury, which basically solves things where the organic way of doing it would take much longer. The point that you make, of course, if you buy another platform, another technical platform, then you have to do another technical integration.
Before we buy a company that has another technical platform with a lot of integration stuff, we'll think 4x , 5x, 6x about it before we do something like that. I see M&A more in really acquire kind of directions instead of really big technical platforms and things.
Okay, that's maybe an add-on from my side. Let's not forget that this unification represented around 90% of our revenues. If we integrate further on the CTV side now, this is still ahead of us, but that's a way smaller revenue share, and therefore the effects will hardly be seen because you might have a hiccup of 5%- 8% or something again based on these revenues that this platform stands for. In this case, that was the majority of the whole company.
Yeah, that's the point. I mean, it's a good point you're making. This migration was huge. I mean, we have done a lot of migrations before that were not noticed or let's say that where you didn't see the things, but this one was really big and in that sense very exceptional.
Okay, thank you. That was very helpful. Just a short one. You mentioned that you didn't lose any of your existing customers during the unification process. That was mainly more about onboarding new customers. How did that impact? Did you lose any potential new customers because of that that needed to conduct their business immediately and can't wait or couldn't wait for you to finish your unifications?
Yeah, thanks for the question. First of all, it's not that we didn't lose customers. We had a 98% retention rate. That means that we lost some customers. We're not many, but you always lose customers as a company. I think that's important to just put it right. We indeed had to delay, how to say it, onboarding of customers. As far as I'm aware, they're all patient and waiting and we didn't lose any of that. Everybody in this market is not dependent on one intermediary. You have several, so it's not a problem to wait. The only problem that we would have had, and I don't think we have it, is that if you run towards Q4, people then will further postpone an integration because nobody wants to do any technical work in Q4 because that's your main quarter.
That's where you make most revenues and you don't want to have any risk of disrupting your Q4 with technical work. In that sense, I'm happy that we now are in the process of integrating well before Q4.
Okay, guys, thank you very much.
Great, then maybe in between, one question from the chat. It's on euro-dollar translation effects. How are we going to handle this in the future? Are we going to do hedging, and what are our strategies regarding this topic?
Okay. Let's just put it in perspective. Obviously, we are now in an environment and have been for, let's say, the last year where the euro-to-dollar translation has had much more swings if you look back in time. It's just a different market environment and more volatile. In such a case, yes, you consider, should you do hedging? We are exposed on our top line in particular, with roughly 89% specifically of our revenues, which are registered in U.S. dollars. We also have quite a bit of our cost base, but not to that degree in U.S. dollars. There's a natural hedge from our cost base that, how do you say, mitigates or offsets, is probably the right word, some of the impact on the top line. We are evaluating the situation of hedging. To do hedging properly, there's a number of elements.
You need to have a quite advanced treasury setup to manage it on an ongoing basis. Otherwise, you go into somewhat, which is just very large bulk positions and you're essentially also taking a bit of a bet on where the market is going. We are looking at what is a good level of hedging for our type of business, the complexity and the exposure we have, but I also want to make sure that we have the right hedging strategy and also setup to handle it on an ongoing basis. It's not quite as easy as people might think it is and just doing one forward contract.
Yeah, thank you. Maybe if I may add to that, I think what's very important is currency fluctuations are a very normal thing if you have to shift currencies between euro and dollar, and they will stay. It's not important that you take out all of these effects, but that you manage market expectations properly. That's why we decided, for the further course of the year and the years ahead, to publish our currency ratio, our exchange ratio that we calculated our guidance with. In this case, I think it was 0.855.
Correct.
Was it right?
That's wrong.
Remember?
For that point, it means all investors know what we calculated with. If dollars go up or dollars go down, you can calculate yourself what the effect will be. Sometimes we're going to have a plus, then we don't want the bonus on our side, but sometimes we're going to lose something. We'll see how far we're going to develop into hedging in the future. I think that is the most important thing that you, as our investors and analysts, know what to expect. Please, Jörg.
Phillip Rai, Warburg. You had a nice chart regarding the softness of the market, some price declines of CPMs in your presentation. In the second quarter, we also saw that the walled gardens had very decent pricing. Meta, for example, at +9%. How do you look at this? First of all, a bit of comment on the difference in pricing regarding the open internet and the walled gardens. Do you expect these positive pricing of the walled gardens to be more a leading indicator of what is to come for you, more or less some gap closing, or what's your view on that one?
Yeah, thanks for the question, Philips. The walled gardens is, of course, something special. We showed before the % of margin that the technology takes and what goes to the publisher. The advantages, and that's what the walled garden has, if you are the publisher yourself and you have the technology, you have much more spread and possibilities in your margin. To take now Google, for example, Google is under pressure of the Department of Justice to be split up. They're especially under pressure because they're unfairly playing the open market. If you would be Google, what are you doing? You say, I'm putting everything on my own supply, which is basically YouTube. You see a lot of tendency on Google that they take market, let's say, revenues out of the open market. Some of our peers are really hit with that strongly.
They take market, let's say, volumes out of the open market and don't put them on, or let's say, don't put them on external inventory anymore, but they put them on their own YouTube. YouTube has become super full with ads if you look at it. Something similar we see also with other walled gardens where they put more and more on their own properties. There's also quite some criticism on that from advertisers because how really well can you measure? I mean, they also deliver their own measuring capabilities. At the moment, I would say short term, we see a bit of winning on the walled garden side. Talking to agencies and others, not everybody is that happy with it. Let's see how that develops in the next quarters. Especially Google is extremely strong in bringing things in-house at the moment.
That's also what you see in the guidances, or let's say in the numbers, the Q2 reports. Most walled gardens have done pretty well, the big ones. If you look at the smaller ones, SNAP, etc., it's the other way around.
Very helpful. One additional question on, you mentioned your targets regarding expansion of your sales teams. How quickly do you expect these to ramp? How much time for profitability, or should we expect some margin dilution until these guys hit the ground?
Yeah, it will be a continuous, thanks for the question again, Philip. It will be a continuous process, of course. You cannot hire immediately 50 people at one go. We had to prepare for that, which means we have sales enabling people, so people that can really build up sales teams. We hired those first. We increased our, how to say it, internal hiring capacity. On the HR team, those things had to get in place. Now there's a lot of interviewing happening. I just had my meeting yesterday with Mishel, and we signed up six people to start next two, three weeks. The good thing in the U.S. is if you hire people, they are very fast available. Sometimes they have still a commission that they're waiting for, so it might delay a bit if they have done a very strong Q3.
They will not start before, how to say it, the second month into the next quarter because they don't want to lose their commission on that. It's a pretty fast process. Normally, you get hired somebody within two or three weeks, sometimes with a little bit delay. We had to first really enable the organization to hire those people that if they come in, that we can train them, that we can really, yeah, make them productive. We had to split up the teams a bit. The team structure was changed a bit. Mishel is nodding in the back, so I haven't said anything wrong so far. It's a process. It's a hefty target, of course, to get from 35 to 150. That's a heavy target. We're working on it. If it's 140, I would not be unhappy. If it's 155, also not.
What we cannot fully estimate, we hire sellers and not all of them will be good. The question is, is 20% that you have to replace afterwards again, or is it 40%? With respect to the cost, we internally calculate that the seller really starts earning back his own salary earliest after six months, latest after 12 months. Otherwise, it has to be replaced. Yes, it will put a bit of pressure, and it did already in Q2, put a bit of pressure on our EBITDA. That's, yeah, how do you say, investment in the future, and you don't capitalize those costs. It's just cost. I think it's super important for us as a company to do that and to further grow.
Thank you very much.
Could I just add to that, that we have the interviewer, the main interviewer in the room today, of course, in Mishel. Of course, from a CFO perspective, what is really important is to make sure that we really set high standards for the people we take on board and not in a rush to get people on board. I don't think I've experienced yet in my career a more tough interviewer than Mishel and with higher standards. From a CFO perspective, I think it's a process, as you say. It's also a balance. We move fast forward, but we also need to secure we have the right quality of people because mistakes on people often are costly. That's the balance we need to keep.
Great. Thank you, Christian. Next question is over here.
Yeah, thank you. Nicola from ABG Sundal Collier . Just one question on your big cash position. You're sitting on something like €160 million in cash after raising some €32 million in equity. That's a big, big war chest. What do you want to do with that war chest?
First of all, let me just remind that we do have certain deferred payments to handle, which is Jun Group in particular. I believe it's $22.5 million in the first installment and then €25 million or dollars. I have to say, I can't remember if it's euros or dollars on the last payment. This is coming up within the next 12 months. Already there, you have some payouts. We have payouts for the earnout for Dataseat. Some of that makes sure that we can pay that. We want to continue the investments that we are doing organically. We, of course, also want to have a good cushion as we move through this somewhat more volatile period than we wanted this year to be.
At the same time, have the opportunity to, if we come across something that is really the right fit, small addition, then we have the flexibility to act on that per your comments. It also explains a little bit the order of priority seen from the CFO perspective in the way that I described it here.
Yeah, wonderful. Thanks for that, color. Just on the guidance downgrade, maybe just some additional nuance on that one. When would you say that you saw the necessity to downgrade the guidance? Did you see any risk of that sometime mid to late June, or did you see it perhaps during Q3 now? How would you characterize that?
I think where we really saw is in the later weeks of July and our expectations for June. We could see that we probably have some effects in July, but it was not recovering at the pace we expected to. Also, initially in August. Now, of course, I'm talking about one part of the company, which is specifically marketplace activities. You have many other things to factor in. When you start calculating on that type of thing, you can see, hey, that we would be looking at a changed view. I think also, as I illustrated, when you then factor in the conversion effect, which is a major part of it as well, and combine that, you see the perspective of the whole thing.
Yeah, very helpful. Thank you.
Maybe also from our side here, it was really the case that Christian approached us in the morning of Thursday, that now all hope is gone to recover that in the remainder of the year. Not all hope. Not all hope. There is you're right.
The probability is.
The probability is very small that we can recover that. Of course, the whole process was started and we decided, let's bring forward the numbers as well so the market knows what's happening. That was really the point when the final decision was made.
Those things are not easy decisions. The problem here is, of course, that first of all, the year gets shorter. The longer something like the unification issue lasts, the shorter the time you have to recover is. On the other hand, the more revenue you lose in the time in between. It wasn't really, yeah, a hefty discussion we had internally. We will do everything to get everything out of this company to go and to really grow. At a certain point, you have then to say and to realize that it's difficult to really make up. What we also had to take into account, we could also have not raised the guidance, for example. You get everybody here in this room and online and many other people are all speculating, hey, the numbers are so weak in Q2. We think they have to take the guidance down.
That is worse. It was really a decision that we took, had to take. Yeah, it hurts. Here, we want to be, yeah, as open as we can be and to also give the backgrounds and the reasons. I'm happy, really happy that it's not like, hey, it's a terrible market and that's why we're losing and we don't know when this is recovering. There is, let's say, majority-wise, it's really an internal thing. That's over.
As an added dimension, you always look at the trading every day and try to evaluate what is the signal of that trading for the remainder of the year, which is very, very difficult. We followed it closely and came to that conclusion when we pulled everything together.
Great. Thank you. Looking at the time, telling me zero seconds left, meaning thank you all for participating in the Q&A. Also, thanks everybody online who participated in this. Really good questions here. Now, let's come again to the agenda. It looks like half the fun is over for the day. Part one is behind us. We're now approaching the lunch break. After that, we're going to have a real interesting tech session with the keynote I'm really looking forward to and two executives following in on the hottest topic of the industry. We'll all meet back here at 12:45 P.M. Also for you online, German pünktlichkeit. Yes. Enjoy your lunch, please. I'm looking forward to see you back in about an hour. Bye.
Good food, and I'm now energized for part two of our Capital Markets Day of the Capital Markets Day of Verve Group 2025. We're now kicking it off with the part I was really waiting for: the keynote speech by Eric Seufert, a well-known expert to the digital advertising industry. Eric is a media strategist, a quantitative marketer, an author, and investor who spent his career working for transformative consumer technology and media companies, including Skype and Rovio, the developer of Angry Birds. He's also the Founder of Agamemnon, a startup that built a mobile marketing analytics platform, which was acquired in 2017. He then launched Heracles Capital, an early-stage venture capital fund focused on the mobile ecosystem. Eric, it's really a pleasure having you here today with us, and thank you so much for traveling all the way from the U.S. Everybody, a big applause, please, for Eric Seufert.
Hey, good afternoon, everyone. Thank you for that very generous introduction, and thank you for having me. It's fantastic to be in Stockholm at this time of year. Thank you for organizing this in the summer, not the winter. I'm really happy to be here, and I'm really happy to be discussing one of my favorite topics, which is what I believe is the total transformation of digital advertising through AI enablement. I've spoken about this on the blog, in my podcasts, and came up with this catch-all description of how I see this transformation unfolding, which I call commerce at the limit. This is the agenda for the presentation. I'll start with an overview of what I believe is at stake with this.
I'll move into where we are, just a status check on the application of AI to digital advertising, and then I'll talk through where I believe we're headed with this. A little bit about me before we dive into the content. I run a blog called Mobile Dev Memo. It's a trade blog dedicated to what I call the digital economy, so digital advertising, distribution, and merchandising. I am the author of Theseus, which is an open-source Python library for marketing cohort analysis. I run Heracles Capital, which is an early-stage venture fund devoted to the mobile economy. Prior to doing those things, I worked for a number of companies that you might recognize, especially those of you that are Europe-based. I started my career at Skype as an analyst. I was Head of Marketing at a company called Wooga, which makes casual games. It was acquired by Playtika.
I was VP of User Acquisition and Network Engagement at Rovio, which makes the Angry Birds franchise. My startup was acquired by a company called Network, which was later acquired by a crypto platform. What's at stake? If you follow any companies in the digital advertising space, which by definition you all do, you're familiar with the application of AI to digital advertising and how it's unlocking a lot of value now, right? I say that now because Mark Zuckerberg says it's happening now. I think a lot of people look at the application of AI to digital ads as this far-off opportunity, right? Something that maybe will materialize, and it's a risk, right? If you look at the way the largest platforms are applying it, it's actually happening right now, and it's delivering real value right now.
This is a quote from Mark Zuckerberg on Stratechery, which, fun fact, I was the previous guest before Mark on Stratechery. He said, and I'll quote it, I'll read it, and it's kind of awkward, but I think it's a powerful quote. In general, we're going to get to a point where you're a business, you come to us, you tell us what your objective is, you connect to your bank account, you don't need any creative, you don't need any targeting demographic, you don't need any measurement except to be able to read the results that we spit out. I think that's going to be huge. I think it is a redefinition of the category of advertising, right? When people heard that, there was, I think, this kind of like this sort of pulse of fear, right?
That sounds like it's just going to displace a lot of people, right? That's going to maybe put a lot of agencies out of business, or that's going to lead to layoffs in marketing teams. I think there's a flip, and some of that might be true, right? I think there's a flip side to this, which is that it's going to enable essentially anybody to become an advertiser. It's going to enable anybody to participate in the advertising economy. When you think about the largest platforms, you think about them onboarding kind of any sort of mom-and-pop shop as an advertiser. If you think about the entire ecosystem of advertising platforms, everyone can benefit from that, right? Everyone can benefit in unique and idiosyncratic ways that might not be available to the largest platforms.
I think for any company operating in the space, there is the sort of looming threat of dislocation just because any sort of new big wave of innovation unleashes that. When I think about the application of AI to advertising, I'm sort of overwhelmed by all the positive. I do believe that those benefits, they don't explicitly and specifically accrue to the largest platforms. They will accrue to anybody that finds a way to apply them to their business. Digital advertising is just a massive space, and it's a great space given the nature of the ecosystem. It's a great space for this to be one of the entry points for the application of AI commercially, right? Dentsu expects the digital advertising market to reach $678 billion in 2025. That's almost 70% of total ad spend.
Digital ad spend will grow by 8%, which is faster than the broader advertising spend growth rate of 5%. A meaningful proportion of this growth will be delivered by channels that don't adhere to traditional measurement methodologies, right? What do I mean by that? I've got this catchphrase, you may have heard it, everything is an ad network, right?
The idea behind everything is an ad network is after Apple's app tracking transparency policy change, you saw this mass proliferation of new ad channels emerge because when you revoked that sort of identity spline that was provided by the mobile ad ID that Apple had called the IDFA, when you revoked that, then any channel, any property with sort of unique proprietary first-party data could spin up an ads business and use that first-party data for targeting in a way that wasn't really viable prior to the revocation of the IDFA because Facebook just owned everything. They were the central store of data for the entire internet.
When these platforms that previously wouldn't have dared to compete with Facebook had the opportunity to monetize this first-party data set because there was nothing else to really target against, they decided, hey, I've got this asset, let me just monetize it by launching an ad network. Everything became an ad network. A lot of the everything is an ad network phenomenon is a result of channels that are not sort of like click-based building ad systems. Or even if they are click-based, they don't integrate well with other sets of data, right? Thinking about CTV, right? CTV is not measured by traditional measurement methodologies. You can't do last click. You have to kind of focus on probabilistic attribution, multi-touch attribution. Think about a lot of retail media. You can't measure that with traditional measurement methodologies.
As AI, these methodologies in AI emerge to handle that, you unlock a lot more of that value that SMBs, for instance, couldn't access because they couldn't measure it. Measurement has become more challenging as the ecosystem grows more complex. Platform policy restrictions, I just mentioned ATT, but there are others. Apple is traditionally more aggressive in the space, as you're all I'm sure aware. Google was going to deprecate cookies. Now they're not. We'll see where that lands. We've got the everything is an ad network phenomenon, which again, a lot of these new channels, they're not attributable with clicks, right? CTV, retail media. Then you've got end-to-end automation, right? When you have end-to-end automation, that kind of crowds out a lot of the traditional measurement that was just a function of transparency, right? End-to-end measurement, the platform's making all the decisions on your behalf.
You, A, don't have the ability to impact those decisions, right, with, for instance, targeting, right, or placements. B, they gain by sort of not making a lot of that decision even available to you ex post, right? There's just a level of opacity that precludes kind of traditional, you know, sort of legacy style attribution. Ultimately, AI will touch all aspects of advertising value. It's not just measurement. It's not just, you know, targeting. Like a lot of people probably think when you say AI and digital advertising, the first place your mind might go is to like all of these tools that the largest platforms have introduced. I'll sort of list them later. They're all kind of named along the same convention. The reality is it's going to touch every piece of the value chain, right? You've got measurement and attribution, like I just talked about.
You've also got product personalization and monetization, right? If you sort of remove the levers of targeting, of even creative production from the advertiser, then all they're really left with, they get this sort of like amorphous blob of traffic to the property. All they're really left with to drive performance is what they can send to the ad platform to indicate whether that was good or bad traffic, right? This is kind of this emerging topic called signal engineering, right? I'm playing around with the onboarding session in an app or the registration flow in some sort of e-comm website. I'm trying to figure out ways to surface intent, surface a signal of intent from that user. Traditionally, the way this has been done is just looking at what's instrumented on the page or what's instrumented in the app and just sending all of that to the ad platform.
Now, if that's the only lever that I have, I have to actually be much more proactive in determining what is a signal that this person will become a high-value user. That's signal engineering. An example of this, which I found really interesting, was this: I was on this podcast recently called The Marketing Operators. We were talking about signal engineering. After that podcast, one of the people on the podcast is the CMO of a large e-comm brand. He wrote on Twitter that as a result of that conversation, he got the gears working in his head about how they could implement signal engineering into their—it's like a beauty shop, right? How could they integrate signal engineering into the registration process for this e-comm brand? They built a CAPTCHA that was a math problem that did not gate access to the site as a normal CAPTCHA would.
It was not really trying to determine whether this person was a bot or not. It just was creating an obstacle for that person to get into the shopping experience. If a person is going to go through the trouble of answering this math problem, they're probably pretty interested in your products. They probably have a higher propensity to buy than someone who will not answer the math problem or who would not answer the math problem if presented with the math problem. What they did with that is they just sent that back. The person completed the math problem. They sent that back to the platform. The platform knew, okay, this is a high-value user. Obviously, this would be tested. There could be no correlation with propensity to buy in doing that.
If there is, and you have to experiment with these things and design these things specifically and proactively to determine intent, if that works, that's a really high-value signal that the ad platform just got that it would not have otherwise had, that it can then use to optimize the traffic that's being sent to the website, right? That is this kind of like an emerging field of signal engineering. It is really powerful. That is left to advertisers currently. There are any number of ways that ad channels could facilitate that, right? You can imagine that there are advantages for smaller ad channels relative to the largest, right? I might not trust one of the largest ad channels to do that in a way that benefits me. A smaller channel, I might trust.
To my point from earlier, there are idiosyncratic ways for AI, call it AI, or call it just any of these tactics to be deployed that generate value for platforms outside of the biggest. There is campaign optimization, which I think is kind of like the canonical way that AI is applied to digital advertising. Those are things like, you know, Pmax and Google Advantage+ . There is a whole list of them. Those applications make obvious sense. The real value of doing that is, A, you tend to maximize advertiser spend when you take control of all optimization levers, right? The goal of these systems is not necessarily to sort of maximize ROAS for the advertiser, it's actually to maximize spend while adhering to some performance target, right? B, you get to do, you make decisions faster.
You really do sort of optimize faster than like a human being could, which again maximizes spend. I think one of the other things is that these systems, they can be improved incrementally over time. The really fascinating thing about the digital economy, but specifically digital advertising, is that since advertisers approach, you know, smart advertisers, capable advertisers approach digital advertising with kind of like a ROAS mindset. If you unlock 5%, right, 6% incremental ROAS, that all gets reinvested, right? We were talking last night at dinner about the opportunities with Apple being forced to open up the iOS ecosystem to allow Link Out, but also alternative in-app payments. The reality is, there is value in just sidestepping the commission. There is a lot more value in just bringing a sort of modern approach to the app storefront, right?
If you think about the app store as a storefront, and you compare it to Shopify, for instance, it looks like it comes out of the Stone Age. I mean, it's not fit for purpose. It's not a modern storefront, you know, relative to what Shopify can do with like personalized pricing, dynamic bundling, even just A/B testing based on different, you know, sort of demographic features. The app store can't do any of that. You just promote commerce by having these sort of modern tools there. When you unlock 5%, 6% like incremental ROAS, then that just gets reinvested back into marketing. Who wins from that? It's the ad channels. I mean, it's the advertisers too, because they're driving higher top-line revenue, but it's the ad channels that are going to absorb more of that.
Creative production really is kind of the tip of the spear, I think in a lot of ways. I'll get into this a little bit later, but creative production, I think, is just more visible as an application of AI. People sort of instantly invoke that when they think about applying AI to digital advertising. The way creative production is done now through AI tools, I think it's mostly a cost-saving exercise. It's not really generating a lot of value. The way that you generate value with creative production, to my mind, is actually doing more rigorous hypothesis testing and figuring out what kind of concepts work the best and using historical data to inform that. That's actually one aspect of digital advertising that I think is pretty primitive. I think AI will unlock a lot of value there.
AI will influence the totality of the customer relationship that any advertiser has. It's obviously customer acquisition, and that's everything I just talked about. That's where, how, and with what messaging customers are reached. Also, customer onboarding. I talked about that. Signal engineering. Getting them into the product, making sure that they're primed to be successful with your product. Better instructions, more personalization. Linking what they see in the product and their first product experience with what they saw in the ad. These things can be facilitated with AI to create a more productive customer journey, which probably leads to higher rates of monetization. There's customer activation, the specific moments that are chosen to surface monetization opportunities to customers. Do I show an offer early in the customer's first session, or do I wait? Do I let them interact with the product more before I try to monetize them?
Do I show an ad at some point in a game, at any specific point? What's the best moment for showing the ad? Paul's going to talk a little bit later about a lot of their on-device stuff that surfaces these signals to unlock more surgical monetization. All that kind of stuff, when we talk about AI, we're really talking about high-dimensional machine learning, just machine learning that can make sense, can parse relationships out of high-dimensional data. All of that is an opportunity. Customer retention and merchandising, the content and opportunities that are exposed to the user keep them engaged with the product, right? It's really just kind of like elongating the retention curve. Who can win? I've kind of touched on this already. The largest platforms have natural data and R&D advantages in developing and deploying AI tools, but they face concrete limitations, right? Inventory.
Advertisers want to reach audiences on CTV, retail media, and elsewhere. It's not just social media, right? Diversification. A lot of SMBs are not really sensitive to sort of concentration. They could be fully concentrated. Some won't, just for the sake that they don't want to introduce that risk into their business. What we saw in the Google DOJ suit was that there were customers that would allow for lower price floors on non-Google channels for no other reason than they didn't want to be totally reliant on Google, right? You might say that that's not a rational choice. They should accept the highest bid from whoever's providing it. For them, it was because they felt that if they were totally concentrated just on Google, they faced kind of business risks. They wouldn't otherwise if they had sort of more demand. On-site use cases, right?
The influence of large platforms doesn't extend into the product experience. I talked about that before. Maybe there are trust issues, right? There are other channels that could have a better sort of ability to influence what happens in the product experience where that publisher wouldn't trust the same kind of intervention from a larger platform. That could be for a number of reasons, right? The platform might not have their best interests at heart. It could be that they don't want to expose that data to the platform because it'll be used to benefit their direct competitors. There's any number of reasons for that, right? That are totally rational. I want to go through where we are at the moment. How is AI being used currently? There are two use cases that are mostly trapped inside the walled gardens currently. The first is creative production, right?
If you think about using generative AI tools to create ad creative, that's mostly a big platform phenomenon right now. There are a number of SaaS tools that have emerged to try to handle this. The problem with that use case is it's a loss leader. It doesn't really make any money. There's no value. There's no sort of ability to monetize that if you're just a third-party SaaS tool, right? The way you monetize that is you pair it with the data that you have to drive better outcomes if you're a large platform, right? Meta announced in Q2 that in its most recent earnings that it has 2 million advertisers using its GenAI tools. That's a lot, right? There's just a natural advantage there because they have 10 million. To get to scale up to 2 million is, you know, they're able to do that fairly quickly.
The second is campaign optimization. These are tools like Advantage+, Performance Max, which is Google Pmax, Smart Plus from TikTok, Performance Plus for Amazon, and Pinterest has one. These allow advertisers to optimize campaigns automatically. It's an obvious use case. There's a tension there because I think it's fairly obvious. It's fairly self-evident that the motivation for this is to maximize spend. It's not really to provide the best possible outcome for the advertiser. It's to maximize spend against some performance threshold. There are other opportunities beyond that for other channels to embrace AI. These are totally up for grabs as far as I'm concerned, right? The first is targeting. Advantage+ manages targeting on Facebook. What about everything off of Facebook, right? Particularly the channels that I referenced earlier, retail media and CTV where Facebook has no presence.
Google has a large presence through YouTube, but the largest channels are not necessarily operating there. There's no reason why they ever will, right? Targeting consumers outside of the walled gardens is a really important use case for these tools. Second is measurement. Advertisers want unbiased measurement that accommodates their entire channel portfolio, right? It's great if a large platform has measurement baked in, but if I'm running on other channels that singular focus on that platform doesn't help me. I need measurement that accommodates everything. Then creative, right? Creative production, again, is mostly viewed as a cost reduction exercise, but other use cases can be tailored to channel-specific contexts, right? CTV, retail media, these things need the sort of tools to accommodate that exhaustive testing that the large platforms have built natively. There is an obvious opportunity there for someone to solve that problem.
None of this is superficial and none of this is temporary. This is a very significant tectonic shift. Direct response marketing teams are embracing the overhaul of the media buying process with open arms, right? If you go back to that quote that I opened with, there was opposition to that idea by brands. A point that I made was that, well, you're about five years too late, right? This is not some new thing. This has been happening for a number of years. Brands and agencies, and if you talk to any sort of SMB digital advertiser, not to be a small company, but just a company that operates in a lean way, their hair is on fire for this. They want more of it. They want it now. They want more AI enablement because that helps them do more with less.
If you talk about the sort of entrepreneurs or just very small businesses that are excluded from the advertising economy now, they probably don't know that they don't know this, but they'll be very pleased to have the opportunity to advertise their businesses as a result of these tools, right? I think the opposition is a little bit misguided. I understand it because, again, it's a dislocation, but this is just expansionary, right? This will bring more people into the advertising economy. While this starts at the top of the funnel, it will only be enhanced by bottom of the funnel improvements. I'm talking about better personalization to improve monetization. That's kind of on-site stuff. Then better signal construction to improve optimization and targeting. Targeting needs to apply to every channel outside of the walled gardens too. A lot of the growth is happening outside of the walled gardens.
Ultimately, every ad channel can benefit from these capabilities, and advertisers will expect every channel to offer them. I think that's an important point. It's not just that this adds value to the sort of customer experience that the advertiser has. It's that advertisers will choose the ad platforms they work with based upon these capabilities. They are going to expect it from everyone. AI-enabled advertising is already delivering considerable commercial value. I won't read the quote again, but in most recent earnings, Mark Zuckerberg attributed all the growth, disproportionately said the growth was a function of these investments in AI. These were not big eye-popping numbers. He went through each of the three big AI initiatives. You've got Lattice, Gem, and Andromeda. He talked about a 4% increase on click-through rate or a 5% increase to ROAS, right? These are not 30%, 50% boosts.
As a nature of digital advertising, those improvements compound over time, and they result in more ad spend coming back to the platform. A few thoughts about where we're headed. As a logistical note, I think we're doing a Q&A on all three of the sessions at the end. There is no Q&A after this one, but we'll have one big Q&A. I think AI ultimately delivers what I've called commerce at the limit. I did a monologue podcast about this a few weeks ago. I define commerce at the limit as the fulfillment of complete optimization across every component of the digital advertising process, such that commercial performance attains its theoretical maximum, right? If you think about that, basically every ad is, call it, utterly optimized, right?
In terms of targeting, in terms of intent creative pairing, in terms of the placement, in terms of the sequence when the person's seeing the ad, right? If you imagine that, and then imagine that along one axis, right? Along the other axis, you imagine that every company that could possibly advertise a product is doing so because there's just an ease of onboarding that invites them into this ecosystem. This is a massively creative, right? This is just an enlargement, a considerable enlargement of this economy. My concept of commerce at the limit is that every link in this value chain operates at maximum possible efficiency, right? Just going back to this diagram from before, like measurement and attribution, product personalization and monetization, campaign optimization, and creative production, all of these things sort of attain the maximum potential efficiency.
There's just, you know, just thinking through that, I mean, there's significant opportunity for expansion. Some of the downstream consequences of Commerce at the Limit. The first, I think the most obvious, is increased participation. Commerce at the Limit expands the advertising economy to every business that could possibly benefit from it, right? The sort of light bulb moment for this was I was in Houston, where I'm from, and I was flying out to give a talk the next day, and I needed a haircut. I went on Google, and it's just almost impossible to find a barber. If you go to a barber, you probably just walk by it and you're like, I'll try that out. It's near my house. If you're in a new place, it's really difficult to find a barber and to understand if it's any good, right? I went on Google and I found one. It was terrible. It was horrible. I was thinking this barbershop could be advertising.
Imagine, and that's inventory, or that's an audience that's probably not being effectively monetized at this moment because it's not having local businesses advertise to it, right? You've probably got this uncompetitive audience that's very cheap to reach that is essentially not being monetized. You've got an entire group of not just barbershops, but anything, nail salons, businesses that aren't participating in this economy. You get more commerce from these companies because they're able to reach customers that they otherwise never would have been able to. On the channel side, the ad channel side, you're actually monetizing this group of people with ads that are targeted to them that otherwise just would have seen remnant inventory. On both sides, you're expanding that economy, which leads to probably price inflation, right? As all ads perform to their maximum potential and all advertisers compete on equal footing, bids become the auction focus.
If everyone's ad is maximally constructed for, optimally constructed for conversion, then how do you beat someone in the auction, right? The auction is just an expected value ordinal ranking. How do you beat someone? You have a higher bid, right? I need to monetize my product better. That forces me to monetize my product better, to have a higher bid to push into the auction. The ad channel just makes more money. Personalization as an imperative. With advertising mostly automated, advertisers will shift development and analytics resources to in-product personalization. These improvements will surface better quality signals that can be passed to their channel partners for optimization, right? This kind of puts the imperative.
If I historically just outsourced a lot of that optimization to my ad channel partners and I just had a product, I had my product that I advertised, a singular product experience, everyone had this same experience, right? I just outsourced a lot of the personalization to the ad platforms based on who got targeted or maybe like which of these dozen creatives got shown. Now that the platform is managing all of that at the very top of the funnel, the only way for me to optimize sort of my own business is to make sure that everyone gets the optimal product experience, right? That pushes a lot of that personalization effort onto me, whereas before the ad platform sort of owned that. Finished exactly on time. Thanks very much. I'll look forward to the Q&A after.
Wow, Eric, thank you so much. Really inspirational. Thank you for sharing your thoughts and your insight on the industry. As you see, we're really starting to go down the rabbit hole now. I would really like you to follow the white rabbit together with me. When I joined Verve and had my first days at the Berlin office, it was so inspirational to me, just like listening to Eric now, to see the people working there, having discussions in the meeting rooms. The whiteboards were full. They even started doing paintings on the glass doors because everything was full and they were just so creative. It's really, that is the reason why I wanted to work for Verve and why I think we have a lot ahead of us.
From all the big tech brains that you will meet at Verve Group, I'm now bringing you two further really good tech brains to present to you the next two presentations. One is Mishel Alon, the CB O of Verve Group, and afterwards, Paul Hayton, our C T O of Dataseat. I can tell you this is going to be intense. Buckle up, ladies and gentlemen. Mishel.
All right. Thank you so much, Ingo, and thank you, Eric. Good afternoon, everyone. It's great to be here again. Some of you probably remember I was here exactly a year ago, the same month that Jun Group was acquired by Verve. It's been an amazing year. I know that one of the questions that was asked earlier was about how's it going? I want to answer that before we get to the topic of the presentation. It's all in a way related. It's been exactly a year since we were acquired. If I had to just describe how this year has gone, I would say that it's been a masterclass of how a well-designed and executed integration should work. The way I think about it is the team that we brought in from the Jun Group side, which Remco was referring to, is very tight.
We haven't lost a single person from that strong team that we've had over the years. As Remco mentioned earlier, we had a strong team of individuals, part of the Senior Management team that have been with the company for 8 to 12 to 14 years. We haven't lost a single person. That's something that I don't take for granted and I'm really proud of. It's been an amazing journey with Verve and we're just getting started. Let's talk about ID-less advertising. Obviously, it's a big topic. We've been talking about it already today around the modes that we have in the industry and how Verve is well positioned to win in this market that includes ID-less and also has ID-less environments which are evolving. I'm going to cover it from multiple directions. I'm going to talk about it from a demand perspective and also from a supply perspective.
When I think about the demand perspective, the way to describe it is when you work with advertisers, when you work with the actual agency folks that are looking to work with partners on running their campaigns, that's the demand side. The supply side is where do you place those ads? Where do you find the relevant audiences for your advertising campaigns? We are in a very unique position where we can work with both. We work with both and we're able to show results and outcomes for our clients. We at Verve are very focused on outcomes. My main goal in terms of outcomes in this presentation is if I ask you what is ID-less advertising after this session, everyone will know how to answer that question. I'm going to use a few real-life examples that hopefully will make it easier to understand. All right.
I'm going to start with a couple of questions. By show of hands, who here is concerned about privacy when it comes to advertising? All right, the majority of the room. The second question is, by show of hands, who in this room feels like they have a good understanding of how targeting actually works in the advertising space? Okay, smaller proportion. My goal is that by the end of this presentation, everyone will be able to lift their hands. Let's start from the demand side of the business. Demand is working with advertisers. Verve has made a few acquisitions, a couple of notable acquisitions in the demand space. One of them, obviously, I just mentioned, Jun Group a year ago. The second one is Dataseat. Dataseat, I'm going to talk more about the products that we have, the solutions that we have for our clients.
Both of those companies that Verve acquired were super focused on a privacy-first type of approach to reaching audiences. We're doing it from taking different directions, different angles to it. At the core of what we do, we want to be in a position where we are able to perform for our clients while also respecting privacy regulations, privacy laws, privacy best practices from the Apples and Googles of the world. If I double click on Verve Dataseat for a second, in 2019, Apple introduced to the market a concept called ITP. That concept was around how do you limit the usage of cookies in Safari browsers on mobile.
Because the founders of Dataseat, Paul and David, learned Apple for many, many years, they knew that that was just a first indication of what Apple was going to do in a broader way when it comes to mobile in-app advertising and that it would not be limited just to the browser environments on the devices. Over the years, we've spent a lot of time looking into what Apple was doing, learning their practices. I'm going to get into that more later in the presentation when we think about the TAM and the future, which are going to just get bigger and bigger as time goes by.
That vision in 2021 became the reality with the introduction of ATT, where, as Remco showed earlier in the presentation, when you install a new app, the app will prompt you a message, ask you, are you willing to share information across apps, to be targeted across apps? 80% of the users say no to that. This is not stopping. GDPR, the evolving regulations in the U.S. around privacy laws, that's all going to continue. Being strong when it comes to respecting privacy regulations and best practices is going to get, at the same time, more challenging and also create a lot of opportunities for us because we are the experts in this space. How does it work? I think that it's really important to define both ID-based and ID-less approaches just so you get a full and holistic sense of how this actually works. Let's start from ID-based.
I'm going to use a simple example to explain how ID-based advertising goes. At the core of ID-based advertising, you have some value, an ID that you tie information to. When you tie that information, you basically create a profile, a behavioral profile that you can then use for targeting. Let's use a simple example. Let's say someone is a single person and they're looking to start dating a person. They install Tinder. They start swiping left and right. They schedule a date night. It all goes well. They want to watch a show on Netflix. They install Netflix. They want to order some food. They will order something from Deliveroo, let's say some Vietnamese food. It's all going great. At some point, they decide to book travel.
What happened there in the background is there are third-party partners called MMPs, mobile measurement partners, that will take all these signals that I just mentioned, what did the user install, when, etc., and they will build a profile for that user. You can see the profile that is being built for Alex in that example. You see what kind of data points will be added to that profile. Now, if there is an ID available there, then they will tie that information to the ID. They will sell it to advertisers. Advertisers will use that information to target the relevant users. In this example, you know this person booked travel via Expedia.
The next thing that you'll see is that they are now being targeted by an ad from Booking, telling them to book travel via Booking to the same place that they wanted to go to, let's say Barcelona in this example, trying to convince them to book via Booking.com. You can take it one step further and you can use the additional information that you got, for example, the type of food that that person ordered and use that also as another adjustment to the targeting in a way where the creative, going back to what Eric briefly talked about earlier, where the creative can also be adjusted to also reflect that to drive better outcomes for the advertiser. Now, you're probably asking yourselves, wait, you just heard that 80% of the users on iOS are not even going to say yes to cross-app targeting.
How do these profiles get built? The answer is that those MMPs that I mentioned, the mobile measurement partners, think about them as players in the market that are responsible for aggregating data and creating profiles around users, are taking this approach that is referred to in the industry as fingerprinting. What is fingerprinting? You basically take the data points that you have about the user and you build a profile around that. Those data points could be the IP address. Every device that gets connected to the internet has an IP address. They will take that information and they'll try to combine it with other information that they have. For example, the type of device that the user is using. Let's say they're using iPhone 16 Max, and you take those data points, combine them, and you create a fingerprint for the user.
The issue with that is that it's a gray area right now. We're going to talk more about what Apple is planning to do down the road around it. It's also not always very accurate. For example, in this room, anyone who is connected to the Wi-Fi is sharing the same IP address to the external world. If there are two people who have the same type of device, which is very common, and they click an ad and make an install, for example, that attribution might get conflicted between those two individuals. You don't know who is the user that is actually the one that installed the app in this example. It has issues and limitations. It's not always very accurate. Those are aspects that you want to keep in mind. On the other side is idealistic contextual targeting.
When it comes to contextual targeting, there are many different approaches that you can take in order to find the right context in which your target audience is. What you're seeing here is the way that we're doing it on our end, which is you basically look at the overall context where the user is coming from holistically. That includes the publisher that they're visiting, the time of day, the geo, and based on the type of advertisers that you work with, the type of campaign that you're running, you'll see different attributes, different signals from the context that will have better accuracy when it comes to performance.
For example, if I go back to the example I mentioned earlier today, earlier in this presentation, if I think about the dating app, time of day will play a bigger role in terms of giving you an indication whether it's the right audience to reach. Publishers are always an important part of the context that we use in order to target the right audiences. It could be language, it could be time of day, it could be geo, many different signals that we take from the context in which the user is at in order to make the right decisions. This approach, the idealistic approach that we're taking, we're going to talk in a minute, like who is working with us in this capacity. If you know what you're doing and you do it right, you're able to deliver great outcomes for the advertisers.
You're able to be in a position where you perform for the clients while also respecting privacy laws. There is something that is happening in the background. I mentioned regulation. Obviously, that's an important part. There is also the overall Apple approach. As I mentioned earlier, Apple in 2019 started caring more and more about privacy. Over time, that became a bigger priority for them. At some point, they decided that instead of letting those MMPs that I mentioned, the mobile measurement partners do the tracking, Apple wanted to be the source of truth when it comes to attribution. They wanted to be the ones who are giving information to the advertisers about which campaigns are effective and how the performance of their campaign is going. Obviously, it's also benefiting Apple to be the one controlling the truth and the information. There is that.
They came out with this concept called SKAdNetwork, where they are the ones who are responsible for providing that information. Let me use another simple example to describe how it works. Let's say Alex from our team has a pizza shop. Alex really wants to promote his pizza shop to audiences. He prints out flyers and starts distributing them to individuals. If you take an ID-based approach, then you want to know that Sameer, for example, received a flyer from Alex and he ended up coming to the pizza shop and made a purchase. That's ID-based advertising. However, Apple said with the introduction of SKAdNetwork, we're not going to allow that. We're still going to allow you to track how the flyers that you are giving out are working, but we're going to do it in aggregate. What Apple did is basically Alex delivers, let's say, 100 flyers per day.
He does that with different colors, different font types, different times of day. Apple will tell them how he's doing with his flyer campaigns in aggregate, but they won't tell him that individual X, Y, or Z are the ones that came into the store. They will tell him your campaign or the creative that you used in the morning was able to deliver X purchases, and the campaign that you did in the evening with pink instead of black delivered Y number of purchases. That is the approach that Apple is pushing for in the industry as a whole. As part of that, they're trying to continue going against fingerprinting and banning it. There's another thing that is happening in the background. The introduction of LLMs has resulted in less users, less traffic going to web publishers.
Because of that, there is a need, even an increasing need, for advertisers to reach audiences on mobile. Why? Mobile is one of the only places where you can reach an individual and not a bigger group. When you run a CTV campaign, you reach a household. When you run a digital ad of home campaign, you reach even a broader audience. The thing is, if you want to reach an individual, you need to rely on mobile. That is creating an environment where mobile is going to continue growing in its size. That context is even more important when you think about the overall opportunity that we have here in terms of TAM. In the U.S., over 50% of the users use iOS, whereas in Europe, it's a lower percent, closer to 30%.
Another thing that is interesting about the U.S. market when it comes to iOS versus Android is that many of the marketers, many of the advertisers are actually trying to reach iOS users. Reaching iOS users is becoming more and more difficult for those who are relying on ID-based advertising for the reason I just mentioned. If you think about it from that perspective, and if indeed Apple goes and bans fingerprinting, that's going to at least 10x our opportunity in terms of promoting our ID-less solutions in the market. Why do we think that that can happen? Apple did something very interesting in iOS 26, which will become available in September. They banned fingerprinting when it comes to the Safari browser. They're not doing it in-app yet. As we just discussed earlier, when they do something on Safari, in many cases, it's an indicator for what they're going to do in the overall in-app environment.
If indeed they ban fingerprinting in-app, that's going to continue accelerating our growth when it comes to our ID-less solutions from a demand perspective. I was alluding to this slide earlier. Who do we work with when it comes to these types of solutions? We work with the largest, biggest names in the world. Just think about it for a second. Apple created SKAdNetwork. They are the ones behind all that is happening in terms of privacy laws and privacy best practices on iOS. They work with us to promote their products and apps. Apple is the one creating it, and they trust us that we are capable in delivering results for them. That's huge. It doesn't stop with Apple. I don't think I need to go into all the names on this slide. It just shows you the strength of the products and capabilities we have in this market.
In terms of the outcomes that our advertisers are looking for, they're pretty broad. It depends on the advertiser. It could be as simple as impressions and reach, and it could be as more involved in terms of funnel, in terms of installs, subscriptions, post-install events. You want to know what happens with the user after the install happens, and we are able to deliver those outcomes for our clients at scale. Let's take a look at one example, OTTO. Remco Westermann mentioned them earlier today, third biggest retailer in Germany, which is very focused on privacy-based approaches to advertising. We've been working with them for almost a year, and the return on ad spend has been significant for them. Eric mentioned return on ad spend earlier.
In the end of the day, it's all about how much do I invest in my campaigns and how much do I get back from an increase in sales or whatever KPI the client has. This product is performing for them, and when it performs for them, they continue working with us because it's a no-brainer for them to do that. With OTTO specifically, we're also giving them visibility into post-view events. The user watched the ad, installed an app. What happens next? Are they spending on the app? Are they making purchases? If you are operating in an idealistic environment and you don't have the capabilities that we have and that we have built over the years, you don't have visibility into those. We are uniquely positioned to deliver those types of insights for our clients. That's another reason why they want to continue working with us.
Another approach that we take when it comes to idealistic or privacy-first advertising is our zero-party data. You heard earlier today the concept first-party data mentioned. First-party is information that you get from the publisher itself. Zero-party is you take it one step further and you get it directly from the user. The user is volunteering information that you can then use for targeting. In this example, you're asking a user if they're interested in buying new sneakers. If they want to answer the question, they'll answer it. If they don't, they don't have to. Based on the response they give you, you are able to adjust the ad experience for them. We've been doing this at scale on the Jun Group side for many years. Now it's offered across the board as part of our brand and agency offering in the U.S. and in other markets.
Here I also have another case study that we did for Mars. It was for a pet product. We asked the user, do you have a pet? Based on the ones that said yes, we delivered the relevant ad. The return on ad spend was significant. In terms of outcomes, this client was able to see a significant increment in sales, which in return drove them to continue working with us and spending more with us. I covered the demand side. Let's spend a minute on the supply side. When ATT came out, what happened was the DSPs, the demand side platforms, had less visibility into the users that they were getting requests from. Why? There was no data. They could not use data to target. They were blind to what was happening. As a result, you saw a big decrease in effective CPMs.
That decrease impacted both the demand side and also the publishers. What's interesting is that publishers still had a lot of access to data. The key was how do you use that data and how do you create an environment where you still drive value for the consumer and also for the advertiser. There was a huge opportunity there. We decided when we saw that opportunity to get into that space in a way that was unique. That way was ATOM. ATOM gives you the ability to identify your target audiences on the device without any sensitive information leaving the device. This on-device targeting capability that we have gives us the ability to identify cohorts of users. When we identify those cohorts of users, we're able to pass it back in a privacy-compliant way. Advertisers can use that to target the right audiences.
We are, again, uniquely positioned with this solution. It gives us the ability not only to look at things from a demand perspective, but also from a supply perspective. Let's show you a quick video of what is ATOM.
In a mobile-first world where engagement drives revenue, advertisers face a paradox of reaching the right users at scale without compromising privacy. Embedded in hundreds of millions of devices, ATOM unlocks mobile inventory at unprecedented scale. Through integration with the Verve SDK, ATOM connects advertisers to one of the largest mobile programmatic ecosystems in real time. ATOM is built on a privacy-first architecture. No personal data, no tracking IDs, no server-side profiling. Everything stays on-device. On mobile, ATOM breaks through ID limitations, generating real-time on-device cohort intelligence, even in ID-restricted environments. Using on-device behavioral analysis, ATOM produces predictive, privacy-safe cohorts with up to 90% accuracy in inferring user attributes. Advertisers can choose from over 200 behavioral and contextual cohorts or build their own. For publishers, ATOM powers real-time insights that reshape how inventory is valued, optimized, and activated. The future of mobile ad targeting is private, intelligent, and scalable.
ATOM, make every impression count.
You saw a few examples of audiences and cohorts that were mentioned there. We will take ATOM, will take the information that we are getting directly from the device without the information leaving the device to calculate the cohort. For example, someone who is very active in the morning will be profiled as such. Obviously, we take a lot of data points. We have over 180 signals that we take into account to calculate those cohorts. There is a lot that is happening in the background. What's basically happening is that you have AI that is spread around many devices. Paul is going to talk more about that in his presentation. How does this all come together and what are we going to do next? AI is really at the core of what we do in terms of what I described from a demand perspective, supply perspective.
We are currently working on the next generation products where we bring all that together. Our goal is to continue delivering better outcomes for our clients. When we provide those better outcomes, they tend to spend more with us. They trust us. They rely on our ability to generate the outcomes that they're looking for. It also creates an opportunity for us to become more efficient. AI has a lot of efficiency promise in it. Paul is going to spend a lot more time talking about it. Our goal is to continue optimizing our campaigns, deliver better outcomes, and to do it in a more efficient way. Just to wrap things up, when it comes to ID-based or ID-less targeting, our approach is to use what we have access to. When we have access to IDs, we will use them as long as it's compliant with the privacy laws.
When we don't have access to IDs, we will use our secret sauce, which gives us the ability to succeed in the market. That puts us in a very unique position overall in the market. The opportunity, as I mentioned, is going to continue growing. It's going to continue growing in terms of our ability to deliver outcomes for advertisers. It's going to continue growing in terms of our ability to strengthen the relationships with publishers by driving better results for them and better revenue. It's also going to give us the ability to be in the forefront of what's happening in the industry because, again, our capabilities in this space are very unique. We're going to continue investing, building on them. We're just scratching the surface. We're going to continue evolving our products and offerings. I'm going to let Paul talk more about that. Thank you.
Good afternoon, everybody. Artificial intelligence has been part of ad tech for over 10 years. It's grown steadily in its importance and its power with each year that's gone by. Over the next 30 minutes, we're going to have a very quick tour through the AI that powers Verve 's advertising platform. We're going to look at the AI that allows us to be more productive and more efficient. We're going to look at the AI innovations that make Verve different, that make us unique, that stand us apart from the rest of the industry. My name is Paul Hayton . I joined Verve three years ago. I am the Co-Founder and CTO of Dataseat, our contextual demand side platform. I began my career in academia quite a while ago now.
I was a university researcher for five years at the Neural Networks Research Group in Oxford University, researching applications of neural networks. I've been part of ad tech since 2011. It's great to be here to talk about a subject that I enjoy so much, which is artificial intelligence. Before we get too much into our tour of AI, a little bit of some clarifications, shall we say. AI is about taking data and training a model that captures the underlying patterns, generalizes what that data is about, learns the truth that that data came from. From that, it can make predictions, classifications, and make decisions. In the past, AI has usually been about taking a particular set of data for a particular task.
You take, for instance, images of road signs, and you train an AI to recognize speed limits, and you put them on the dashboard of your car as you're driving along. One particular task-based AI. Recently, though, we've started seeing large language models come along in the last year or so. They're trained on language. They're trained to provide an appropriate response to a prompt or a question. This gives us just a whole new set of opportunities of ways in which we can make use of it to drive better outcomes and efficiency. We do need to watch out, though, because AI is not perfect. There are lots of well-known examples. See, an AI is great if you take some data and you give it another question or input that was similar to the ones it's been trained on.
It'll do a great job of providing a very accurate prediction if it's inside of that training data that it had. If you ask it to do something that it's never seen before, it will often fall flat on its face. If you ask ChatGPT for a picture of a person writing with their left hand, you will get a picture of a person writing with their right hand. If you ask for a picture of someone wearing a jacket inside out, you will just get a picture of someone wearing a jacket. You've asked for something that it's just never seen before, and it just falls over, and it does something completely unpredictable. You just have to be aware of what AI can do and what it can't do. It often doesn't capture the truth that's underneath it. It just learns from what it's seen.
Onwards, we go into our three chapters of our tour. First of all, we're going to look at ad targeting. What's at the heart of Verve 's advertising platform? The need for data, how we match advertisers and consumers, and how we optimize for KPIs. We're going to look at efficiency, the tools we use, intelligent AIs, and how we and our strategic Google cooperation, how that is going to enable us to produce better AI models going forward. Lastly, differentiation, the things that make us different, the things that make us stand out, what makes us unique, and what gives us that competitive edge: ATOM, SDK integrations, and Helix. Going into the first chapter, targeting within Verve , first off, we need to recognize that we need data. In fact, there are three things you need to build an AI.
You need a lot of data, you need a lot of computing power, and you need a team of data scientists who understand how to get the most out of those things. We have all three at Verve . We have a huge amount of data, trillions of pieces of data from our exchange, millions of bits of data from our SDK coming from those devices that have our SDK in, things like, in some cases, from ATOM, gender and age, and other SDK signals, how people use their devices. Then we've got ID graphs, links between different devices that a user owns and uses. We can also take in external data from other sources like Experian. All that data can come together in what looks very simple because it's just a triangle on the screen of just AI optimization.
It's really an opportunity to use all that data and optimize it to provide the best ad served to the best target group, the best ad to the right target group. That's our goal. Now, what's inside that triangle of AI optimization is a lot more complex than it looks. We're going to dive in and have a look and see inside. See, our net revenue that we get out is a function of a whole bunch of things: the ad requests that we send out, what we get back in, the bid rate, which is how often our demand partners are bidding in the auctions that we are supplying, how often they are winning, because if they don't win, that means nothing to them or to us, the revenue share that we get, the CPM, which is the prices that they're bidding.
The higher they bid, the better for us, the better for our publishers, and the render rate, whether or not those adverts actually show. All of those aspects of this process can be optimized. All of those have opportunities for us to optimize them with AI models. We're going to pick up on just one of them and look at the impact it has. There are lots of initiatives on the screen there: demand shaping, bid floor pricing, Helix. We're going to touch on some of them, but we haven't got time to go into all of them in detail. Just picking up one and the impact it has for us, demand shaping is a question of taking all of the available ad requests that we could have access to and choosing which ones to send to our demand partners. It's like fishing.
It's like putting the right bait on the right hook. If you do it right, you get lots of bids back at high CPM, high prices. You make lots of money. If you do it wrong, you get no bids back. We have access to an absolutely enormous, enormous amount of available advert opportunities. This is over a 31-day period. There were 1.251 trillion opportunities to show advert requests. Of those, we're picking out the 2%- 3% where there's a very, very high chance that our demand partners will want to show an advert where everything matches: the right geography, the right type of publishers, the right demographics. If that's done well, it's a huge efficiency gain. If we get it wrong, it costs us a lot of money. This is demand shaping at its best.
We routinely measure the impact of this versus not doing it. It has a huge boost in revenue. We can do better. We can make this more granular. This is a constantly evolving, constantly improving piece of technology because our demand partners change what they do. They're constantly evolving too. They're doing more subtle bidding on our inventory. We have to make sure our demand shaping matches that to supply them with the requests that they want to bid on. To do that, we're looking to use more complex models, including, for instance, deep neural networks. They are super complex. They are harder to train, but they provide us with the most complex means of generalizing underlying patterns that you can have.
We're also looking at using a technique called reinforcement learning that will allow us to have a model that's trained and have that model adapt as new data comes in. This is not static. It actually continues to learn in real time as new data arrives. If our demand partners change what they're doing, the model will change what it supplies. This allows us to have more complex models that are trained less frequently because training them is the costly part of it. We've looked at one example of an AI technique within Verve of quite a few that there are in place. These AI techniques allow us to select the adverts that we supply to our demand partners. They allow us to optimize the number of adverts that we show, and they allow us to maximize the revenue and the margin that we take as a business.
Now we have to move on to the second chapter of our tour, which is about efficiency. We're going to look at three aspects: team efficiency, an example of one intelligent UI that we've built, and our strategic Google collaboration. First of all, team efficiency. As well as building AI tools, we also are consumers of AI. We have the opportunity to make use of AI to make ourselves more productive and more efficient. As I'm sure many of you have tried playing around with AI tools, we are the same. We've started activating certain AI tools, making them available for our teams in an evaluation mode, shall we say, because not all AIs are created equal. There have been academic studies that have shown that some AI tools actually decrease productivity. They're more gimmick than they are tool.
What we're doing is evaluating how well these work, how much efficiency improvement we get, where we become more productive, and where they're just a distraction. We have GitHub Copilot, which is an AI for our developers, which aims to lessen the amount of typing they have to do because it predicts what they're starting to code and offers completions, offers highlighting mistakes that when you mistype. It's very easy when you're coding just to miss one character, which changes what your software does completely. AI can pick up on that for you and highlight it. Code Rabbit is an interesting AI tool that does code reviews for you. Now, developers like solving problems. They don't particularly like typing. They like solving problems, and they like coding. They don't tend to like reviewing other people's code so much.
It's a necessary part of the job, but it's not people's favorite piece of their working day. Code Rabbit does a basic review for you. It picks up on occasions of bad practice or imperfect structures, and it can speed up that review process by highlighting as a first pass things that could be improved. We're not going to get rid of the human code reviewer, but we can make their life a little bit easier. We also have Gemini, Google Gemini, added to the whole of our organization so that we, rather than searching through documents, can ask Gemini to find answers for us. The way to do this is not just to enable all these tools and hope. We are doing it in a structured way.
We have an overview of which tools are active and making sure we validate that they are beneficial, that they are actually giving us productivity improvements, and giving guidance for best practice to our team. We need to have a proper rollout of the tools that we agree. The ones we find do give us productivity improvements. We also need to be careful that we're not disclosing information by using AI. We need to make sure that these AIs are effectively closed. The data we put into them doesn't go and get used for further language model training. That way, we get the most out of them, ensuring that we benefit and don't get hindered. Another way in which we can be efficient is to take advantage of the new large language model style AIs and build an Agentic AI user interface.
As an example, this is a media brief or a media plan. These are used by our users to create what's called a deal. A deal is a segment of advertising inventory chosen for a particular advertiser campaign. It might be set up targeting a particular geography with a particular set of advertising ad formats on a set of publishers targeting a particular type of person. Typically, a media brief will be a document which will list a whole set of objectives. It'll give some dates, as you can see here, an audience that you're trying to reach, some, in this case, various deals. We want to do some display. We want to do some video.
The traditional way of doing this, we have got a user interface that we've always had, a user interface that does this, is to go in and set these up with lots of controls and selectors and date boxes. It involves quite a lot of time and effort. Understanding a document is what a large language model does brilliantly. What we've got under development is an Agentic AI, an intelligence UI. This is our next-generation deal portal. We are going to demonstrate it. What we're going to show is that we can upload that media plan. The large language model will process it and summarize it, because summarizing is a huge strength of these models, pick out the key features, and produce the deals. Let's have a look. This is the UI. We'll watch it go. We select the upload a file button.
We're going to select that file, that media plan, and upload it. The large language model processes or passes through that document, picking out the key features. It spits out, that's maybe not the right word, but it produces the deals that it believes should go along with that media brief. Let's go back to the top. We'll have a little look at what it found. It passed the media plan. From the summary, you can see it's picked out the fact that this media brief is to take Verve supply and send it to the trade desk. It's picked out the dates. It's figured out that it needs eight deals: four display deals, four video deals. Those are listed below as queries that it used from the deals that are available. It's produced some forecasts and suggested price points.
Having reviewed all those, we can either decide that we want to refine it by adding some more instructions to say, change it in this manner, or we can, if we're happy, just say, let's create the deals. When we create the deals, we go to our sort of more standard look. We can see all the deals ready to go. If we're happy, we just hit save. Those are then being trafficked from Verve supply to the trade desk in a matter of seconds, whereas previously this would have taken, at best, minutes. We can use the power of large language models to give us an intelligent user interface, one that makes life easy, that doesn't require huge amounts of training, and allows people to get things set up quickly. The third aspect of efficiency is our Google Cloud partnership.
Having moved onto Google Cloud, we have lots of opportunities. We're going to look at three of them. The first is we can upscale or upskill our data science machine learning operations. We have access to better tools. We can take advantage of Vertex AI. We can look at and use state-of-the-art notebook environments for our data scientists. This allows them to efficiently prototype, test, and put into production a new AI model. If you can do that, you can try things faster. You can get through the things that don't work and find the things that do work. You produce AI at scale in an efficient manner. We're also, as I mentioned briefly, looking at reinforcement learning so that we can take a trained model and adapt it or let it adapt itself as it sees new data coming in. It isn't just trained.
It becomes something that's being trained on an ongoing basis. It's training throughout its time. This means we can use a more complex model. It gets trained less often, and it adapts to the data it sees in real time. The third aspect is to start approaching deep neural networks. These are the most complex architectures of neural networks. We have one deep neural network in production that I'm aware of in Verve, which is within Dataseat. It does recommendations, but it doesn't have to do that in real time. We're looking to see if we can use deep neural networks at scale. It is difficult to get them right because the complexity can be anything from simple to enormous. That does take specialist knowledge. The revenue impact that this could have is to be seen. This is something that will take us. We're beginning the journey now.
This is something that will bear fruit. We've looked at AI tools that allow us to be more productive and more efficient, our Google Cloud partnership that lets our data scientists go from an idea to a prototype much faster with state-of-the-art machine learning operations. We've looked at our agentic AI user interface, an intelligent user interface that saves time and allows humans to do the bits that they do best, not just typing numbers into boxes. We'll move on to the last chapter of our tour, which is what makes us different, what makes Verve unique amongst the rest of the industry. We're going to look at ATOM. We're going to look at Dataseat. We're going to look at Helix. Privacy, we've mentioned it quite a lot today. Various people have mentioned it already. Mishel's mentioned ATOM.
I want to dive into some of the just highlight some of the AI parts of what ATOM does. Privacy is super important. We've always believed in it as being important. It does mean that you're saying to users, do you want to share your data? If you don't, that's okay. If users don't share their data, that means you have no data to train an AI. That means you need to find other ways to optimize, other ways to train models. Privacy protects users' data. That means we have to do different things with AIs in response. The answer, as you've heard lots about, or one of the answers that you've heard about, is ATOM. What ATOM does is it looks at all the signals. The thing that is amazing about ATOM, just before I dive into what it does, is it's an AI that's on users' devices.
It's inside their iPhones or their Android devices. It's running there. It's there in millions of devices, and it's running all the time. It's listening out for signals, device content, what's going on their device, the apps that they're using, how well or how that user engages with their app, how much attention they provide to different parts of the app, and whether they engage in adverts. All of these features, these hundred signals, are going into the AI that's on that device. It's being trained. It's out there in our SDK. It's taken a long time to get there. It's now on its upward trajectory of growth. That AI can provide us with these cohorts, 200 different categorized, classified cohorts. That AI is listening to these signals.
When it gets to a certain point, it will say, okay, I have a high probability now that this user is currently at home, or this user is interested in video games, or this user is of a high-income demographic. That data doesn't breach someone's privacy. It's not identifying the person, saying where they live and who they are. You can't link it to other data. It's invaluable for advert targeting. It's invaluable for showing the right adverts to the right groups of users. This will be like a drug to our demand partners because once you get addicted, you can't get off it. This will be something that will drive up prices because it's going to be super valuable for targeting. As an example of one of those, I have a teenage daughter. She uses her phone in a very different way to me.
Her fingers tap away at it at a far higher speed than I can ever hope to imagine. Looking at the way people pinch, tap, zoom on their devices are the kind of features and signals that allow ATOM to classify into one demographic, in this case, 18- 24. Those classified cohorts get sent to our demand partners. Our demand partners can then appropriately choose the advertising campaign that is right for that user, and we get better advertising whilst upholding people's privacy. ATOM has been scaling up. You've heard about it for a long time because it's been in development and slowly rolling out. It's now really on that upward trajectory. On iOS, there are 700 apps now with our SDK, with ATOM running, producing ad requests with ATOM cohorts, 1.7 billion ad requests per day from 700 apps. Android was released much more recently, only in June.
It's currently at 40 apps sending ATOM cohorts, 3.3 million ad requests per day. There's a lot more to come. As this gets bigger in scale, it becomes more important, and it drives up prices for advertising. We've looked at ATOM, and one of the customers of ATOM is dear to my heart, Dataseat, because Dataseat is a contextual bidder. Dataseat relies on contextual features, things like the publisher app that a user is in, how long they've been using the app, the city that they are currently residing in. Now we can also look at the ATOM cohorts as an extra feature. Those features together are what goes into the Dataseat AI models that allow our models to pick out the optimum bid price or the value of showing an advert to a given user for a given campaign.
As the scale gets bigger, the value of this feature, the cohorts, get bigger as well. It's a virtuous circle because the advertising gets better and better. As well as providing data, we've also talked a bit about audiences and deals through our deal portal. When we're looking at providing an audience as a different sort of aspect of what we're doing here, if we're looking at providing an audience and making that available for our demand partners or for particular advertisers to use, we now have this split between data that's got IDs in it and data that is ID-less, personally identifiable information and non-personally identifiable information. In iOS, it's at about 75% doesn't have IDs, no personally identifiable information. 25% does. On Android, it's kind of the other way around. 80% has IDs. 20% doesn't have IDs. The challenge is you've got this split.
What we really want is to take the best of both worlds. We need something that gives us a unified view of all of this data, the ID-based and the ID-less. We want to take all of it and put it into one single audience store. That way, we get everything. We can monetize all of it. We can provide the best audience possible. That is what we call Helix. It's an audience intelligence engine covering both ID-based and ID-less. It gives scale where previous audiences would have only been based on IDs. Helix allows us to have an audience that covers ID-less as well. It covers all of our channels, CTV, in-app, and web, to give us the maximum view of an audience. I'm going to demonstrate another AI user interface that allows us to get that audience as simply as possible.
We can have a look at insights on what that audience will cover, who it will reach, where they'll be, how they interlock with each other in the categories. We can optimize that audience as we go along by making changes to it and interacting with the prompts. We can make that audience, when it's finished, available through our deal portal as Verve supply trafficked to a particular demand partner. Let's have a look. This is the user interface. It is quite small because this is a screenshot that we took on Thursday. We went through various prompts. We had settled on the prompt we're going to use, which is people like me. I want to target men who are interested in electric cars and a healthy lifestyle. It's a very, very simple prompt to use. The data that's in this tool at the moment is from the U.S.
What we get is our summary of what that audience will be. There can be 21 billion daily matches. That's not users, but that's matches, 2.4 billion from ID-less, and 18 billion from ID-based, covering mobile, CTV, and web. We can also look at how the pieces of this audience interlock together: healthy living, sports, automotive, holidays, the links between those to gauge behavior affinity. We can see the location of where this audience will be spread and the different pieces that it makes up that are part of it: healthy shoppers, summer travel, healthy living, and so on. We can look at the different ad formats that will be part of this audience, from some display through to video. We can look at the price point. We can look at what volume of inventory, what volume of our audience we'll get at any particular price value.
At $5, we'll be reaching about 50% of that available audience. It'll be the cheaper inventory. If we go higher, we'll get more video inventory at a higher price point. We've seen ATOM, we've seen Dataseat, and we've seen Helix, three parts of Verve that are quite unique that make us stand apart from everyone else. Helix addresses audiences across ID-less and ID-based. Dataseat, contextual DSP, targets in a privacy-centric advertising setting, and ATOM, providing data from our SDK integrations that is not available in a privacy-based way from any other source. I'm going to finish by looking at our vision for the future going forward. This is a diagram of the Verve technology structure. We've got publishers and advertisers on the left. We've got our marketplace, Helix, control center, curation center, and our DSP. Each of these pieces has an opportunity for AI-driven programmatic optimization.
Within our publishers, we've got ATOM, which provides us with intelligence on the edge, giving us better data coming in. Our curation center is our deals portal that you saw earlier, where we can curate packages of media very simply. In our control center, we're looking at techniques for anomaly detection, so we don't have to have humans staring at screens and metrics. Instead, we can train AI models of what's normal and let them flag up when we've gone outside of normality. Our marketplace, we dove into demand shaping, but there's plenty of other techniques that were already in place: floor pricing and dynamic margin as I mentioned t oo.
On our DSP side, we have got pacing and targeting, but we're also looking at creative design, bid shading to get the optimum prices that we're bidding, and finally, Helix, that is our audience intelligence engine to provide an audience. The idea of all of these is to use the right tool for the right job, to use an AI where we're doing what an AI does best, to take data, to generalize that data, to make classifications and predictions, and free up other humans to do what they do best, which is to make strategic decisions and to be in touch with our clients, ensuring that they get the outcomes that they're looking for. Thank you very much.
Paul, great. My head just exploded. Good. Thanks a lot. May I ask Mishel and Eric back on stage as we're running for our second Q&A session?
Oh my God.
This is going to be the tech Q&A session, so we're not going for financials here or anything. I understand it's, I think we all don't have a PhD in AI, so there's no stupid questions. Please just ask. For everyone who asks a question, there will be somebody else in the room celebrating this guy for just asking because he didn't dare to. I give you, as usual, please raise your hand, and then we'll bring you the microphone. Yes, thank you so much.
Is it on?
Yes, now it's on.
I started with a stupid question.
No.
I liked your presentations, and I just wondered a bit, giving some context in terms of the competition. You historically won market share when ID-less solutions in iOS were introduced, basically. What are you thinking? What is your closest competitor or best competitor, and how are your approaches differing between your competitors?
Let's see.
I have to go first. I think that a unique position in the market is that we're able to look at it from both demand and supply perspective. There are not many players out there that have those types of perspectives. I think that one example to highlight there is what Paul just described with the combination of Dataseat and ATOM. You're able to take information from the supply side, make it available to the demand side and DSP side in a way that makes our accuracy even higher when it comes to our ability to target people at scale. Another thing to mention there is our SDK. There are many players out there that don't have that direct access to the supply, which gives us a lot of control and visibility into the ad experiences, into how much we pay out, into our ability to protect our margins, etc.
When you think about it from those two angles, you understand the unique positioning that Verve has in the market.
Please come a bit closer to me so that our online viewers can also see you. That's a good point you just stated about the SDK integrations because that seems to be a question a lot of our online audience has. How does ATOM get access to the devices, referring to the SDK integrations, and then how many other competitors in the market are offering similar SDK integrations, and how is the real competition in this market to bring your SDK integration into an app?
In terms of who has an SDK, that number of really leading vendors in the space that have SDKs is not that large. The reason for that is it takes many years to get your SDK from an engineering perspective to a place where it's stable, capable, and also drives the revenues that publishers are looking for. When you think about it from an app developer perspective, they don't want to install more than eight to ten SDKs in their device. The reason for that is every SDK you add to your platform, you create some risk. If an SDK crashes, that can have an impact on your app. You want to be very careful about who you integrate, and you have to be very selective to only include the ones that are the best performers.
The way that it works is, as part of the SDK integration, the app developers have the ability to also add ATOM. As you saw in one of the slides that Paul presented and also I presented, we've been able to scale things up. We're looking at 1.7 billion auctions just on iOS on a daily basis. That gives us a pretty robust user base.
Maybe one question to you, Eric. As the visionary on what might happen, there is a cohort of investors, I would say, that do not dare to invest into such a tech company because they say, "Oh, I cannot really say." Not something completely new will come up and then push out all the players that are currently in the market. What would you answer to an investor stating something like that?
Yeah, that's a good question. I think there's a structural resiliency that you'd want to look at in the company itself, right? There was, I mean, ATT was like an extinction-level event, right? It's five years in the rearview mirror now. A lot of DSPs went out of business. A company that survived that has proven resiliency, right? And adaptability. Just going back to the SDK comment, because that's actually a really important component. If you have the SDK, you kind of have that direct connection that gives you access to data in a way that obviates the need for a lot of the privacy-violating stuff, right? You talk about, Paul talked about, you both talked about ATOM, but like doing that sort of edge inference is really important. There are other ways, like privacy-preserving ways, like things called federated learning, right?
You can update the model from just the edge device without leaking any personal information. That kind of capability is kind of privacy. Any sort of new intervention in the market would not disrupt that, being able to do that, right? If you look at the issue with getting SDK, and I can talk about this from the advertiser side, it's the bloat, right? To your point about the fragility, every SDK adds essentially some new opportunity for your app to fail. What a lot of advertisers do is say, "I'd never do an SDK." If you want my data, I'll send it on the back end. That's all MMP-based, right? If you have an SDK, that really is a very deep connection. I'd say that kind of footprint is really important in assessing the risk of any new privacy intervention.
I would also just add that I guess what sets us apart is we have a demand-side platform that's about privacy and contextual bidding, and a supply-side platform that's about providing the right signals for that. That's a great link together. There are other companies with SDK integrations that just provide data, please bid. There are other companies that do bidding. The two together, both with that same synergy, both with that same vision of doing ID-less advertising, that's how it works best.
Thank you.
Hello.
Yeah.
I just get some clarification. This is a little bit building on what just has been said. If we say ATOM, so ATOM, B- TOM, C- TOM, so competitors. Is it your point that we have ATOM and there are no comparables such as, call it B, C, T- Tom? Is that what you said before?
There isn't another solution like ATOM in any other SDK.
Okay. So we can tick the box? W e can tick the box here and say we are unique and we are market leading. You have said before in how many devices ATOM is active in.
Yeah.
Is there not a technique to call this installed base? That we have an installed base, basically. If we are in 200, I think you said in 2 billion devices we are active. In one of your drafts, you said ATOM is in 2 billion devices.
We are getting 1.7 billion auctions every day, requests every day. You know we have that data available in that scale. In terms of the install base, we have about 700 apps on iOS that have ATOM enabled already.
Okay, if we are in the device, in the iPhone, is there value in terms of being in more iPhones?
Yeah.
Sectors position us well for continued growth.
These are 700 apps, each of those apps has 1 million daily active users. That gives you 700 million people.
I call this in reach. We have the reach into these hundreds of apps, a hundred, and then a million, million of devices and so on. Is this not value, which is difficult for competitors to match?
Absolutely.
Because they don't have the reach.
Yeah, absolutely.
How do we systematically expand that reach? We spoke this morning about investing in the salesforce, not these 100 people you want to recruit, but how do we systematically invest in expanding the reach?
There is an effort on the supply side and on the marketplace side to increase that scale, to further increase that scale. That is an ongoing thing. We are at a point where the level, the volume that we're getting on a daily basis is significant. This is not, you know, a walk in the park type of volume when you think about it from a daily request perspective. We are constantly pushing that, so it will be an even bigger part because that increases our addressability.
Yeah.
It's always tied to the demand. In the end of the day, the side that needs to act on the information that comes from ATOM is the demand side, which uses that information to target the right audiences on the SDK.
This is understood. The question, I just want to find out how strategic we are to expand the reach. If we have the installed base, the reach, and all of this, we have a more credible source or quality of information argument on the demand side, which you're going to invest in in terms of the salesforce.
Yeah, our goal is to have ATOM enabled in as many apps as we have direct SDK reach into, and also to increase that volume on a daily basis, that it will be even higher.
Maybe to add some words to that, to the question, thanks for the question. ATOM took a long time. I mean, we have been developing it for a long time. We have been testing it for a long time. In the beginning, you have to be super careful if you put it in apps because the worst thing is that the app crashes and that the publisher doesn't want to work with you anymore. We had to scale it up slowly. We have now a decent number of apps where we're in. We have our account managers that are talking to the publishers, and they are now really making sure that it gets into, let's say, we have 65,000 apps, so we want to roll it out everywhere. ATOM, to be careful, we just also still, let's say, carefully rolling that out. Sorry, not ATOM.
On Android, I wanted to say, on Android, we carefully roll it out, but on iOS, it's proven, it works, it is at scale, and the only thing now is to get more scale. Our account managers, Sameer's here, Sameer's team is really extremely much pushing that we roll it out everywhere now and really get an installed base that is even bigger than where we're now. We are now at a scale where we can prove that it works. We get enough ad requests that we can make it really helpful and that we really get the results. Paul was talking about higher prices for the publishers, but it's even more about, let's say, more results for the advertisers. Now it's really about further scaling that. We have a competitive advantage with it. Sorry, I wanted to quickly fill in here.
Thank you, Remco, for that. Maybe adding to this, the number one drop-dead question that every tech company gets is, if what you're doing is so sophisticated, why doesn't Google do it? What would they have to go through to copy us?
There is a reluctance to let go of ID-based advertising. There are quite a lot of the companies out there who are still hanging on to either the ID segment or trying to do attribution by fingerprinting because that's all they know how to do. Eventually, there will be more of a forced shift over. Mentioned the possibility of Apple absolutely breaking fingerprinting. It's in their rules that you're not allowed to fingerprint, but people are kind of ignoring that, and they're getting away with it at the moment. If that gets enforced, there'll be this seismic shift that Eric mentioned really hitting, and that everyone will be playing catch-up at that point.
Thanks. Any further questions here?
Yes, a question for you, Eric, maybe. If you would like to sort of elaborate or give your thoughts or opinions on where ID-less targeting is currently in terms of efficiency and overall accuracy compared to the sort of still traditional ID-based targeting methods, just general thoughts. I believe it can be quite difficult for you to give any firm answers, but where are we currently? I guess it's still early days, but yeah.
Yeah, I could talk about the potential because I think if you look, I follow a lot of companies. Anyway, just to level set, I don't have any relationship with Verve . I was an advisor to Dataseat, but I don't have, I'm here just to give a talk, right? Not to promote Verve Group . I think if you look at the potential, I would use Facebook as the kind of guiding example here. What they've done with Advantage+ has essentially recovered completely from AT&T. I think one of the issues with ID-based advertising is that it kind of created this sort of false trust. First of all, there was essentially no deterministic attribution. It just didn't really exist. The presence of an ID made people a lot more comfortable with that idea.
Once you pulled back from that and advertisers didn't have access to it anymore, what you realized was there were a lot of interaction effects that I was not giving credit to. I was running this influencer campaign that I couldn't measure deterministically, and that was actually benefiting all the deterministic, all the campaigns that I was measuring last click. What I think has happened is the whole market has moved away from the idea of last click. There are very specific ways in which ID-based advertising is much more performant, but there are a lot of benefits to moving away from ID-based measurement that broadly, I would say, the advertising, the cohort of advertisers have embraced. When you embrace those things, you unlock a lot of capabilities that you would never use if you were totally tethered to ID-based advertising.
Those are unlocked with probabilistic advertising and things like that. That is a lot of the sort of AI-driven stuff. I would say certainly there are downsides to not having the ID. You can't build a behavioral profile that's really helpful in a lot of cases. When you're tethered to that, you don't do things like looking at interaction effects with CTV and or TV-based campaigns and all that kind of stuff. The rise of the everything is an ad network set of new channels that couldn't be deterministically measured anyway has awoken people to the value of that. I would say, yes, there are downsides, but I think on net, everyone's sort of better off. It has created a lot of value supporting those use cases with things like attribution and measurement.
Just to add quickly from a Verve perspective, if you consider that the slide where I was showing who is working with us and what kind of outcomes we deliver for them, there is a set of advertisers that is more conscious about respecting privacy regulations and best practices. Those are the ones that would prefer to use the ID-Less approach because they believe that that's the approach that should be used in the market. I mentioned Apple as a great example for that. We expect that that market will continue to expand because of regulation, because of what Apple is doing. At the same time, when you think about those types of clients, they will not work with you if you're not able to produce outcomes and results for them in an ID-Less advertising environment.
If you consider some of the case studies that we showed, we are able to show that performance and deliver those outcomes in a way that makes the clients happy. In the end of the day, it's all about outcomes when we consider what we can do for our clients.
Great, thanks. I think we have time for one last question. Do we have any further questions here? Did you want to? I saw your hand earlier.
Yeah. I hope it's not too stupid, but just technologically, we spoke about ATOM in apps. If I have an ATOM integrated in one app, is it automatically also available as a solution for other apps? Meaning it's basically on device, or do I need an ATOM integration for each individual app?
It's an app.
Yeah, it's each individual app.
Yeah.
Because each app is sandboxed and separate.
Yeah.
Good question. Crisp question, crisp answer. Thanks a lot again to the tech expert. I think an extra applause for them.
Thank you.
It was a real pleasure listening to you. Thank you all.
Thank you.
We're reaching the end of today's Capital Markets Day. Before I hand you over to Remco, also from my side, a big thank you for attending today's day. It's no fun if there's nobody watching. To everybody online, thanks for spending the time with us. For the famous last words, I'm handing you over to Remco again to close the session of the day. Thanks.
Thank you, Ingo. Thank you all for staying with us so long and following us. Capital Markets Day is always a good point. We do it once a year. It's a good point to really show where we are. I think today we were also able to show how complex this market is. That's also one of the challenges. We gave you a lot of information. Our aim today was presenting our Q2 performance, give you an update of what we have achieved since the last GMD. We give a commercial update, financial update. We explained our Q2 unification issue. We went into our strategy and mostly also our confidence about the future. I hope we were able to also show you our, not only show you our confidence, but also make you as investors confident about the great future that this company has. We had external and internal experts.
Thanks to Eric, thanks to Paul, thanks to Mishel that did deep dives. I hope we didn't confuse you too much because, as mentioned, it is complex, but I hope that we rather were able to educate, to clarify, and to make you understand this sector better. We continue our focus to make media better. We showed you our strategic initiatives: verticalization, multi-channel, privacy, platform unification, investment into sales, geo expansion, strengthening the core team, and M&A if it fits, mostly focus on organic growth. Why are we really building USPs? As mentioned before, earlier, this market has a lot of players that don't have clear USPs, that just are there because the market is big. If you want to thrive, if you want to really be successful in this market, you need to build USPs.
With our ID-less advertising potential, with our AI, with our strong supply base, with our strong demand, I think we have really a very strong basis to further grow this company. In summary, we reached a key milestone with our in-app platform integration, and after that, I don't want to talk too much about it anymore. We further built on our strong direct supply position in-app, and we want to scale that to other platforms. We further invest in our sales teams. We continue to invest in our differentiation. We keep growing. We further grow. The target is really to reach the $1 billion. That's our next target. After that, there will be another target. We want to do that in the next three to five years. We're looking forward to that. Also, I'm super proud about the team. They had great presentations.
We see that really this company is making big progress every year. This year's Capital Markets Day presentation was shorter than last year, even though it still was long. We would really appreciate your feedback because we only want to make also this better. I would like to thank the speakers for participating, and not only the speakers, also the team behind it, and also here in the room because they were not on stage today. We have Sameer, who's running the marketplace supply side. We have Alex running the strategy part. We have Tobias Weitzel, our Chairman of the Board, apart from the other people that you saw on stage. The biggest thanks is to investors, analysts, all the people that support us. Thank you very much. Thank you for supporting us. Thank you to go with us through ups and downs.
We had a bit of a down, but we're going up now. Yeah, super proud, super happy. Thank you all very much. That's the end of today.