Stripe's built-in recovery tools are genuinely good at the first pass. A failed charge is retried on a schedule informed by a lot of network data, card update services quietly replace expired numbers, and a hosted payment page gives the customer somewhere to fix their details. For a small subscription business, that first pass may be most of what ever happens. The interesting question is not whether that machinery works. It is what your business does on the morning it does not.
Picture the case that remains after the automatic attempts are spent. The customer has not cancelled. They have not replied. The card on file has failed for a reason the bank described in two vague words, and the subscription is drifting toward the point where your platform will simply give up on it. Nothing about that case is unusual, which is exactly why it is dangerous: it looks like background noise, so nobody owns it.
Owning it starts with reading the decline properly. A soft decline at a business that bills monthly is often a timing story — the attempt landed before payday, or on a card with a daily limit that a rent payment had already used. A hard decline is a different story entirely, and spending more retries on it does not make the story change. The discipline is to sort the leftovers before acting on them, because the right next step for one pile is patience, for another it is a clear email, and for a third it is a respectful stop.
That sorting is the standard behind stripe dunning software judged by what happens after the first smart retry fails: the built-in tools handle the easy middle of the distribution, and the added software earns its place on the tails — the cases that need judgement about timing, wording, and when to stop, rather than another identical attempt fired on a fixed clock.
The email side deserves the same scrutiny. After several silent failures, a customer does not need a louder subject line; they need to understand, in one reading, what was charged, why it did not go through, and what single action fixes it. The best recovery emails read like a short note from a person who checked the account first. The worst read like a countdown. Both kinds get sent by software. Only one kind gets replies that begin with an apology and end with an updated card.
There is also a quieter failure mode to watch: the customer who fixes the card somewhere else. They update their details through your app settings, or on a different subscription entirely, and the new card never reaches the failed case because nothing connected the two events. A recovery system that only watches its own email link will sit waiting while the answer is already sitting in the account. Closing that gap — treating a card change anywhere as a signal to try the waiting case again — recovers money that no amount of better copywriting can reach.
None of this argues against using Stripe's own tools. It argues for knowing precisely where they end. Draw that line on paper for your own account: which attempts are automatic, which messages are native, at what point a subscription is cancelled for non-payment, and who — person or system — looks at what is left. If the last answer is nobody, that is the gap to buy software for, priced against the value of the cases actually sitting there.
That is the working premise at Swipe Again — start where the automatic pass ends, read each remaining case on its own evidence, and let the record of what worked, case by case, set the next attempt.
Top comments (0)