What is dunning, and what is involuntary churn?
Dunning is the sequence of retries and reminders that recovers a failed subscription payment. It matters because two very different things get filed under churn. Voluntary churn is a decision — someone cancels. Involuntary churn is an accident — a card expired, a charge got declined, a bank blocked it. The usual retention and win-back playbook targets people who chose to leave and skips involuntary churn entirely. Before you build any of it, split the two: how you calculate churn should separate voluntary from involuntary, because you fix them in completely different ways.
How big is the leak?
Bigger than most founders expect, because nobody decided to leave. Baremetrics reports that subscription businesses lose roughly 9% of MRR to failed payments — on $50,000 MRR, that is about $4,500 a month walking out with no one clicking cancel. The share of total churn that is involuntary is often cited at 20–40%; Churnkey's benchmarks put the SaaS average near 22%. Treat these as directional, not gospel — your mix depends on plan price, card types, and geography. The point stands: a real slice of your churn is recoverable revenue, not lost customers.
Why do payments fail?
- Expired cards — the most preventable cause; the card on file simply aged out.
- Insufficient funds — a soft decline that often clears on a later retry; Churnkey reports it drives around 40% of failures.
- Fraud and 3DS blocks — the bank or network flags the charge or demands a re-authentication the customer never completes.
- Hard declines — closed accounts or reported-lost cards that no retry will fix; these need a new card, not another attempt.
Smart retries: schedule beats brute force
Retrying a declined card every hour mostly earns more declines and annoys the issuer. Timing is the lever. Stripe's Smart Retries pick attempt times from signals like the hour a card historically clears, and default to 8 attempts over about two weeks. The intuition is simple: an insufficient-funds decline clears after payday, so a retry two or three days out beats ten retries tonight. Whatever tool you use, retry on a schedule tied to when money lands, and cap the attempts.
Pre-dunning: warn before the card expires
The cheapest failed payment is the one that never happens. Pre-dunning acts before the charge fails. Two moves cover most of it. First, a card-account updater — a service from Visa and Mastercard, offered through Stripe and most processors — refreshes a card's new number or expiry automatically, so a reissued card never triggers a decline. Second, a heads-up email a week before a card on file expires, asking the customer to update it. This is the same discipline as a trial-expiration reminder: a small, timely nudge that prevents a silent lapse.
The dunning email sequence
Emails back up the retries — they matter most when a card genuinely needs the customer's action. Keep it to three messages, each with one clear ask and a one-click link to update payment. Do not stack six guilt-laden reminders; recovery per email drops fast after the first, and every extra send risks a spam complaint.
| Stage | When | The ask |
|---|---|---|
| Heads-up | On the first failed charge | Your payment did not go through; we will retry automatically — no action needed yet. |
| Retry in progress | Mid-sequence, after a retry fails | We tried again and it did not clear. Update your card to avoid interruption. |
| Final notice | Before access is cut | Last attempt failed. Update now or your plan pauses on the date shown. |
Why dunning emails are transactional, not marketing
A dunning email is triggered by an account event and asks for a specific action, which makes it transactional, not marketing. That distinction is not pedantic — it decides deliverability. Send dunning from your transactional stream, not the promotional one that gets throttled and filtered. Marketing suppression rules should not apply; a customer who unsubscribed from your newsletter still needs to hear that their card failed. If these emails land in spam, the whole recovery flow quietly breaks.
When should you stop gracefully?
After the final notice, stop emailing. A customer who has ignored three messages and several retries is not going to convert on the seventh nag — they are going to mark you as spam, which damages deliverability for everyone. End gracefully: downgrade to a free tier, pause the account, or move them into a proper win-back campaign weeks later. Fold dunning into your broader lifecycle marketing so recovery, retention, and re-engagement are one system, not three disconnected flows.