The hidden fixed costs of a paid UA operation

Most studios treat paid user acquisition as the purest variable cost in the business. Spend goes up, installs go up; spend goes down, installs go down; the finance team models it as a straight line through the origin. The claim here is that this model is wrong in a way that matters. A meaningful share of a paid UA operation doesn't move with spend at all, or moves only in steps, and pretending otherwise leads studios to cut spend at exactly the moment cutting it is most expensive per install.

The mistake is understandable. The media invoice is the largest line, it's the one that moves month to month, and it's the one the CFO can see. Everything else that makes the media invoice productive sits in other cost centres under other names and never gets divided by the number of installs it produced.

What is actually fixed

Start with the people. A UA manager, a creative strategist, a producer for playables and video, a measurement analyst. Even a lean team is four or five salaries before a single impression is bought. Those salaries don't fall when a third comes off the budget; they fall when the studio makes someone redundant, which is a different decision with a different timescale.

Then the contracts. Mobile measurement partner agreements typically come priced on tiers with annual commitments and minimums. Creative tooling, ad intelligence subscriptions, playable authoring platforms, a data warehouse plus the connectors that feed it: each is a fixed monthly charge negotiated on the assumption of a certain scale. Some networks and agencies also carry minimum monthly commitments in exchange for better rates or dedicated support.

Then the switching and maintenance costs that nobody budgets for until they arrive. Unity's announcement in late March that the ironSource Ads network would shut down on 30 April is a clean example. Every advertiser on that network has to re-plan the spend, re-upload creative elsewhere, re-map postbacks, then pull historical data before the export window closes at the end of May. None of that is media. All of it is cost, and it's cost that scales with the number of channels a team runs, not with the number of installs it buys.

Finally, creative volume. AppsFlyer's State of Gaming for Marketers 2026, published in January with Unity and Newzoo, put the top advertisers at 2,400 to 2,600 creative variations a quarter, up 25 to 30 percent year on year. Creative production has a variable component, but the machine that produces it, the people, the pipeline, the review process, is a standing cost. A studio that has built that machine can't switch it off for a quarter and switch it back on without losing most of what made it work.

Why the fixed share changes behaviour

The mechanism is arithmetic. If a studio spends 900,000 a month on media and carries 300,000 a month in fixed UA cost, the fully loaded cost is 1.2 million. Say that buys 400,000 installs, for a media CPI of 2.25 and a fully loaded CPI of 3.00.

Now cut media by a third to 600,000. Installs fall roughly in proportion, to around 267,000, and media CPI stays at 2.25; but the fixed 300,000 now spreads across fewer installs, so the fully loaded CPI rises to about 3.37. The studio has saved 300,000 a month in cash and made every install it still buys 12 percent more expensive on a true basis. These are illustrative figures, but the shape is universal: the fixed layer acts as a lever that amplifies the per-unit cost of any cut.

The second-order effect is the one that hurts. Payback models are almost always built on media CPI, so the cut looks neutral in the dashboard while the studio's real unit economics have deteriorated. If the studio cut because payback was marginal, it has just made payback worse and can't see it.

The same lever works in reverse when scaling. Increasing media spend spreads the fixed layer thinner and improves the true CPI, at least until the team hits a step: another hire, a new MMP tier, a second creative pipeline. Fixed costs in UA aren't flat; they're a staircase. Knowing where that step sits matters more than knowing the current fully loaded CPI.

A rule for the review meeting

The practical fix is to carry two numbers in every budget review. The first is media CPI or media cost per payer, which is what the platforms report and what the team optimises against day to day. The second is fully loaded cost per install, which includes every recurring cost that would not exist if the studio stopped buying users tomorrow.

A simple decision rule follows. Before approving any spend cut, ask what the fully loaded CPI becomes after the cut, and whether the payback model still clears at that number. If it doesn't, the choice isn't "cut spend" but "cut spend and reduce the fixed layer", which is a slower, harder conversation involving contracts and headcount. Making that explicit stops the quiet version of the decision, where someone trims the media line and leaves the fixed layer to erode the economics unobserved.

The reverse rule applies to scale. Before approving a spend increase, identify the fixed-cost step ahead and its size. If the increase crosses the step, the incremental installs carry the whole step, and the marginal fully loaded CPI can be far worse than the average suggests.

The cost that never appears anywhere

One more category deserves attention because it appears on no invoice at all. When a UA operation runs below the scale its fixed layer assumes, the team's time doesn't get repurposed; it gets diluted. The creative strategist produces the same number of concepts for fewer live campaigns, the analyst spends more time reconciling smaller datasets with wider confidence intervals, and the whole operation becomes slower to learn. A studio that halves its spend doesn't halve its learning velocity. It usually loses more than half, because statistical power falls faster than spend.

That's the argument for treating the fixed layer as a deliberate design choice rather than an accident of hiring history. A team sized for 300,000 a month in media shouldn't be carrying the tooling and contracts of a team sized for a million, and vice versa. The audit is simple: list every recurring UA cost, note the spend range it assumed, and compare that range to the budget the studio actually expects to run for the next two quarters. Anything sized for a scale the studio no longer plans to reach is a fixed cost with no installs to absorb it.

Related archive reading

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

• D7 ROAS targets punish long-retention games
• Layoffs and the UA function