Churn Prevention System Builder

What it does

Produces concrete systems to reduce SaaS subscription churn on two separate fronts — not vague advice, but copy-ready flows + copy + a measurement plan. It splits churn into voluntary (customer chooses to cancel, typically 50-70% of total) and involuntary (payment fails, 20-40% but the most recoverable).

It runs in three modes: (1) build a cancel flow from scratch, (2) optimize an existing flow, (3) set up dunning. The cancel flow follows the canonical sequence Trigger → Exit Survey → Dynamic Offer → Confirmation → Post-Cancel. The core insight: match the offer to the reason. "Too expensive" → discount/downgrade; "not using it" → pause/onboarding; "missing feature" → roadmap. A blanket discount fails most reasons and trains customers to cancel-for-deals.

On the proactive side it builds a health-score model (login 30% + feature 25% + support 15% + billing 15% + engagement 15%) mapped to Healthy/Needs-attention/At-risk/Critical actions, triggering intervention before "Cancel" is ever clicked. For involuntary churn it produces a dunning stack: pre-dunning (card-expiry alerts, card-updater), a decline-type smart-retry schedule (soft decline 3-5 retries / hard decline no retry), and a 4-step email sequence (Day 0 friendly → Day 10 final notice).

When to use it

• Churn is high and you don't know why customers leave
• Cancellation is instant with no save mechanism — leaving money on the table
• Failed payments silently cancel accounts (no dunning)
• An existing cancel flow has a low save rate and needs offer-reason tuning
• You need an FTC click-to-cancel-compliant flow with no dark patterns

Method / frameworks

• Voluntary/Involuntary split — root-cause segmentation driving two distinct strategies
• Cancel Flow sequence — Trigger→Survey→Dynamic Offer→Confirmation→Post-Cancel
• Offer-to-Reason matrix — primary + fallback offer per reason (discount 20-30%/2-3mo sweet spot)
• Weighted Health-Score model — 5-signal 0-100 score with risk-band actions
• Dunning Stack — pre-dunning → smart retry → dunning email → grace → hard cancel
• Smart-retry decline typology — soft/hard/auth-required distinction
• Recurly/Stripe benchmarks — save rate 25-35%, dunning recovery 50-70%, involuntary 20-40% share
• Cohort + A/B test plan — statistically rigorous cancel-flow experiments via the ab-testing skill