Search for a dunning solution and you will be offered at least four different products under the same label. One is a retry engine: it re-attempts failed charges on a smarter schedule. One is a messaging tool: it sends the emails and texts that ask customers to fix their cards. One is a card-update page: a hosted, branded form that collects a new payment method. And one is a service layer: people and systems that work the backlog, classify failures, and report what came back. Vendors blur these together because the blur sells. Buyers should unblur them, because each product solves a different failure and fails at the others.
The retry engine solves exactly one problem: a charge that would succeed if attempted at a better moment. That is a real problem — timing failures are common — but it is a minority of any queue. A retry engine pointed at a replaced card will fail politely, repeatedly, until the subscription cancels. The messaging tool solves the awareness problem: the customer who never knew the payment failed. It does nothing for the customer who knew, clicked, and found a confusing update form. The hosted page solves that friction, and does nothing for the customer who never opens the message. Each layer is necessary for somebody and sufficient for nobody.
This is why a sober comparison has to start from your own failure mix rather than from feature lists. Pull a month of failed payments and sort them by what would have fixed them: a better moment, a new card, a clearer ask, a second chance weeks later. Most merchants who do this exercise find all four piles are non-trivial, which is the real finding — the category's honest answer is a stack, not a product. A useful map of what dunning solutions actually cover, what they cost, and how the layers differ is one that keeps the four products visibly separate and prices each on its own terms.
Cost comparisons deserve the same unblurring. Percentage-of-recovery pricing attaches to whichever layer the vendor owns and grows with your best months. Flat pricing reads higher in a quiet month and lower in a strong one, and has the virtue of being known in advance. Per-seat and per-email pricing fragment the bill across the stack until nobody can state what recovery costs end to end. Whichever model you choose, insist on being able to answer one question in one number: what did we pay, in total, per recovered subscription last month? A vendor who cannot help you answer it is selling you a layer and calling it an outcome.
A practical way to run the unblurring is a four-column sheet. Label the columns Moment, Card, Ask, and Residue. Walk one month of failures down the sheet, placing each in the column of the product that would have fixed it, and total the revenue in each column. The sheet takes an afternoon and ends most buying arguments, because it replaces opinions about vendors with the shape of your own losses — and it shows, in your numbers rather than a salesperson's, which layers you can defer and which you cannot.
There is also a fifth consideration the feature lists rarely mention: who works the failures that survive the automation. Every stack produces a residue — customers who need a human-toned follow-up, a re-approach after a payday cycle, a save attempt tied to a cancellation. In some stacks that residue is yours, worked by whoever has spare time, which usually means nobody. In others it is the service itself.
For Stripe-based stores, that full-stack shape is what Swipe Again sells at a flat $179 a month: retries timed by failure type, update requests in the merchant's own branding, the ninety-day backlog worked instead of forgotten, and recovered revenue reported with its arithmetic visible. Whether you buy it or assemble the stack yourself, buy it unblurred — four products, four prices, one honest total.
Top comments (0)