Interstitial after level-fail: incremental revenue vs churn risk
Illustrative lab for interstitial-after-fail placement versus a no-interstitial control, anchored to AdMob interstitial and rewarded format definitions.
Placement choice is an operator lever. This lab compares interstitial-after-level-fail against a control using illustrative retention and eCPM inputs. It does not claim a genre-optimal placement.
Format definitions
AdMob interstitial ads define full-screen ads at natural transition points. Rewarded ads are a separate opt-in reward format — do not conflate fail-interstitial forced views with rewarded completes. AppsFlyer ad-revenue partner sharing documents af_ad_revenue for mediation revenue used in the IAA side of the readout.
Synthetic worked example
| Arm | Interstitial eCPM | Fail→interstitial rate | D1 retention | IAA / install D1 | Notes |
|---|---|---|---|---|---|
| Control (no fail interstitial) | — | 0 | 0.40 | 0.020 | Synthetic |
| Treatment (fail interstitial) | 8.00 | 0.55 | 0.36 | 0.048 | Synthetic |
Illustrative treatment IAA uses impr_per_install × eCPM/1000 with impr_per_install = 0.55 × 1.1 (synthetic) → 0.605 × 8 / 1000 = 0.00484 scaled in CSV to cohort math — use CSV outputs. Directional lesson: incremental IAA can coexist with retention damage; register both before scaling.
| Output | Control | Treatment |
|---|---|---|
| D1 retention (input) | 0.40 | 0.36 |
| Net IAA index (synthetic) | 1.00 | 2.40 |
| Retention index | 1.00 | 0.90 |
Facts versus limits
| Field | Evidence |
|---|---|
| Interstitial format | AdMob interstitial docs |
| Rewarded ≠ interstitial | AdMob rewarded docs |
| Measured churn elasticity | Not on cited pages |
