L’interface est localisée ; les présentations et documents des Skills sont traduits automatiquement par IA dans votre langue. Le code, les commandes, chemins et termes techniques restent en version originale, et le document de l’auteur fait foi.
Explorer les Skills
Design et création · SKILL.md

churn-prevention

churn-prevention

Design cancel flows, save offers, dunning, and failed-payment recovery to reduce involuntary and voluntary churn.

Independent GitHub snapshot · Automated classification · Original documentation governs

Documentation de la Skill

GitHub

name: churn-prevention description: "Design cancel flows, save offers, dunning, and failed-payment recovery to reduce involuntary and voluntary churn."

Churn Prevention

You design programs that reduce churn — both voluntary (cancellations) and involuntary (failed payments). Read product-marketing-context first.

Voluntary churn levers

  • Cancel flow — ask why first (radio button, max 5 reasons), match the response to a save offer.
  • Save offers by reason:
    • "Too expensive" → 1-2 months discount, downgrade to cheaper plan.
    • "Not using it" → pause subscription, win-back schedule, success-team intro.
    • "Missing feature" → roadmap visibility, beta invite, manual workaround.
    • "Switched to competitor" → no save offer; ask for honest feedback and let them go.
  • Frictionless cancel — never hide the cancel button. Trust > short-term retention.
  • Pause > cancel — offer 1-3 month pause as the default save.

Involuntary churn levers (dunning)

  • 4-email cadence over 21 days: D0 retry + email, D3 retry + email, D7 retry + email + in-app banner, D14 last chance, D21 cancellation notice.
  • Update-card link in every email.
  • Smart retries (Stripe Smart Retries / network-tokenized retries).
  • Card-expiry preempt: warn 30 days before expiry.

Metrics

  • Voluntary churn rate (canceled / paying customers).
  • Involuntary churn rate (failed-payment-driven).
  • Save rate (saves / cancellation attempts) and save value (MRR retained / MRR at risk).

Output

  • Cancel-flow wireframe with question tree and offer logic.
  • Dunning email sequence (hand off to email-sequence).
  • A test plan for save offers (ab-test-setup). Adapted from coreyhaines31/marketingskills (MIT).
Skills enregistrées