| name | churn-prevention |
| description | Use when the user wants to reduce customer churn, improve retention, design a cancellation flow, or build a win-back strategy. Also use when the user mentions 'churn rate', 'customer health score', 'dunning', 'cancellation flow', 'win-back', or 'voluntary churn'. |
Churn Prevention
Expert knowledge for diagnosing churn, building health scores, designing cancellation flows, and recovering lost customers.
Churn Taxonomy
Voluntary churn — customer actively cancels. Indicates perceived value failure or better alternative found.
Involuntary churn — payment fails, not recovered. 20-40% of churn in most SaaS is involuntary. Entirely preventable with good dunning.
Silent churn — customer stops using the product but doesn't cancel (common in freemium / annual plans). Dangerous because it's invisible until renewal. Detect via login frequency and feature usage.
Churn rate formula:
Monthly Churn Rate = Customers Lost This Month / Customers at Start of Month
Annual Churn Rate ≈ Monthly Rate × 12 (approx — compound for precision)
Benchmarks (B2B SaaS): < 5% annual = healthy. 5–10% = watch. > 10% = priority problem.
Customer Health Score
A leading indicator of churn risk. Compute weekly per account.
Formula:
Health Score = (Login Frequency × 0.3)
+ (Feature Usage Depth × 0.4)
+ (Expansion Signal × 0.2)
+ (Support Sentiment × 0.1)
Signal definitions:
- Login Frequency: Sessions in last 30 days / expected sessions (based on plan type)
- Feature Usage Depth: Core features used / total core features available
- Expansion Signal: Has added users, upgraded, or integrated in last 90 days (binary: 1 or 0)
- Support Sentiment: Recent ticket sentiment score (positive = 1.0, neutral = 0.5, negative = 0.0)
Score thresholds:
- 0.7–1.0: Healthy
- 0.4–0.69: At Risk — trigger proactive outreach
- 0–0.39: Critical — escalate to CSM or automated save sequence
Adapt weights to your product. If login is a weak signal (e.g., API-first product), increase feature usage weight.
Cancellation Flow Architecture
A good cancellation flow has one job: surface the right save offer at the right moment.
Step 1 — Pause option
Before asking for a reason, offer a pause (1–3 months). Resolves 15-25% of cancellations from customers who are temporarily budget-constrained.
Step 2 — Reason collection
Ask for one reason (not a long survey). Use it to route to the right save offer.
| Reason | Save Offer |
|---|
| Too expensive | Discount (20–30%) or downgrade |
| Not using it enough | Usage coaching / feature tour |
| Missing feature | Roadmap preview + feature request |
| Switching to competitor | Direct comparison + migration help |
| Business shutting down | Acknowledge + offboard gracefully |
Step 3 — Save offer
One specific offer, matched to the stated reason. Never offer a generic "here's a discount" — it signals you were overcharging.
Step 4 — Confirm cancellation
If they proceed: make it one click. A difficult cancellation creates a negative brand moment that gets shared.
Dunning Sequence (Involuntary Churn)
| Day | Action | Channel |
|---|
| 0 | Payment failed — retry immediately (often transient) | Auto |
| 1 | Notify customer: payment failed, update card | Email |
| 3 | Retry payment | Auto |
| 7 | Reminder email: account at risk of suspension | Email |
| 10 | Final retry | Auto |
| 20 | Account suspended (access revoked, data retained) | Email |
| 30 | Account scheduled for deletion | Email |
Smart retry: Retry on different days of the week — many card failures are temporary credit limit or bank-hold issues.
Offer: On Day 7 email, include a one-click link to update payment. Remove every possible friction from the recovery path.
Win-Back Strategy
Timing: Wait 60–90 days after cancellation before win-back outreach. Too soon feels desperate. After 90 days, they've often committed to an alternative.
Win-back sequence (3 emails over 2 weeks):
- Day 60: "What brought you back" curiosity email — no hard sell. Ask what would make you reconsider.
- Day 67: New feature announcement relevant to their cancellation reason.
- Day 74: One-time offer (discount or extended trial) with clear expiration.
Segmentation: Customize by cancellation reason. A customer who left for a competitor needs a different message than one who left for budget reasons.
Never win back: Customers who cancelled due to a bad support experience without first resolving the support issue.
NPS and Churn Correlation
NPS detractors (0–6) churn at 3–5× the rate of promoters (9–10).
Action by segment:
- Promoters (9–10): Ask for referrals, case studies, G2 reviews
- Passives (7–8): Identify the one thing that would move them to promoter and address it
- Detractors (0–6): Assign to CSM within 24 hours. Understand root cause. Detractor rescue before churn, not after.
Common Rationalizations
| Rationalization | Reality |
|---|
| "Churn is normal, some customers just leave" | Voluntary churn above 5% annual signals a product or value-delivery problem, not natural attrition. |
| "We shouldn't make cancellation hard" | Easy cancellation ≠ no save attempt. A good save offer is helpful, not friction. |
| "Win-back doesn't work" | Win-back to previously happy customers converts at 20–40%. Ignore it at your expense. |
| "Involuntary churn isn't our fault" | It's entirely recoverable with dunning. Accepting it is choosing to lose revenue. |
Verification