mcpField
R
retention-cohort-review

retention-cohort-review

github.com/sylphxaiData & databases150 installsUpdated today
Repository

About retention-cohort-review

A skill for analyzing product retention cohorts and turning chart patterns into diagnoses, measurement fixes, and experiments. It is for product teams evaluating activation, repeat use, renewal, and other retention decisions, with guidance on cohort comparisons and their caveats.

Key features

  1. 01Defines product loop and cohort
  2. 02Checks metric against decisions
  3. 03Separates causes of retention changes
  4. 04Specifies cohort cuts and event fixes
  5. 05Produces hypotheses and experiment plans

Use cases

  1. 01Diagnosing weak subscription renewal
  2. 02Reviewing repeat use by acquisition source
  3. 03Finding onboarding or product-value gaps

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/sylphxai/skills --skill retention-cohort-review

After installing, type/retention-cohort-reviewin Claude Code to confirm it loaded.

Bundled filesagents/openai.yamlreferences/retention-cohort-systems.md

Details

From the project README

Retention Cohort Review

Use this skill to turn retention charts into product decisions instead of dashboard theater.

Workflow

  1. Identify product loop, activation event, cohort definition, time grain, segment, and business model.
  2. Read references/retention-cohort-systems.md.
  3. Check whether the metric answers a decision: activation, habit, subscription renewal, content return, game progression, marketplace repeat, or developer production use.
  4. Separate measurement errors, segment mix, acquisition quality, onboarding gaps, value gaps, lifecycle messaging, and monetization effects.
  5. Produce a diagnosis, cohort cuts, event fixes, product hypotheses, and experiment plan.

When not to use

  • Do not use for generic advice the base model already handles without this skill's specific artifact contract.

Guardrails

  • Do not compare cohorts without controlling for acquisition source, platform, geography, version, seasonality, and activation state.
  • Do not optimize short-term return if it increases fatigue, refunds, churn, or low-quality usage.
  • Do not treat signup as activation unless signup is the product value.

Output format

Retention question:
Cohort definition:

Diagnosis:
- signal, evidence, caveat

Cohort cuts needed:
- <segment> -> why

Action plan:
- product change, metric, expected movement, guardrail

Instrumentation gaps:
- <event/property>

Comments

Markdown supported