Podcast: The AppsFlyer investment and the new dynamics of mobile attribution (with Olivia Kory)

On this week’s episode of the podcast, I am joined by Olivia Kory, the CMO of Haus, to discuss the recent $1BN investment in AppsFlyer by Meta, Google, Unity, and Moloco, and the shifting landscape of mobile measurement it portends. We explore the strategic motivations behind major ad platforms banding together to ensure measurement neutrality and how the industry is moving toward more sophisticated incrementality models. Among other things, we discuss:

• Why a consortium of major ad platforms decided to invest in a major mobile measurement partner at this time
• How the tension between platform-owned measurement tools and independent third-party services will influence advertiser trust moving forward
• Whether the failure of Apple’s SKAdNetwork to provide actionable insights has forced the industry back toward traditional attribution
• If the entry of private equity into the measurement space changes the long-term incentives for data neutrality and transparency
• What the rise of connected television as a performance channel reveals about the limitations of current mobile measurement frameworks
• When advertisers should prioritize long-term incrementality testing over the immediate gratification of last-click attribution data in their dashboards
• Whether the consolidation of ad tech assets under single entities creates a conflict of interest that ultimately hurts performance

Thanks to the sponsors of this week’s episode of the Mobile Dev Memo podcast:

• ⁠⁠INCRMNTAL⁠⁠⁠. True attribution measures incrementality, always on.
• Xsolla⁠. With the Xsolla Web Shop, you can create a direct storefront, cut fees down to as low as 5%, and keep players engaged with bundles, rewards, and analytics.
• ⁠Branch⁠. Branch is an AI-powered MMP, connecting every paid, owned, and organic touchpoint so growth teams can see exactly where to put their dollars to bring users in the door and keep them coming back

Interested in sponsoring the Mobile Dev Memo podcast? Contact Mobile Dev Memo advertising.

The Mobile Dev Memo podcast is available on:

• Spotify
• YouTube
• Apple Podcasts

Transcript

Eric Seufert: Hello and welcome to the Mobile Dev Memo podcast. I am your host, Eric Seufert, and I am joined today by Olivia Kory, who I think is coming back for her third appearance. Is that right?

Olivia Kory: Third time. I am honored.

ES: I appreciate you coming on because this was short notice. I have been meaning to have you back on, and the perfect opportunity presented itself when yesterday we all learned that an investment syndicate consortium comprised of Meta, Google, Unity, and Moloco have invested $1 billion into AppsFlyer. Given your attachment to the measurement space, I thought you would have a lot of fascinating insight on that news. Welcome back.

OK: Thank you. Before we started recording, I mentioned this to you, but when the news hit, I was thinking to myself that I wish I could talk to Eric about this. This is perfect timing, and I am excited to be here. You did not see this coming, which I saw you posted that you did not see this one coming. Usually, you are ten steps ahead of the industry.

ES: Full disclosure, I am an advisor at Unity. They definitely kept me away from that news. Which, good for them. Not that I would reveal anything, but I had no idea. It was news to me. Before we dive right in, for those who did not hear your previous episodes, could you just introduce yourself?

OK: Olivia Kory, I am the Chief Marketing and Strategy Officer at Haus. I usually just talk about my brand side experience when I am introducing myself, but I have a pretty deep background in the ad tech space. I have seen this industry from a few different angles. I started my career agency side at Publicis.

Then I moved over to ad tech to a programmatic DSP called TubeMogul. Their big pitch in 2013 or 2014 was independence, neutrality, and objectivity. They were going up against the likes of big tech in Google and Meta. Trade Desk ended up winning out, but they had a successful acquisition to Adobe. I saw the ad tech ecosystem from that angle. Then I moved over to the brand side and spent a few years at Netflix. Netflix is where I got my MBA in incrementality testing and really learned from the best.

Then I moved on to Quibi, which is where I got the mobile app experience, and then to Sonos, where I got the e-commerce experience. I was doing incrementality testing at Sonos and saw the impact it was having on the organization and joined my founder Zach at Haus in 2022. Haus was born from the ashes of iOS 14.5 and I built the business with him. I have been doing it now for four years, and we have a couple hundred customers. We have seen a lot and have a lot of battle scars from this movement from attribution over to incrementality. You will find me maybe a little more tired and perhaps burned out than I have been in past appearances, but transformation is not easy.

ES: Also, congrats on the promotion. That was what, three months ago or something?

OK: It was pretty recent. It was a few weeks ago, but I am excited. I figured if I am going to be giving advice to CMOs all day, I need to prove myself.

ES: Well deserved. Let us get your high-level general thoughts on that investment. What was your initial reaction? How did you interpret that news?

OK: This was fascinating, and you can imagine from where I sit, it is a novel way to think about M&A in this space. If you have to rewind a little bit, let us rewind maybe five or six years to the AppLovin and Adjust acquisition.

We experience this all the time. We talk about this in the context of Haus quite a bit. Independence, neutrality, and objectivity are our biggest assets. It is why brands work with us. It is so core to our mission and to our value proposition. We are nothing without it. There are even moments where we really put the customers at the center of everything we do. It is very core to our mission.

Think of an example like ad credits, for example, where if a customer is signing up with us, we have this thing called the partner benefits program where an ad platform will provide our brands with ad credits to test a new channel. Even in that moment, it starts to become a gray area of whether we are out beyond our skis in terms of our involvement in this partnership between the ad platforms and the brands. We really try to be very thoughtful around objectivity.

So you have this acquisition of Adjust by AppLovin, and I want to talk to you about this. I do not think anybody would say that was a success. I do not know if we will ever really know why. It could have been a clash of cultures. It could have been for reasons separate from the neutrality piece, but I often point to that as an example of what you do not want to have happen, where this thing that was prized on being independent ends up in the hands of the platform that is perhaps grading their own homework. I do not think anybody would point to that and say that was successful.

Then we have this news yesterday of AppsFlyer, which by all accounts is doing well. They feel ubiquitous to me in this industry. It feels like they have emerged as the clear winner. Attribution is still table stakes in this space. So, they are doing well. They want liquidity, and instead of selling to one entity, they sell to a consortium. Thus, it is the case that no single entity can bias their platform or their results in a way where they are putting their thumb on the scale. In that way, I feel like it is a win-win. AppsFlyer gets the liquidity, the team gets liquidity, and Meta, Google, Unity, and Moloco get to keep a good thing going in terms of this independent arbiter of truth.

We loved it. The team here was really applauding the acquisition. But perception is reality. The open question is whether brands and advertisers will perceive them as neutral with this deal. That is the open question that I have been wrestling with. How do brands feel about it? I have not really talked to them yet.

You see this all the time with things like Meta’s conversion lift product or Google’s open-source Meridian. Brands, even if it is built with the best of intentions, they just do not want to trust the platform who is grading their own homework. I think there is a big open question around how brands feel about this, but it is certainly interesting and unique in the way they have approached it. I saw your coverage and want to hear your thoughts. We also just need to talk about the why and the AppLovin of it all, and whether this was just to keep it out of AppLovin’s hands.

ES: I do not think it was to keep it out of AppLovin’s hands. My sense is that Adjust has been for sale for some time. I do not believe that AppLovin sees that much strategic value in that asset anymore. Going back, I wrote why did AppLovin buy Adjust in February 2021. My argument was financial engineering because AppLovin had these locked-in service contracts and long-term contracts. There was a revenue boost. The app ecosystem component of it because AppLovin was still aggregating games at that point in time, and they had the IDFV. There were maybe ways you could line IDFVs up across publishers if you used the IP address, and then conversion value management, which was going to be the new thing everyone thought in the wake of SKAdNetwork 2.0 being launched with ATT. None of that really ended up being the case.

What they said was that they have a great sales team and they can upsell their clients on our ad network. Their ad network was growing a lot then, but their ad network was an underdog at the time, and that made some sense. My sense was that they saw ATT as just an opportune time to get good valuations on assets. A bunch of people took this approach, like Digital Turbine or Zynga. They just bought stuff because we are going to get good prices and we will find a way to stitch it all together. My sense is Adjust probably never really materialized as a meaningful part of a strategy.

AppLovin is a totally different company now than it was then. It is not an underdog anymore. It is the dominant app install network in many cases. I will go in and look at what companies are spending money on, and it is number two behind Google. It might even be number one in a lot of cases, ahead of both Google and Facebook.

I do know that my belief is, and I have talked to people that worked at the company and they said they really did keep a firewall between those two entities. One guy showed me he has an Adjust Slack and there is no AppLovin Slack. They had no access. My understanding, and I believe this to be true, they had no mutual back-end access. They kept those two things separate.

I do not think that they would want to buy AppsFlyer. My sense is it was a defensive maneuver against AppLovin by those companies because AppLovin is dominant in mobile gaming now. But I think also the existing investors wanted out. I actually advised the 2017 series D or series C Goldman investment, and I said that I think Apple is going to deprecate the IDFA and I do not know if an attribution company is a great investment right now. I wrote my piece, “The Coming War Between Apple and Facebook,” in 2017. That deal was 2017. I remember believing at the time that Apple is going to deprecate the IDFA. They do not want these companies usurping the direct customer relationship and building these big ad businesses and cutting Apple out of not just the ads piece, but the commissions piece and also control of the App Store.

My sense is this is just core infrastructure now. The existing investors wanted out. The IPO fell through and the acquisition by PE fell through. If they really wanted out, they would have probably forced a sale, maybe to somebody that you just would not want controlling that critical infrastructure. There are a number of companies that probably fit that description. So my sense is that is what it was.

Attribution has been a very durable practice, but measurement has evolved beyond just attribution. No one thinks of measurement as attribution anymore, which used to be the case with app advertising. It is just part of a toolkit. Even for a lot of gaming advertisers, it is part of a toolkit. These companies might have said that we require stability and predictability with the largest single attribution platform. It is worth it to us to put money into it to drive the valuation up beyond what any company would reasonably pay. That just protects that interest.

My sense is, and I do not have any inside information about this, the vast majority of that billion went to buying out the existing investors so that now they have a lot more patient investors that are behind them that are not going to try to force a sale anytime soon. I do not know how much of that hit the balance sheet and will be used for R&D. My sense is this was just exchanging these investors that wanted a payout with much more patient investors who see a lot of strategic value in having this be independent and neutral and not influenced by a buyer that might not have those same interests.

OK: I have a few questions for you on that. Number one, whose hands does this get into in terms of the wrong hands? Who are Google and Meta and Unity and Moloco nervous about? If not AppLovin, who else does this end up with where it hurts them in terms of the ad networks?

ES: You could imagine a ByteDance. Now there are issues there, so I do not know that they can necessarily pull that off. But you could imagine an Amazon. I think about who Google is, and there are different sets of competition fears. Google and Meta, that may have been more about creating this stable base that sits in opposition to Adjust.

You could think about gaming companies that could afford it. They were trying to sell at 1.9, and apparently, as reported by Costis, the PE funds thought they should pull back on that valuation because of the AI stuff. I guess the existing investors would not accept that. Think about two billion. It is pretty big. But you could think about an OpenAI. You could think about any of these companies that directly compete and will directly compete with Google and Meta for advertising dollars. ByteDance, Amazon, OpenAI with its advertising ambitions, two billion dollars to them is chump change. They spend 150 on TVPN.

There are any number of these companies, and then you see the sort of consolidation with CTV. Having this be a cross-channel measurement system makes a lot of sense. From Google and Meta’s perspective, there is a lot of defensive value there. From the Moloco and Unity perspective, I think you could make the case that you would not want private equity owning that and doing another rollup.

Now imagine a lot of these assets are cheap now. You see private equity saying that we can buy a measurement company, we can roll a couple of other things into there, other types of advertising surfaces, maybe even some apps, and then take another swing at the circa 2020 AppLovin approach, which was owning a big content portfolio plus the measurement and then having all that data from everybody else. That is the Moloco and Unity perspective.

But there is also just the fact that AppLovin owns the second largest attribution platform. The first largest, we should ensure that it is neutral and independent.

OK: I think you are right when you talk about attribution as infrastructure at this point. The question I have always wondered is if attribution is infrastructure and if it is more or less commoditized at this point, why do not brands do this in-house? This was one of the things that I could not believe when I joined Quibi, when we learned about the ecosystem. Nobody would ever build this in-house. Everybody has an MMP. It is just not something that you would do. Whereas in other industries, a lot of brands have their own in-house attribution model. If it feels like infrastructure, why do not more people build versus buy?

ES: That is such a good question, and I am working on a long-form solo podcast that is a history of the MMP program. It is because of Facebook. It is a protected cottage industry that Facebook established with the MMP program. Going back, Facebook called this the PMP program, product marketing partner. It was FMP, Facebook Marketing Partner program, then became the MMP program. If you wanted to attribute Facebook installs, you had to use an MMP and they got to decide who got to do that. It was a total cottage industry. They created it.

I remember King had its own attribution. When I was at a company called Wooga, we built our own attribution system. We did attribution for everything but Facebook. If you wanted to attribute Facebook, you had to go through an MMP and they did not give any exceptions. The tech is kind of, I do not want to say commodity, because I think that minimizes the complexity of it, but you could build it if you had to, and you could build it and maintain it. Wooga was not a huge company. The only reason you had to go through an official company in the MMP program was that was the only way you got access to the API to do Facebook attribution. That was it. So they kept this whole segment essentially, they prevented real competition from emerging.

These people got to charge prices that were not consistent with the quality of the service. You talk about the Apple tax, but I think there is an MMP tax. It is pretty expensive. A lot of the MMPs offer ancillary services and a lot of companies basically use MMPs for core analytics now, and that is why there is a lot of resilience built in. But that is why. That is why you have to have an MMP, because otherwise you cannot attribute Facebook app installs. If Facebook is your number one, number two, or number three, you need to be able to do that, and so you build everything around that.

OK: We will talk about this in terms of how MMPs are trending over time, and it still feels ubiquitous. My question was just why, and that helps answer it. It was propagated by Facebook. What is your perception of how customers and how the market perceived the AppLovin/Adjust deal? Do you think that part of the reason that has not been a very strategic acquisition is because the value of the asset was diminished by it belonging to AppLovin in the first place?

ES: A little bit, but not to the degree that you might have expected. This was in my consulting days, so I worked with a lot of gaming studios at the time. People explore the prospect of switching, but switching costs with MMPs are high. It is not a trivial task to swap out MMPs. There was a question of, well, AppLovin says they are not accessing that data. I believe that they were not, in the main. So what is the real risk here? Given the lift, the engineering resources, and the distraction of swapping out MMPs, does it make sense?

I would say that the number of defectors from the Adjust platform was a lot lower as a result of the acquisition than the number of people that stopped working with AppLovin when they launched their gaming studio. When they actually officially launched Lion Studios, there were more scaled gaming companies that said we are not going to work with AppLovin anymore after they announced their own gaming studio initiative than there were people that switched off of Adjust to other MMPs, at least in the short term.

I do think that you saw AppsFlyer eclipse Adjust in terms of feature set after the acquisition just because Adjust is a portfolio company essentially of AppLovin. That is how they are treated, and so I do not think that they invested as much in R&D as AppsFlyer was. You did see more of a trickle of people away from Adjust. I think they have got a pretty stable customer base. I do not know that it has dwindled down to something immaterial to the company. There was not that immediate exodus that you might have expected.

OK: That bodes well for this deal for AppsFlyer in terms of a minority stake, no influence, and in a consortium of platforms. To know that there was not a mass exodus in the hands of AppLovin makes me feel pretty confident that brands and advertisers are not going to see this as problematic, which is exciting because I think it is the right thing for the industry.

ES: Just one more thing on this because you keep saying brands, but we are talking about app advertisers. An app advertiser’s mentality about all of this is very different than a traditional brand. Even with the gaming studios stuff, it is like, okay, well AppLovin is now a serious competitor of ours when they are publishing games, but what are we going to do? Where are we going to go? If we cut AppLovin out of our mix, we are basically cutting off a stream of revenue. I think that is how they think about it. They can be much more calculated in those kinds of decisions.

Talk to me about what you were seeing in terms of app install attribution before this. What were the trends? What kind of insights do you think app advertisers were most interested in, and how had their measurement stacks been changing and adapting to support that?

OK: These MMPs are still very much ubiquitous. I think one of the last times I was on here, we were talking about this evolution in terms of measurement from attribution to incrementality. That has happened and it is underway. But we still see that there is a role here and that attribution has value.

The reason why is, despite it being clicky, it picks up these lower-funnel channels. It gives you signal on the lower funnel. It really misses impact in terms of upper-funnel view-based channels, like YouTube, CTV, or ChatGPT ads. It is definitely biased. But these advertisers need two things: they need daily signal and they need ad-level granularity. We have seen time and again that is non-negotiable for them.

What we are starting to see is incrementality-adjusted attribution in that we understand the bias. We understand the ways in which attribution is biased, and we can adjust for that bias, get to an incrementality adjustment, but still give them what they want, which is the daily ad-level granularity. You made a joke about these being app advertisers. You know this ecosystem even better than I do. They are in Meta ads all day. They are in Google all day. They are clicking around all day long. To expect that they are just going to be able to hand-wave and say, no, we will get you that incrementality readout at the end of the month, is just unrealistic.

That is what we are seeing in terms of this movement. We are actually bringing the best of these experiments and causal data into the dashboard. They can pull it into their system, into the view that they are used to looking at hourly, if not daily.

ES: Was that a function of having that incremental lens on it? Was that a function of exploring new channels? Or was that a function of leveling up and probably evolving from the last-click mentality that dominated app install advertising from 2012 when the practice really took off to 2019 or 2020 even, going into ATT? Was it an evolution or was it just that we need more channels and then we cannot use last-click on those, so we need a new approach?

OK: It is both. The bias in terms of correlation and click-based attribution is well understood at this point. You run a UAC standard campaign in Google, and it looks extremely efficient. Then you run a UAC video campaign that is delivering primarily on YouTube and other video channels and the video campaign looks horribly inefficient. The standard campaign looks great. You run an incrementality test and it is actually the reverse. You see that these things normalize when you are adjusting for incrementality, and budgets are adjusting accordingly.

You are seeing shifts out of the typical lower-funnel search. Apple Search Ads is a good example of a channel that looks amazing on a last-click basis in the MMP that tends to be less incremental. That has been underway for a while. Then you have the persona who just wants to expand into new channels.

There is this constant belief, whether rational or not, and I think there is some truth to it, that Meta and Google are perpetually delivering to your existing customers and the people who were going to come in anyway. Their algorithm is so good at finding intent that it tends to be non-incremental. There is just this belief across the industry that this is what is happening with their models. When folks sign up with Haus, their number one use case is that they want to introduce their brand and their app to new audiences. They want to break out of this bubble and go deliver to new customers, to new users.

The way in which they see themselves doing that is both signal engineering in Meta and Google, pointing it to shallower kinds of signals, but also expanding into new channels. Enter CTV, YouTube, and most video-based advertising channels are seeing a huge rise in popularity, and then the LLM ad formats of the world as well.

What is interesting here is they all come in really excited about this, and then something like the tariffs happen or there is a war in Iran and oil prices are going up, and the appetite to actually see through that channel expansion and to have the patience to actually go crack those channels tends to be a little bit lower than perhaps it was when they started the year. I am not seeing the rise in spend match the rise in excitement around a platform or a channel like CTV, especially in the mobile app space. I am not seeing CTV budgets taking 20% share of wallet, even though again you might think that based on the headlines. Every time times get tough in terms of macro economy or pacing behind on goals, they end up back in Meta and Google.

ES: Flight to quality, but flight to perceived quality and on the basis of the measurement, which I think most people would agree is not very sophisticated or robust if you are just doing last-click. But to your point, that is how they consume it, and consumption defines the framework in a lot of ways. If the CFO looks at this dashboard, then whatever we replace it with has to fit in the dashboard, and that is not as easy to do with pure-play incrementality or whatever other approach like MMM.

OK: What we are seeing though is the rise of the longer experiment. If the hypothesis is that these channels, like a Vibe or Universal Ads, is introducing your brand to new audiences, then you need a longer test. That plays out over a longer time horizon. One of the big learnings that we have had over the past year or so at Haus is a two-week test is just not enough time to even see anything on a CTV channel. What we are seeing is a lot of that lift is coming in the back half of a four- or six-week experiment. That is a reality that you do have to contend with, that these different channels have different payback windows and what is the value of a dollar that you get today versus a dollar that you get back tomorrow. We have learned is that you need a longer test, but that also comes with its own drawbacks.

ES: Good luck going to a CFO and saying we cannot rigorously make any determination about this unless we spend 500k over three months. You are probably not going to get approved.

OK: Marketers are not very patient. We have so many examples, and we just published some data on this, of most of that lift on CTV coming in the back half. If you cut it at three or four weeks, you are missing all the value.

ES: And the interaction effects. People do not appreciate that. It has value when paired with these other channels, and that is a big piece of it. A lot of MMMs deliberately do not even capture interaction effects, and so you are just going to miss that completely. Even if we do a longer test, but at a very tiny budget, you are going to run into the same issues. It is going to get absorbed by the bigger channels, and you are not going to really be able to distinguish. You have to really deliberately and thoughtfully engineer these tests.

People just do not, or it will be messy, and all of this in marketing suffers from collinearity anyway. It is very challenging. It is the reason why Haus exists. That is the whole point, to get that marketing expertise applied to your campaigns. I do want to talk about CTV and I do want to talk about the Vibe acquisition that was announced today.

Before we get there, I think another takeaway from this investment into AppsFlyer is that these companies clearly believe that SKAdNetwork and AdAttributionKit is going nowhere. If they thought this will evolve over time and this will become something meaningful, they probably would not do this. But no one does. This is dead. I chose violence one day. I was on a flight and I was like, are you kidding me, Apple? So you are just done even giving us updates?

You pushed companies into adopting SKAdNetwork wholesale. You convinced them this was the future. People spent millions of dollars, I was attached to projects where probably millions of dollars were spent in a single company adapting to SKAdNetwork, and you are not even going to give an update anymore. You are not even going to tell us what is new because nothing is new because you do not care about it. You just decided one day that you were going to force everybody to adopt this new methodology under threat of basically being deplatformed, having your company be killed if you get your app removed from the App Store because you were deemed to be non-compliant. You are talking about a potentially existential threat.

You are not even following through. What is this, five years later? You are not even following through with updates anymore. This is dead. This was just so disrespectful and so inconsiderate and so callous to the entire mobile app ecosystem that supports you. And you cannot even deign to give an update anymore about what the roadmap is or any new features, if there have been any, that have been added to AdAttributionKit. Are you kidding me? So it is dead.

OK: Well said. We have got to clip that. I think about this all the time. It is still infuriating. They did it all under the guise of privacy with the billboards everywhere in 2020. I do not understand what the end game was because my thought at the time when they did it is they are getting ready to launch a big ads business of their own. This is the only logical explanation for what they are doing, that they want to win in the ads business. But they are not really doing that. They are not really trying to build a much bigger ads business, at least it does not seem like that to me. Did they just not realize what they were getting into here? Did they not realize how hard it was going to be? Does that bode well for the AppsFlyers of the world if this is just painful and complicated? Strategically, it makes no sense. I feel like someone at Apple should have to explain themselves.

ES: I would hope they would have had to do that in some kind of congressional testimony or something, but that never happened. It just seems like they have abandoned it. They are not even giving updates anymore. They have just left people with this complete, dysfunctional, incomplete framework that is effectively useless and walked away from it as if that was going to solve any problems or fill this void that they created or that they conjured up. I do not want to get into the ATT stuff because that is old. That dead horse has been mutilated. But it was just the fact that they cannot even go through the motions of pretending like they are continuing to invest in it. They have completely moved off.

OK: They have completely abandoned it. There is no explanation for that other than they just did not realize what they were doing at the time. It was not well thought out.

ES: I think that is right. I also think there was not a consensus internally and there were just a lot of people that made the case that this would not have that big of an impact. Maybe the experience from ITP was informed by that decision because ITP did not really change that much because Safari was not the dominant browser and people found workarounds. But they just underestimated the impact.

I want to talk about CTV because that is a really interesting, you mentioned that you see the app advertisers are not shifting meaningful budget. But is it trending up into CTV? Is there more experimental budget going there? How do app advertisers from your perspective use CTV?

OK: I think what the Vibes of the world and the Universal Ads of the world have done really well is position TV as a performance channel. Good luck telling the CFO you are going to invest more money and for a longer period of time and you won’t be able to show returns for six months. That is a very difficult conversation to have. Where Vibe is winning is they have positioned TV as a performance channel.

Universal Ads has the same messaging in market of, run this like your Meta ads and we are going to hold ourselves accountable to a performance outcome. I think that is working really well. Everybody talks about brand marketing and wanting to invest in building the brand, but the reality is they only care about brand in so far as it is driving downstream outcomes. You eventually have to tie the brand spend to an outcome or it is going to be the first thing on the chopping block. It is going to be the first thing that gets cut.

I think Vibe is doing a really good job of that. Similar to AppLovin when they launched the e-commerce pilot, they have leaned into third-party measurement quite a bit. They are funding a lot of Haus tests where they will put credits up and they will make it really low risk, very easy to test. Universal Ads is the same way. Thus, the marketers and the teams can justify it as a performance channel.

That being said, I do like to be realistic about what is possible here. We have some data now on how quickly you see the effects of a Meta, how quickly you see the effects of an AppLovin. It takes a little bit longer to play out in terms of these effects. You are introducing your brand or your app to new audiences. They are going to need more frequency. They are going to need more time.

For a channel like CTV, you have just got to run a longer test and you have to be realistic about expectations there. You have to make sure you can protect that long enough to see the effects. That is the thing that I think the teams that we work with struggle with, letting the bet play out for long enough to actually make it an always-on channel that is there alongside the Metas, the Googles, the AppLovins of the world.

ES: And that timeline is important to think about too. No serious person would say brand marketing cannot possibly improve my commercial prospects or contribute to my marketing success. No serious person would say that. But you do hear a lot of people saying brand marketing will absolutely contribute to your commercial success, and that is just as absurd of a statement. You have to test it. You do not know that.

Upper-funnel marketing can possibly improve the impact of my marketing on my business. On what timeline? Because it has also got to do it on a timeline that I care about. If it is in five years after $100 million of expenditure, well, maybe I will leave some money on the table then and not pursue it. I understand that I am under-investing by that standard or from that perspective, but the time value of money and the prioritization and the return on other things is just greater.

It should not be driven by religion or blind faith. You need to test it. But you need to determine when I take into account the time value of money and I take into account the ROI when I rank that expenditure against other places I can put that money, is it going to be the best possible use of that money. That is an important exercise to go through.

OK: The conversation I am having a lot right now comes up every year. Every year there is this big, “Hey, we need to crack TV.” And then they will find reasons not to test. This actually happens more in the world of linear TV than CTV because CTV is pretty easy to test and it is pretty easy to get going. In linear, it is a little bit harder to run a geo test.

Every year they will look at doing a geo test, they won’t run it, they might try it for a few months, they won’t be able to see any discernible impact to the business. You mentioned collinearity. It is a multi-collinearity problem where they are lighting up TV right around the time that their business seasonality tends to ramp up. Then nobody has any idea what it delivered to the business, and then they turn it off, and then they have the same conversation again next year.

So I have been pushing really hard. I know this is going to take longer and it is going to be more expensive than you would like, but you need an answer to this question. That is what we have been seeing a lot of. These delayed impacts feel very imaginary in a lot of ways, and we are trying to help these brands actually assign some data to it. The existing research and data on long-term effects is not good. There is not much of it, and so we are finding that we are just having to go and run these tests ourselves and collect the research.

ES: When you approach this, it is important to have some kind of measurement framework. A lot of the other people that advocate for brand advertising as a necessary part of the toolkit often say that it cannot be measured. Well, that is not really helpful. I had Carl Mela on the podcast and what I loved about his framework is that he is measurement first. It is, “Here is how you measure that. I am not just making a claim that you must be building a brand.” This is the approach you should take to measuring it. I like also that they rooted that in a lot of financial metrics because then that is an easier sell to your finance team.

OK: We are working with some very large brands, but for the most part, we are not talking to P&G here, like a big CPG conglomerate. We are talking to a series B or series C startup that is fighting every day to survive. The brand conversation is just a very difficult one at that stage. You can look at these shallower proxy metrics, but you need to tie them to downstream sales. I think that is part of the reason we have been doing well as a business. We speak and we report on the language that the finance teams understand and the board understands. We are speaking in the same language in terms of these metrics.

ES: I also think that is what is important and differentiating and meaningful about the approach taken by Universal Ads. It is not so much that it is measurable from a click standpoint because it is not. It is about letting you be flexible with the budget and with the targeting, because if you do not have that, there is no way to do real experiment-based measurement. There just is not. And if you set the bar for a campaign to be a million dollars minimum, then forget it. You are never going to get a smaller company investing in that. They need that money-at-risk to be much lower than that.

OK: Vibe has made it incredibly easy to do business with them. They are so accommodating and so easy to work with. There is something to be said for that versus a Trade Desk on the other hand with minimums and a lot of barriers to entry. That team just has been hustling like crazy. I am really happy for them.

ES: I want to talk about the report you recently published on AppLovin. They just today went GA with the self-serve, so maybe it was in response to the AppsFlyer news to try to suck all the oxygen out of the room, but I think that has been on the schedule for a long time. Tell me about the report. What did you find? What were the primary findings of the report?

OK: The conversation around AppLovin has been very polarizing for a very long time. I started off very skeptical just with a background in programmatic open web. It is really hard to differentiate in that space and there is just not a whole lot of great inventory. So I myself personally started off skeptical.

We started testing. We looked at 15 months as part of this most recent analysis. Last year throughout 2025, we were very encouraged by results and we were posting them along the way, but this meta-analysis looks at 15 months of data. What we consistently see is that AppLovin is hanging in as a strong number two or number three channel in the mix. We looked at how it is comparing to the other experiments running at the same time, and we saw efficiency was 11% better than the average of other channels that those same brands tested.

So it is looking good. It hits hard. AppLovin works quickly and that can be a feature. That can be a really good feature. It is very easy in terms of working it into your models and your payback cycles. It works fast and it is also primarily driving D2C. It lowers the measurement complexity because most of the impact is happening on digital channels versus offline on an Amazon or a retail.

Those are the key stats. The big one that we looked at, the big question was, is it hanging in as these brands scale the channel? The data there is a more mixed story. I think this is probably similar to a lot of new channel stories where you see really good efficiency, and then the next question is how that efficiency hangs in there when you start scaling the channel.

What we looked at was how AppLovin is performing versus the typical test run at the same time to control for seasonality. AppLovin was beating the typical test between 60 and 80 percent of the time in 2025, so just consistently winning in terms of performance. In 2026, that number has come down and it has settled to closer to a coin flip, which means only half the time is it beating the average of other tests in other channels being run at the same time. That suggests that early edge that they have is normalizing as brands scale up spend.

But we are seeing AppLovin’s share of wallet increase very clearly. We have data, I think it is 5% in January to over 8% share of wallet in May. So it does seem like brands are leaning in. Obviously, this data is mixed. It is not perfect in terms of the data on how it is scaling, but they are commanding a pretty respectable share of wallet in terms of 8%. We do not see anyone else in terms of the secondary platforms growing at that clip.

ES: That is an incredible share. 8%. That is impressive in the e-commerce space in the course of 15 months.

OK: That is API spend we can track. It is the core digital channels. It would not include TV as an example. It is the Meta, Google, AppLovin, Snap, TikTok, Pinterest.

ES: Wow. Okay. By the way, it just occurred to me, you are not chatting with me from the beach in Cannes. I do not see turquoise water in the background here.

OK: If I were there, then I would not be here with you. This is what I live for, Eric. I am better behind the desk.

ES: I just saw the most appropriate description of Cannes from Matt Belloni. I’ll just read it. “Welcome back to what I’m hearing, thankfully not coming from the Cannes Lion Ad Conference. I shouldn’t have to say this, but to everyone asking if I’ll be at Cannes this year: Cannes is the Festival de Cannes, a prestigious global film event where talented creative people display and sell their work and look glamorous doing so. Cannes Lion is not. It’s a tacky business conference for selling advertising and announcing brand partnerships populated mostly by punchy middle managers and YouTubers. They call it an international festival of creativity, but that’s just to make the suits and the media people feel better about their soulless corporate boondoggle. Enjoy the Rosé.”

OK: I wish the Haus experience was ten years ago or ten years from now where my kids are a little bit older. It is just so hard to justify. I agree with a lot of what he said.

ES: I hear you. Olivia, thank you for coming on the podcast with essentially no notice. I appreciate your insights as always. Talk to the audience about Haus. How can they learn more?

OK: Haus.io. We just launched a lot of very ambitious new products in terms of roadmap. We have been building toward this since the early days of Haus in terms of how you reduce the friction from insight to action. How do we get better at operationalizing incrementality in a way where you are actually using these tests to improve business outcomes? Our roadmap is all in pursuit of that and we would love to talk to more folks about what we are building and get thoughts and feedback from smart people. Feel free to message me. My DMs are open.

ES: Thank you so much.

OK: Thanks, Eric.