A webhook is a promise a platform makes in the moment, and like most promises made in a hurry, it occasionally does not arrive. Endpoints time out, deploys eat events, a retry queue gives up quietly at 3 a.m. None of this is malice. All of it means the same thing for a system built on events alone: your picture of reality has holes, and the holes are exactly where the weird failures live.
Recovery cannot afford holes, because a missed failure event is a subscriber nobody tried to save. So Swipe Again treats events as the fast path, never the only path: alongside live ingestion, reconciliation sweeps re-read the platform's own records and diff them against ours, and a ninety-day backfill at connect time establishes ground truth before the first live event ever fires. Anything the webhooks dropped, the sweep finds — late, but found, which in this business is the difference between a save and an apology.
The design principle generalizes past payments: events for speed, reconciliation for truth. Any system whose errors are invisible by construction will eventually present its operator with a dashboard that is calm, green, and wrong. The recovery flow assumes its own inputs lie by omission and budgets a second source of truth accordingly. Boring, redundant, and the reason the numbers can be trusted.
Top comments (0)