Decide whether to launch Group Story

A proposed Group Story feature could divert activity away from regular stories while still increasing overall engagement. Outline the experiment and decision framework: 1) Specify success metrics and guardrails: feature adoption, total session time, regular-story posts, creator activity, and crash rate. 2) Handle network effects: choose between user-cluster randomization (such as communities) and switchback at the feed level; justify your choice to minimize interference. 3) For powering heavy-tailed session lengths: select among variance-stabilizing transforms, winsorization, or quantile treatment effects; justify. 4) Quantify cannibalization versus net new creation; define incremental lift decomposition. 5) Duration and ramp strategy with pre-period CUPED. 6) Heterogeneity analysis by cohort (new vs existing creators) and by group size; specify how results would change your launch/no-launch decision and what holdout you would keep for long-term effects.

Overview: This question assesses a data scientist's ability in experimental design and causal inference, including defining metrics and guardrails, handling network interference and randomization, statistically treating heavy-tailed engagement metrics, planning power and duration, decomposing incremental lift to separate cannibalization, and analyzing cohort heterogeneity. It appears frequently in Analytics and Experimentation interviews because it tests both conceptual understanding of causal and interference problems and practical use of statistical experiment methods at product-launch scale.