If app stores become optional, what happens to UA

Google Play has been surfacing third-party Android stores to US users since July under the Epic injunction. Apple's revised EU fee structure, announced in August with a flat five percent Core Technology Commission for apps distributed outside the App Store, is due to take effect on October 1. The Supreme Court will hear Apple's appeal in Epic v Apple this term, and Apple filed its opening brief this month arguing the court should vacate the contempt finding. Whichever way the details fall, a world where the store is one option among several is now the planning case rather than the speculative one.

The comfortable reading is that optional stores lower the cost of acquisition, because the thirty percent goes away. I think that reading is wrong. The store bundled distribution with payments and with trust; making it optional unbundles the three and re-prices each separately. The commission falls. Distribution and trust get more expensive, and the party best positioned to sell them is the ad network. Acquisition does not get cheaper. The toll moves.

What the store actually sold you

A store commission looked like a payment fee because the platform collected it on payments. It paid for more than that. It paid for a single install path every user already trusted; for ratings and reviews that did conversion work for free; for a search box that delivered organic installs; and for a payment sheet nobody needed persuading to use.

Xsolla's August announcement (a vendor figure, and worth treating as one) described web shops at roughly fifteen percent of global mobile IAP, on the back of eight hundred or so mobile game web shops in its own ecosystem. That is the payments piece unbundling first, because it is the piece studios could take without asking permission. Distribution is unbundling now, through the Play injunction and the EU terms. Trust is the last and hardest, because it was never a line item.

The mechanism: unbundled tolls find new collectors

When distribution is no longer routed through one store, someone still has to put the game in front of the user and get them to accept a less familiar install path. That job has a name. It is advertising.

The store's cut was a percentage of revenue, paid after the fact, only on success. The ad network charges up front (per impression or per action) whether or not the user ever pays. Moving a dollar of toll from the first structure to the second changes who carries the risk. Under commission, the platform carried it. Under advertising, the studio does.

This is why the ad networks are the natural winners of optional stores. Unity said in August that Vector had passed a one billion dollar annual run rate, and AppLovin's second quarter revenue grew by more than half year on year according to its August results. Those numbers predate any meaningful shift in distribution. As the store's gravity weakens, the share of a game's total customer acquisition cost paid to the network rather than the platform rises, because the network is now doing part of what the store used to do.

Two second-order effects

The first is fragmentation cost. Every additional install path is a separate creative and QA surface. This site published a day-one UA playbook when third-party stores went live on Play in July. The larger point is that each path needs its own end card, its own store-page equivalent, its own install instruction and its own measurement integration. A studio with one path had one set of assets to keep in sync. With four, the coordination overhead grows faster than the paths do.

The second is trust. Ratings and reviews on the mainstream store did quiet conversion work; outside it the ad itself has to carry that load, which means longer creative, more explicit gameplay proof and probably lower conversion at the install step until users acclimatise. Creative teams should expect the persuasion burden per install to rise on non-store paths, and budget for it rather than meeting it as a surprise in the CPI.

An illustrative toll ledger

Take a purely illustrative user worth one hundred in gross lifetime spend.

Through the mainstream store today: roughly thirty goes to the platform on spend, and say twenty goes to the network to acquire the user. Fifty remains.

Through a third-party store or direct path: platform take falls to somewhere between zero and five depending on region and terms. Payment processing costs about three. But the network toll rises, because the install path is less familiar and the ad has to do the store's trust work, to something like twenty-eight. Add a couple for the extra creative and QA work per path, and for the attribution overhead. Around sixty-two remains.

That is a real improvement; it explains why studios are moving. But it isn't the fifty to eighty jump the headline commission number implies, because most of the saving goes back to the network and to your own operations. A team that plans on the headline number will over-spend into the new paths and wonder why margin did not follow.

The decision rule for buyers

The test for whether an optional path is worth opening is not whether the commission is lower. It is whether the margin after the new distribution toll beats the mainstream store by more than the fragmentation overhead, for the segment of users who will actually take that path.

That last clause is where most plans fail. Not every user will install from an unfamiliar store or buy from a web shop. The ones who will are disproportionately your engaged, already-acquired players. For them, the path is a margin play on revenue you would have earned anyway. For new users, the trust deficit means the network toll is highest exactly where the commission saving should be largest. Model the two populations separately, or the blended number will flatter the case for a path that is only working for players you already had.

Related archive reading

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

• Alternative app stores will not lower your CPI
• The Scopely Niantic Acquisition and Location-Based UA