Event-week retention uplift that is not UA-attributable
Calendar effects look like UA wins. Isolate organic vs paid retention in the same event week using documented cohort filters.
Live ops often claim UA wins that are calendar effects. If organic D1 retention jumps in the same week as a festival event, paid cells riding that tide are not automatically “creative wins”.
Documented retention definition
AppsFlyer Cohort and retention (source read 20 September 2026) defines Retention as on-day unique users who opened the app divided by conversions × 100. Organic users can be isolated with Media source is Organic. Prefer complete data when comparing finished cohort days so partial dashes do not masquerade as lifts.
Illustrative week
Event live Tuesday–Monday. Install cohorts Monday (pre) vs Wednesday (in-event):
| Cell | D1 retention pre-event | D1 retention in-event | Delta |
|---|---|---|---|
| Organic | 28.0% | 34.0% | +6.0pp |
| Paid network A | 26.0% | 33.0% | +7.0pp |
| Paid network B | 27.0% | 31.0% | +4.0pp |
Network A’s +7pp is only +1pp above the organic calendar lift. Crediting +7pp to creative without the organic row overstates UA. Figures are synthetic.
Facts vs limits
| Fact | Limit |
|---|---|
| Retention and organic filter are documented | Event taxonomy is studio-defined |
| Complete vs partial modes matter | This lab invents no genre retention norms |
| Same-week organic cell is the control | Cross-title benchmarks are out of scope |
