Most recovery systems are explained in nouns: dashboards, workflows, journeys, engines. The nouns make the work sound larger and murkier than it is. Strip a functioning system down and three verbs carry the whole load. Something must notice a failure and understand it. Something must decide what happens next and when. And something must bring the money in or close the case honestly when it cannot. If any one of the three is missing, the other two do not compensate — they just operate more expensively in the dark.
Detection is more than receiving a failure event. A webhook that says a charge failed is table stakes; the useful detection layer reads the reason, checks whether the credential on file has changed since the last attempt, and looks at what has already been tried on this case. Without that reading step, the system treats a replaced card and a short balance as the same event, and every downstream decision inherits the error. Good detection is quiet work, but it is the work the whole outcome is priced on.
The retry step is where judgement lives. A schedule printed on a settings page — retry on these days, at these hours — is a starting point, not a decision. The decision is per case: does anything about this failure suggest the next attempt should wait, go now, or never be spent at all because the subscriber first owes an action only they can take? Systems that separate the ask from the attempt get this right more often. When a card must be updated, the email is the recovery action and the retry is what happens after it works. That ordering is the substance of how recovery actually runs when detection, retrying, and the subscriber's own step are kept in the right sequence.
Recovery, the third verb, is a definition before it is a result. A case is recovered when the failed dollars collect — by a retry, by the subscriber updating a card and the charge then succeeding, or by a replacement payment for the same subscription and amount. It is not recovered when a subscriber promises, clicks, or reopens an email. Keeping the definition at the money keeps every report honest, including the ones used to judge a guarantee. It also clarifies the losses: a case that ends in cancellation during recovery counts plainly as a loss, not as an exit from the statistics.
Notice what the three verbs leave out. There is no fourth verb for persuading, chasing, or engaging. Subscribers in a failure case are not an audience to be warmed; they are people mid-transaction who need one clear fact and one clear action. The messages inside a recovery system earn their place only when they serve the retry decision — a card ask, a timing note, a final notice that means what it says. Everything else is decoration that costs deliverability.
When you evaluate any system, including your own platform's built-in tools, grade it verb by verb. Show me how a failure is read. Show me how the next step is chosen for this case rather than for the average case. Show me how a recovery is defined and counted. A system that answers all three plainly will outperform a larger one that answers them in nouns.
The verbs also give a team a shared language for the weekly review. Instead of debating a recovery rate in the abstract, walk the cases in order: which failures were misread at detection, which retries were spent without a reason, and which recoveries were claimed before the money landed. Most weeks, the losses cluster in one verb, and that cluster — not the headline rate — tells the team what to fix next.
That three-verb discipline is the design Swipe Again is built around: detect and read each failure, retry or ask according to the reason, and count a recovery only when the money actually lands.
Top comments (0)