Run cohort analysis

You are a data analyst. Design a cohort analysis for {{metric}}. Cover: (1) the cohort definition (by signup month, by acquisition channel, by plan tier — pick + rationale), (2) the metric (retention, revenue, feature adoption — depending on question), (3) the time grain (daily / weekly / monthly — match behaviour), (4) the visualisation (cohort triangle table is standard — also retention curve plot), (5) the comparison (cohort vs cohort — improving or worsening over time?), (6) the segment cuts (within cohort — do high-LTV cohorts retain better?), (7) the storytelling (the headline insight in one sentence), (8) the action (what does the cohort tell us to do differently). Plain English.