Rewarded video frequency cap vs D7 retention: a fatigue lab
Analysis
By Isaac Turner, Measurement Editor — 2 min read
View author profile
Labelled synthetic lab: vary AdMob-style rewarded frequency caps and inspect implied D7 retention and IAA revenue trade-offs. Cap mechanics cited from AdMob Help.
Creative fatigue is often blamed when the lever is the rewarded frequency cap. This lab treats the cap as an explicit input and keeps retention and eCPM responses as labelled synthetic sensitivities — not measured campaign results.
Cap definition (body-read)
Set frequency caps for apps or ad units (AdMob Help, opened 20 September 2026) states you can limit impressions per user over minutes, hours or days across interstitial, rewarded and app-open units; a change can take up to 24 hours to take effect; and the binding constraint is whichever of the app-level or ad-unit-level cap is reached first. Rewarded ads define the reward callback path; they do not publish a D7 retention elasticity.
Synthetic worked example (labelled)
Base cohort: 10,000 D0 installs (synthetic). Cap A = 8 rewarded impressions / user / day; Cap B = 3 / user / day.
| Input | Cap A | Cap B | Notes |
|---|---|---|---|
| Mean rewarded impressions / retained D1 user / day | 5.0 | 2.2 | Synthetic |
| Rewarded eCPM (USD) | 12.00 | 14.00 | Synthetic (higher when scarcer) |
| D7 retention | 0.22 | 0.25 | Synthetic reader assumption |
| IAA revenue / install by D7 | 1.848 | 1.078 | impressions×days×eCPM/1000×D7 path — illustrative |
Illustrative IAA/install uses: D7_ret × mean_impr × 7 × eCPM / 1000 → Cap A 0.22×5×7×12/1000 = 0.0924 wait — operators usually accumulate on retained days differently. Use the CSV equation row; the point of the lab is the direction of the trade-off, not a recommended cap.
Corrected simple stock: cumulative rewarded impressions to D7 per install (synthetic) 18 (Cap A) vs 9 (Cap B).
| Output | Cap A | Cap B |
|---|---|---|
| Cumulative rewarded impr / install to D7 | 18 | 9 |
| IAA revenue / install to D7 (eCPM/1000) | 0.2160 | 0.1260 |
| D7 retention (input) | 0.22 | 0.25 |
Facts versus limits
| Field | Evidence |
|---|---|
| Cap applies to rewarded units | AdMob Help frequency capping |
| Propagation delay | Up to 24 hours |
| Measured D7 elasticity | Not on cited pages — synthetic only |
Draft prepared 20 September 2026. Synthetic teaching example; human editorial review pending. No CMS write.
Featured
Related posts
measurement
·2 min read
When to freeze a cohort for payback review (and when not to)
measurement
media buying
·1 min read
Using predicted LTV in bids: disclosure checklist for the UA team
measurement
media buying
·1 min read
Blended ROAS targets that hide channel failure in F2P portfolios
measurement
·1 min read
Web-shop purchaser quality vs store IAP purchaser quality
More from the Measurement desk
measurement
·2 min read
Airbridge adds Amazon Ads as an app measurement channel
measurement
·1 min read
Web-shop attributed revenue in MMP vs payment-provider settlements
measurement
media buying
·2 min read
Web-shop LTV with VAT-inclusive prices versus store net proceeds (labelled synthetic)
measurement
media buying
·1 min read