Braintree merchants sit on an asset many of them never inspect: the vault. Cards get replaced by issuers, account updater services refresh credentials, and the vault quietly holds newer payment data than the subscription's last successful charge used. When a renewal fails on a stale credential while the vault holds a live one, the failure is not a customer problem. It is a plumbing oversight.
The oversight compounds because renewal failures look uniform in a report. Expired, superseded, and genuinely dead credentials arrive wearing similar decline codes, and a retry schedule that cannot tell them apart spends its attempts evenly across the living and the dead. Reading the failure against the vault's current state is the sort that turns a pile into a queue.
The fresh read includes the credential itself
Good dunning discipline already insists on one habit: before every attempt, re-read the invoice fresh, because the world moves between failure and retry — a payment settles, an amount changes, a native attempt runs first. On Braintree, that same habit extends naturally to the credential attached to the invoice. What does the vault hold for this customer right now, today, before this attempt fires — the same card that just failed, a replacement issued since, a credential refreshed by the account updater, or nothing usable at all? Each answer dictates a different action, and none of them is "retry the stale card and hope." If newer payment data is sitting in the vault, the attempt belongs on that data, once, cleanly recorded. If the vault holds only the credential that already failed hard, the correct move is not another attempt at all. The fresh read is cheap. The attempts it prevents, and the attempts it redirects onto live data, are where its value sits.
Sorting the queue the vault reveals
Once failures are read against current vault state, the familiar sorting can do its work with better information. Hard declines stop immediately and route to a single clean card-update request, because a dead credential in the vault confirms what the decline already said: there is nothing here to retry. Soft declines are re-read and re-spaced, treated as timing rather than refusal, with the next attempt placed patiently instead of fired in a burst. Insufficient funds, which says nothing about the credential and everything about the calendar, rides a payday ladder out toward day 25 rather than consuming its attempts in the first week, when the account it draws on is in the same state that just failed. The vault does not replace that sorting; it sharpens it. A decline code tells you what the issuer said. The vault tells you what you actually hold. A queue built from both answers spends its attempts where a live credential and a plausible moment coincide, and closes the rest without ceremony.
Weave, cap, and keep the customer out of the machinery
Braintree's own native retry behaviour is already running underneath all of this, and a recovery layer that ignores it creates two rhythms on one card. The disciplined pattern is to weave outside attempts into the gaps of the platform's native schedule, under a combined cap of 15 attempts in total, with Swipe Again's own ceiling at 8. The customer, meanwhile, should never see any of this plumbing. What they see, when they need to see anything, is a short sequence in the merchant's own brand: email on day 1, day 3, and day 7 pointing to a card-update page, with SMS where the merchant's program allows it. If they later choose to cancel, that is a separate moment with its own honest shape: one branded save offer at the cancel click, with exit one click away. Behind the scenes, recovery attribution is stamped at a single choke point — retry first, then card update, then save offer, then honest external — so a renewal recovered because the vault held a live credential is credited once, to the step that did it, and never double-counted by the email that happened to precede it.
A field guide to Braintree dunning, the account updater's role, the decline codes that mean "look in the vault first," and the ones that mean stop, is the difference between retrying a subscription and actually recovering it. Braintree's route to market runs through the PayPal partner program, which is its own lesson in how this platform expects to be extended: deliberately.
Published by Swipe Again - failed-payment recovery for subscription businesses.
Top comments (0)