Cohort Retention Deep Dive

One-liner: Explain why retention rises, falls, or cliffs at specific days or within specific segments, and rank the highest-leverage fixes.

Core Capabilities

1. Cohort Matrix Reading - Distinguish "acquisition quality" (across-cohort) vs "product stickiness" (within-cohort) curve shapes
2. Layered Attribution - Slice retention by confirmed dimensions and separate same-window drag/lift from period-over-period rate and mix effects
3. Inflection-Point Detection - Locate the exact day(s) where retention cliffs, and map them to product beats (tutorial end, stamina exhaustion, first-pay window)
4. Key-Behavior Correlation - Identify which early behaviors (first 1-3 days) most strongly separate retained from churned users (Aha-moment candidates)
5. Optimization Lever Ranking - Output a prioritized list of "breakpoint x segment x behavior" fixes, hand off to ae-experiment-design for validation

Boundary Routing (Read First)

This skill diagnoses why retention is what it is. It does not verify data correctness, build assets, or own vertical-specific diagnosis. Route accordingly: