The consolidation endgame for mobile ad tech

The consolidation story in mobile ad tech is usually told as a race to a single winner. It is not that. The structure the market is settling into has two or three vertically integrated stacks, each combining a demand-side model, supply through mediation or owned inventory, and increasingly its own creative tooling, sitting on top of a measurement layer that the stacks collectively part-own. The endgame is not monopoly. It is a small oligopoly with shared plumbing, and that is a harder structure for a buyer to manage than either a monopoly or a fragmented market.

Practitioners who have lived through the last three years of network churn may find the shared-plumbing point overstated. The events of this year argue otherwise.

The year the shape became visible

Take the sequence. In March Unity announced the ironSource Ads network would shut down at the end of April, redirecting its Grow business to Unity Vector. In August Unity reported Vector past a one billion dollar annual run rate two quarters ahead of plan, and sold the Supersonic publishing label to Tripledot for about forty million dollars. Unity kept the model and the mediation. It shed the parts that did not feed the model.

AppLovin reported Q2 revenue of 1.92 billion dollars and announced the acquisition of Wurl, a CTV and streaming ad distribution business. Having opened Axon Ads Manager to global e-commerce self-serve in late June, it is now assembling supply outside mobile games entirely. Moloco, which Bloomberg reported in February had appointed banks for an IPO, launched Performance CTV in April and won Costco's onsite commerce media business. Liftoff listed on Nasdaq in June.

Then in June AppsFlyer raised more than a billion dollars at a 2.7 billion dollar valuation, with Google, Meta, Unity and Moloco as minority strategic investors. Read that list again. The measurement company that buyers rely on to arbitrate between networks now counts four of those networks, or their parents, among its shareholders. Minority stakes are not control. They are, however, alignment.

The mechanism driving it

Each stack is doing the same thing for the same reason. A delivery model improves with data, and the most valuable data is the closed loop from impression to install to in-app event. Owning the mediation layer gives a stack the impression and the auction context. Owning the creative tooling gives it the ad features. The one piece it does not own is the post-install truth, which sits with the MMP and the advertiser.

There are two ways to close that gap. One is to build attribution into the stack and encourage advertisers to trust it, which every network has tried and buyers have resisted. The other is to make the independent layer a friend. A strategic stake buys a seat at the table when data-sharing terms are set, and a shared interest in the MMP's survival as the neutral clearing house rather than a challenger to it.

The incentive for the MMP runs the same direction. In January it was reportedly in talks to sell at around a two billion dollar valuation, a deal that never closed. Six months later strategic money valued it higher. Neutrality that everyone pays to preserve is a better business than neutrality nobody funds.

The second-order effect for buyers

The obvious risk of vertical stacks is concentration: a handful of vendors control most of the spend and price accordingly. The less obvious risk is that the layer buyers use to check the stacks against each other becomes structurally reluctant to embarrass any of them.

That does not require anyone to behave badly. It only requires that product roadmaps, data-access decisions and methodology choices get made in rooms where the stacks have a voice. Over time the measurement layer's definitions of a good install, a valid attribution window or an acceptable overlap converge on what the stacks find comfortable. Buyers who rely on the MMP as their sole arbiter will not see this happen, because the dashboard will keep working.

The consequence is that the only measurement a buyer fully owns is the one it runs itself: holdouts, geo tests and incrementality reads that do not route through a party with a stake in the answer. UA Ledger's April piece on what Q1 consolidation does to CPI looked at price. The measurement dependency is the quieter cost.

A dependency map and a rule

The practical response is not to boycott the stacks. It is to know where your dependencies overlap and to cap the overlap.

For each vendor relationship, note which of five functions it performs for you: demand (buying installs), supply (monetising your inventory), mediation (deciding which network fills), creative (producing or optimising ads), and measurement (attributing or validating). Then look for vendors that hold three or more, and for measurement providers with shareholders among your top demand sources.

An illustrative rule for a studio spending across several networks: no single stack should hold more than two of the five functions for any one title, and no more than about forty per cent of paid spend should sit with stacks that also mediate your inventory. Where you cannot avoid a triple overlap, which for many ad-supported casual studios means the stack that runs their mediation, keep a standing incrementality test on that spend that does not depend on the stack's or the MMP's numbers.

What to watch in the next two quarters

Three things will tell you whether the shared-plumbing structure is hardening or loosening. Whether the MMPs with strategic shareholders change data-sharing terms or default attribution settings in ways that favour any investor. Whether Moloco's listing, if it proceeds, brings in another stack with its own supply ambitions or leaves it as the one large demand-side player without owned inventory. And whether any of the stacks moves on a creative tooling acquisition, which would close the last gap between what the model sees and what the advertiser controls.

The buyers who come out of this well will not be the ones who picked the winning stack. They will be the ones who kept their own scoreboard while the stacks were agreeing on the rules.

Related archive reading

These articles provide related context and remain subject to their stated review status.

• Scenario planning for a world with no device IDs
• Ad tech integration hurts mid-sized studios most