Designing a Payment Retry Schedule That Recovers More
How many retries, how far apart, and which declines to skip entirely. Retry design that recovers revenue without annoying customers or upsetting card networks.
- Day 0Renewal failsRecord the decline reason
- Day 2Retry + email 1Helpful: “this happens”
- Day 6Retry + email 2Firmer, still helpful
- Day 13Final attemptSay plainly what happens next
- AfterPause, don’t deleteRecovery stays one click away
Retry logic is the least glamorous part of a subscription business and one of the most consequential. Good retry design recovers a substantial share of failed renewals. Bad design irritates customers and can attract network-level penalties.
The default that works
Four attempts over roughly two weeks:
| Attempt | When | Paired email |
|---|---|---|
| 1 | Day 0 — the scheduled charge | — |
| 2 | Day 2 | Email 1: helpful notice |
| 3 | Day 6 | Email 2: firmer |
| 4 | Day 13 | Email 3: final attempt notice |
Then a defined outcome: pause or cancel, with the customer told which.
Beyond four attempts the recovery curve is essentially flat, and continued retries increase your decline ratio — which some processors and networks care about.
Skip retries that cannot succeed
This is the highest-value refinement and the one most commonly missed.
Retry these: insufficient funds, do-not-honour, issuer temporarily unavailable, generic processing errors. The condition may change.
Do not retry these: expired card, invalid card number, card reported lost or stolen, closed account. Nothing about a second attempt changes the outcome.
For the second group, go straight to the email. Retrying an expired card three times delays the fix by two weeks and teaches the customer that your system is broken.
Time around pay cycles
Insufficient-funds declines cluster before payday and clear after it.
If your subscribers are salaried and paid monthly a retry scheduled for the 1st or 2nd recovers materially better than the same retry on the 28th.
A simple version of this: if attempt 1 falls in the last week of the month, schedule attempt 2 for the 1st rather than at a fixed +2 day offset.
Do not retry everything at midnight
Two problems with a single nightly batch:
Gateway rate limits. A burst of thousands of charges can produce throttling errors that look like declines and pollute your data.
Concentrated support load. Every dunning email goes out simultaneously, so every reply arrives simultaneously.
Spreading retries across a window smooths both.
Authentication-aware retries
Some declines are not about money at all. In Europe an issuer can respond to a renewal by asking the customer to authenticate it, and no retry schedule fixes that — the customer has to act.
A retry system that treats authentication requests like funds failures burns its sequence on charges that cannot succeed without the customer, then cancels someone who would happily have confirmed if asked. At minimum, record authentication requests as their own outcome and put an authentication link in front of the customer. SCA and subscriptions.
What to do at the end
Pick one and be consistent:
Pause the subscription. Contract survives, customer can resume in one click, reporting stays clean if you count paused separately. Best default for consumer subscriptions.
Cancel with easy resubscribe. Cleaner, slightly lower recovery.
Never: leave it retrying indefinitely.
Instrument it
Record, for every failed renewal: the decline code, each attempt with its outcome, which emails were sent and opened, and the final state.
Without that you cannot tell whether a low recovery rate is a retry-timing problem or an email problem, and those have opposite fixes.
Super Subscription retries failed renewals on a fixed schedule, sends the customer a payment-method update email, and records renewals that need the customer to authenticate as their own outcome rather than as ordinary declines.
Part of the guide to Failed Subscription Payments: Recovering the Revenue.