DEV Community

Swipe Again
Swipe Again

Posted on

Recurly Keeps Good Records. The Gaps Live Between Them.

Recurly is, among billing platforms, one of the tidier record-keepers. The invoice history is legible, the retry attempts are logged, the dunning campaign leaves a trail a finance person can follow without a flashlight. That tidiness creates a particular blind spot: when the records look complete, it is easy to assume the recovery is complete too. It usually is not. The gaps live between the records — in the hours when nobody is watching the case, and in the facts that changed outside Recurly entirely.

Consider the most ordinary gap. A renewal fails on Monday. On Tuesday the customer, prompted by nothing more sinister than a banking-app notification, updates their card inside your portal. On Wednesday Recurly's schedule fires its next attempt — at the old arrangement, or into a case that no longer needs one, depending on how the pieces were wired. A tidy log will record that attempt faithfully. It will not record that the attempt was unnecessary, because from inside the log there was never anything wrong. The discipline that prevents it is a check performed outside the schedule: before any attempt spends itself, confirm the world still looks the way it did when the attempt was queued.

A second gap opens around sibling charges. Subscriptions are messy in ways invoices are not: a customer retries checkout manually, a support agent re-bills from another screen, a plan change mints a fresh charge for the same period. Any one of those can collect the money while the original failed case sits open, still counting down toward its next retry and, eventually, its cancellation step. Closing that gap takes a rule, stated plainly: if the same member, on the same subscription, for the same amount, has paid under a new charge, the original case is recovered — recovered by replacement — and it closes. Without the rule, the operation eventually double-collects from its most cooperative customers, which is the worst possible audience for that mistake.

Timing is the third gap, and it is quieter than the other two. Recurly will retry on the cadence it is given, and most teams give it the cadence that shipped. But a decline for insufficient funds is a statement about a date, not about a person. The same account that cannot pay on the second can often pay on the sixteenth without anything else changing. A recovery operation earns its keep by aiming attempts at those dates — after common paydays, clear of the weekend settlement drag — instead of spacing them evenly out of tidiness. Even spacing is a property of calendars, not of customers.

The last gap is the ending itself. Records make it easy to see attempts; they make it harder to insist that every case close with a reason. Recovered is one reason. Paused alongside a paused subscription is another, and it belongs off the scoreboard entirely. Honestly lost, after a sensible number of well-timed attempts, is a legitimate ending that too many queues refuse to declare, so their open-case counts swell and their recovery rates sag for accounting reasons rather than operational ones. A winback handoff is an ending too, provided it is a handoff and not a euphemism for forgetting.

For teams running their recovery on this platform, a plain guide to failed payment recovery for Recurly and the checks that belong around its native retries lays out the pre-attempt checks, the replacement-charge rule, and the closing discipline in working order.

Keep the records; they are genuinely good. Just refuse to mistake them for the operation. The operation is the set of decisions made between one logged attempt and the next — and decisions, unlike logs, do not make themselves.

Those between-the-records decisions are the whole design of Swipe Again: live state checked before every attempt, sibling collections recognised and closed as recoveries, retries aimed at paydays, and every case closed with its reason on the record.

Top comments (0)