# Cohort Analysis

> Plan or interpret cohort analyses for users, customers, accounts, revenue, retention, activation, or behavior. Use when Codex is asked to define cohorts, compare groups over time, identify cohort effects, or write cohort analysis memos.

- Skill: `jeremylongworth-source/cohort-analysis` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add jeremylongworth-source/cohort-analysis`
- Raw SKILL.md: https://api.skillmd.com/api/skills/jeremylongworth-source/cohort-analysis/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Data & Analytics
- License: MIT
- Author: jeremylongworth-source (https://skillmd.com/u/jeremylongworth-source)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/jeremylongworth-source/cohort-analysis

---


# Cohort Analysis

## Core Workflow

1. Identify question, cohort key, cohort start event, measurement window,
   metric, and segment.
2. Define inclusion/exclusion rules, data grain, time buckets, and censoring.
3. Separate cohort effects from calendar effects, seasonality, product changes,
   and acquisition mix.
4. Specify table/chart shape and comparison method.
5. Note data quality caveats and confidence level.
6. Convert findings into decision options or follow-up analysis.

## Safety Rules

- Do not infer causality from cohort differences without supporting evidence.
- Do not invent cohort sizes, time windows, or source fields.
- Escalate analysis used for finance, investor reporting, compensation,
  compliance, or customer commitments.

## Deliverable Shape

For cohort analyses, provide:

- Question and cohort definition
- Metric and time window
- Inclusion/exclusion rules
- Analysis table or chart plan
- Findings or expected comparisons
- Caveats and confidence
- Decision implications
- Follow-up checks

## References

- Read `references/cohort-analysis-checklist.md` when planning or reviewing
  cohort analysis.