Mobile apps fatigue faster than any vertical we model. The median creative lifespan is 5 days. Hypercasual games sit at 3.
The cause is not that game creative is worse. It is arithmetic: install campaigns run at high volume against deliberately broad audiences, so a single asset reaches effective frequency inside three to five days. Nothing is wrong with the ad. The audience has simply finished with it.
That changes what a creative strategy has to be. At a 3-day lifespan you are not looking for a winner to scale. You are building a rotation you can feed.
The economics you are working inside
| Segment | CTR | CPA | ROAS | Lifespan |
|---|---|---|---|---|
| Hypercasual games | 1.65% | $3 | 1.4x | 3 days |
| Mid-core games | 1.15% | $14 | 1.7x | 5 days |
| Utility / productivity | 0.92% | $11 | 1.9x | 7 days |
| Health & fitness apps | 0.88% | $18 | 2.1x | 5 days |
| Fintech / subscription | 0.65% | $34 | 2.4x | 8 days |
Two things stand out. Return runs inverse to volume, so the cheapest installs produce the thinnest ROAS. And lifespan tracks consideration: the more someone thinks before installing, the longer your creative survives.
The full modeled set, including CPM and campaign-type breakdowns, is on our mobile apps benchmark page.
Six formats that survive the rotation
1. First-30-seconds gameplay, unedited
Raw capture of the actual opening loop, no polish. It works because it sets an accurate expectation, and accurate expectations lower the day-1 churn that makes cheap installs worthless.
Rotate by: level, character, or starting difficulty. One capture session yields a month of variants.
2. Failure-then-success
The player loses, then clears it. This is the most reliably reusable structure in games advertising because the emotional beat survives repetition better than a feature callout does.
Rotate by: which level fails, how close the miss is, what the reward looks like.
3. Choice-on-screen
A decision presented mid-clip, often with a countdown. It converts because it starts the play loop inside the ad itself rather than describing it.
Watch for: this one fatigues fastest of the six. Treat it as a 2-day asset in hypercasual.
4. Progression compression
Level 1 to level 50 in eight seconds. Sells depth to mid-core audiences who need to believe the game has a long tail before they install.
Best for: mid-core, where the 5-day lifespan and $14 CPA justify a more produced asset.
5. Real player reaction
A person reacting to a moment of play, filmed on a phone. Lower production cost than it looks and it survives longer than studio-produced footage because it reads as a person rather than an ad.
6. Re-engagement, built separately
Not a prospecting format. Existing users who lapsed respond to what changed since they left, and the modeled numbers are the strongest in the vertical: 1.45% CTR, $5 CPA, 3.4x ROAS against 1.5x for broad install.
If re-engagement is currently sharing creative with your install campaigns, separating it is usually worth more than any new prospecting concept.
Building the rotation
Six formats at a 3-day lifespan means roughly ten refreshes a month per format if you run them all continuously. Most studios do not, and should not. A workable structure:
1. Run three formats concurrently, not six. Rotate which three every fortnight. 2. Keep one long-lived format in the mix, usually progression or player reaction, to reduce total refresh load. 3. Treat choice-on-screen as disposable and budget for rebuilding it twice a week. 4. Give re-engagement its own pool and its own cadence, since it fatigues more slowly against a smaller audience.
The failure mode here is not bad creative. It is a production process built for an 8-day vertical running in a 3-day one, which shows up as rising CPA that looks like an auction problem and is actually a supply problem.
If the volume side is the constraint, that is the specific thing creative generation is for, and current plans are on pricing.