Do shipped features earn a return?
Release noise is easy to count. This audit looks for users who actually return after a feature ships, while keeping young cohorts explicitly incomplete.
event_date,user_id,feature,event_type9 sample rowsSample ready. Audit to compute cohort evidence.
No evidence classified yetThe audit will validate every row before grouping post-release uses by feature and user.
Return-use evidence
Valid rows0
Releases / uses0 / 0
Unique users0
Verified return0
One-and-done0
Incomplete0
Repeat users0
Parser errors0
verified return usemature, one-and-doneobservation window incomplete
Interpret carefully. Repeat use is observable evidence, not proof of product quality. Instrumentation gaps, shared accounts, seasonality, cohort selection, and the chosen windows can change the result.