DEV Community

Swipe Again for Swipe Again

Posted on

Hard and Soft Declines Are Not Two Severities. They Are Two Different Jobs.

Every failed payment arrives wearing a label, and the most consequential label in subscription billing is the one that divides declines into hard and soft. Teams often treat the split as a severity scale — hard meaning bad, soft meaning less bad — and then apply the same recovery motion to both, just with more urgency for the hard ones. That reading wastes the information the label actually carries. The split is not about how serious the failure is. It is about what kind of work is even possible next.

A soft decline says the door is still open. Insufficient funds, a temporary issuer hold, a card that hit a velocity limit — in each case the credential itself is valid and the account behind it may be perfectly willing. The work here is timing and patience: wait for the payday cycle to turn, space the attempts so they read as care rather than desperation, and let the customer's own banking life resolve the obstacle. Retrying a soft decline is a scheduling problem, and most of the craft lies in not spending attempts before the conditions that caused the decline have had a chance to change.

A hard decline says the door itself is gone. The card is closed, reported lost, or the account no longer exists in that form. No schedule fixes this, because there is nothing at the other end to approve anything. The only possible work is replacement: the customer must supply a new credential, which means the recovery motion is communication, not retrying. Every automatic attempt fired at a hard decline is wasted motion, and worse, it can read to the issuer as the kind of persistent charging behavior that invites closer scrutiny of everything else you submit.

A worked morning makes the difference concrete. Two renewals fail in the same hour. The first returns a closed-account code: the card is dead, and the only honest next step is a message asking the customer for a new one, sent while the subscription is still warm. The second returns insufficient funds three days before a common payday: the honest next step is nothing at all until that payday passes, then one well-placed attempt. A severity-based program treats the first as the emergency and the second as routine. A job-based program knows the first needs a mailbox today and the second needs a calendar entry — and that swapping those two responses loses both customers.

The operational payoff of the split is that it sorts the queue before anyone touches it. Read the practical split between hard and soft declines and what each one allows next and the shape becomes clear: soft declines go to a calendar, hard declines go to a mailbox, and mixing the two queues is how teams burn retry budgets on dead cards while live ones wait behind them.

There is a grey band worth respecting. Some issuers return generic codes that hide which side of the line a failure sits on, and some hard declines arrive dressed as soft ones for the first attempt or two. The honest practice is probabilistic: a decline that behaves hard across a couple of well-spaced attempts should be promoted to the replacement queue rather than retried on hope. Hope is not a dunning strategy, and the attempt counter does not care how sincere it was.

Getting this one distinction right changes the economics of everything downstream. Retry budgets concentrate where approval is possible. Customer messages go only to people who can actually act on them. And the failure report stops being a single alarming number and becomes two manageable queues, each with its own craft. Most recovery programs do not fail for lack of effort. They fail because the effort was aimed by a severity scale when the labels were offering a map.

That sorting — reading each failure before acting on it, and sending it to the calendar queue or the replacement queue accordingly — is the daily work of Swipe Again for Stripe-based subscription stores, at a flat $179 a month, with recovered revenue reported and the arithmetic shown.

Top comments (0)