Open ten dunning software pages side by side and a pattern emerges within minutes. Every vendor leads with the message: prettier emails, smarter subject lines, a card update page with your logo on it. The retry engine, the part that actually decides whether the money comes back, gets a paragraph near the bottom, usually the same paragraph everywhere. This ordering is not an accident. Messages are easy to screenshot. Retry logic is not, so it is described in adjectives instead of mechanics.
That inversion matters because the two halves of the job recover very different amounts. A well written email with a working update link genuinely helps when a card was replaced or a customer simply missed the first notice. But the largest pool of failures never involves the customer at all. Soft declines clear on their own schedule when the retry lands after a payday, after a temporary hold lifts, after the network updater posts a fresh credential. No email, however warm, changes when a bank balance refills. Software that treats the message as the product will keep polishing the smaller half of the recovery.
So grade the retry first. Ask how attempts are scheduled: on a fixed interval the vendor chose, or against the condition that caused each decline? Ask what happens the moment a customer updates a card mid sequence. In too many systems the answer is nothing until the next scheduled attempt, which can sit days away while the updated card waits unused. Ask whether a replacement charge paid through a different route closes the case or gets counted twice. These questions sound unglamorous next to a template gallery. They are also the difference between a tool that reports activity and one that reports money.
Then grade the accounting, because dunning tools are unusually creative accountants. Some count any payment that eventually succeeds as recovered, including customers who would have returned on their own. Some quietly exclude cancellations from the denominator. Some price as a percentage of recovered revenue, which sounds aligned until you notice the percentage applies to recoveries the tool barely touched. A flat reading of a buyer's guide that grades dunning software on retry mechanics first is a useful corrective here: it works through the category feature by feature and keeps returning to the same test, which is what the software does in the days between the failure and the next attempt.
None of this makes the message layer worthless. Tone decides whether a billing email reads as a service or a threat, and a frightened customer cancels instead of updating. Branding matters for the same reason: a payment request that looks like it came from a stranger gets treated like one. The point is sequence. A tool should earn its keep in the retry queue first and in the inbox second, and any evaluation that starts with the inbox will reliably pick the wrong winner.
There is a simple field test. Take your last fifty failed renewals and sort them by what actually resolved them: a customer action, a later retry, an updater refresh, or nothing at all. The sort takes an afternoon and it tells you which half of a dunning product your store actually needs. Most stores find the retry column is longer than they expected. Buy for the column that is longest, and let the template gallery be the tiebreaker it deserves to be.
That ordering is how Swipe Again is built for Stripe-based subscription stores, at a flat $179 a month: the retry placed against each decline's condition, the card update firing the next attempt immediately, and the customer written to in the store's own name and colour, with recovery counted once and shown plainly.
Top comments (0)