Split churn into two kinds before doing anything. Involuntary churn (failed payments) is often a fifth to a third of the total and is the fastest to fix with retries and dunning emails. Voluntary churn (customers choosing to leave) takes longer: find out why they leave, then fix onboarding, value delivery, and the cancellation flow, in that order.
What AI did Suggested adding discount offers to the cancellation flow.
What the expert decided Limited discounts to accounts with real usage. For everyone else, a discount only delays churn.
How do I reduce churn in my SaaS?
Start with failed payments, because fixing them takes days and doesn't require changing the product. Then work on why customers choose to leave, which takes longer but matters more over time.
| Weeks | Focus | Type of churn |
|---|---|---|
| 1–2 | Measure churn properly; split voluntary and involuntary | Both |
| 2–4 | Payment retries, card updater, dunning emails | Involuntary |
| 3–6 | Exit surveys, churned-customer interviews, usage analysis | Voluntary |
| 5–10 | Fix onboarding and first-value gaps | Voluntary |
| 8–12 | Cancellation flow, at-risk outreach, win-back | Voluntary |
What's the difference between voluntary and involuntary churn?
Voluntary churn is when a customer decides to cancel. Involuntary churn is when they lose access because a payment fails (expired card, insufficient funds, bank decline) and isn't recovered.
How big involuntary churn is depends on the source and the business:
- In Recurly's July 2026 network data, involuntary churn makes up roughly a third of SaaS churn (Recurly).
- Churnkey's analysis puts the SaaS average at 22% of churn, and lower for B2B (16%) than B2C (24%) (Churnkey).
Either way, it's a large share of churn that customers never chose.
What are the quick wins for failed payments?
Turn on smart payment retries, card updater services, and a short dunning email sequence. These need little engineering if your billing provider supports them.
| Fix | What it does | Evidence |
|---|---|---|
| Smart retries | Retries failed charges at times more likely to succeed | Stripe Smart Retries alone recovered 51% of failed payments in Churnkey's data (Churnkey) |
| Card updater | Refreshes expired or replaced card details automatically | Offered by most major processors |
| Dunning emails | Asks customers to update payment details | Each email recovers a small additional share; the first email recovered 2.8% in Churnkey's data |
| In-app payment notice | Shows a banner or blocks access until payment is updated | Combined with retries and emails, Churnkey reports recovering 70% of detected involuntary churn (vendor data) |
Note: Churnkey is a retention vendor, so treat its recovery figures as indicative.
How do you find out why customers leave?
Use three sources and look for agreement between them: exit surveys, conversations with churned customers, and usage data before cancellation.
- Exit survey. One required question with 5–6 reasons, plus an optional comment.
- Interviews. Talk to 5–10 recently churned customers. Ask what they were trying to do, not why they left.
- Usage data. Compare the first 30 days of retained vs. churned customers. Which actions did retained customers take that churned ones didn't?
AI is useful here for clustering survey comments and summarizing interviews. A human should decide what the patterns mean.
What do teams most often misdiagnose as the cause of churn?
"Too expensive." It's one of the most common exit survey answers, and often means "I wasn't getting enough value to justify the price." Cutting prices rarely fixes that; helping customers reach value sooner usually does.
How should you improve onboarding to reduce churn?
Find the action that best predicts retention, then remove everything between sign-up and that action. Churn that shows up in month three often starts in week one.
- Identify the "activation" action from your usage comparison above.
- Cut sign-up steps that don't lead toward it.
- Trigger onboarding emails by behavior (what a user did or didn't do), not just by day.
- Have a human reach out to high-value accounts that haven't activated by a set point.
How should you design a cancellation flow?
Make cancelling easy, and use the flow to learn and offer genuine alternatives. Making it hard to cancel damages trust and may conflict with consumer protection rules in some markets.
A good flow:
- Asks for the reason (one question)
- Offers a relevant alternative based on that reason: pause for seasonal users, a smaller plan for price concerns, help for setup problems
- Confirms cancellation clearly, with no hidden steps
What does a realistic 90-day result look like?
Expect the biggest early gain from failed payment recovery, because it works within weeks. Changes to onboarding and value delivery take longer to show, since they affect customers who sign up after the change.
FAQ
What's a good churn rate for SaaS?
It depends on customer size and pricing, and published benchmarks disagree on both the numbers and whether they're monthly or annual. Compare yourself with businesses at a similar price point, and track your own trend month over month.
Should I offer discounts to customers who try to cancel?
Only when the reason is price and the customer is actually using the product. Discounts for customers who don't get value delay churn rather than prevent it.
How do I spot customers about to churn?
Watch for drops in usage, fewer active users on an account, unresolved support tickets, and failed payments.
Is churn a marketing problem or a product problem?
Usually both. Marketing sets expectations; the product has to meet them. Customers who signed up for the wrong reason churn fastest.
Can AI predict churn?
AI can score accounts by risk from usage patterns. People are still better at noticing context the data misses, such as a champion leaving the customer's company.