When to Scale UA Spend: A Decision Framework for Mobile Games
Scaling is a sequence of evidence-sized bets, not a switch. The question at each step is whether the data you have can support the size of the decision you are about to make.
When is a mobile game ready to scale UA?
When three things hold for the market and source you want to scale: the matured cohorts you have already bought pay back within your acceptable window, the last budget increase still earned an acceptable marginal return, and the product holds up with the broader, less motivated users that more spend will bring. If any one is unknown, you can still increase spend — but label it a learning step and cap it.
What evidence do different decisions need?
- Keep a campaign running or add spend within a cap: stable CPI and encouraging early ROAS may be enough. This buys learning.
- Scale materially: retention and ROAS curves from cohorts old enough for your payback window; marginal ROAS of the last step; stability across a few weeks.
- Enter a new market: its own retention and yield, not the portfolio average — see the Geo Budget Planner.
A small increase in UA spend and a decision to scale should not use the same evidence bar.
What should you check before each increase?
- Cohort maturity. Are the numbers you are reading complete for their age? See data maturity.
- Marginal, not blended, ROAS. What did the last step buy? The Marginal ROAS Calculator shows it.
- Creative concentration. Is one creative taking most of the spend — and does its cohort retain like the rest?
- Segment health. Is the market receiving the next dollar holding up, or is a blended average hiding it? See The Averages Trap.
- Cash. Can you fund the deficit while cohorts pay back? The Bid Runway tool models it.
How big should each step be?
Big enough to learn something, small enough that being wrong is affordable. Many teams step in increments and hold each level long enough to produce readable cohorts at the age that matters. At portfolio peak we ran UA at up to $700K a month at a 45% blended ROAS; the discipline that mattered most was not the ceiling, it was refusing to treat a learning increase as proof the scale question had been answered.
When should you stop scaling?
When marginal ROAS at equal cohort age falls below your hurdle and new creatives or markets do not lift it. That is the current scale ceiling. Raising it is usually a product job — retention, monetization depth — as much as a media one.
Related: A Small UA Increase and a Scale Decision Need Different Evidence · What Is Marginal ROAS · Cheap Installs Can Still Produce an Expensive Growth Problem