Every recovery tool demos beautifully. The dashboard is clean, the retry chart climbs at a satisfying angle, and the sample data — always sample data — shows a business rescuing a third of its failed payments without lifting a finger. None of this is dishonest, exactly. It is just measured on the one day when nothing can go wrong: a day with no real declines in it, no expired cards belonging to real people, no subscriber who replies to a dunning email at 11 p.m. asking a question nobody scripted.
The honest test of recovery software starts where the demo ends, with your own failure file. Take last month's failed payments — the actual list, with its actual decline codes and its actual awkward cases — and ask the tool's documentation, not its sales page, what happens to each class. The soft declines are easy; everyone handles those. The interesting answers concern the rest. What happens when a card is replaced mid-sequence: does the sequence notice, or does it keep writing polite emails about a card that no longer exists? What happens when a subscriber cancels halfway through: does the outreach stop that minute, or does one more message go out because the schedule was set on Monday? What happens when the platform's own retry succeeds on day nine: does the tool stand down, or does it send a success email claiming a rescue it did not perform?
A useful reference for structuring that interrogation is a buyer's checklist for payment recovery software that starts from your failure file — its useful habit is organising the evaluation around case classes rather than feature lists, because features are what vendors choose to show you and case classes are what your subscribers will actually generate.
There is a second test the demo cannot fake, and it is about money rather than mechanics. Recovery pricing usually arrives as a percentage of recovered revenue, which sounds aligned until you price a quiet month. In a quiet month the tool recovers little, so the percentage costs little — fair enough. In a loud month, after a price rise or a card-network wobble pushes failures up, the same percentage bills you for a problem that got worse. Flat pricing inverts the relationship: the fee is known in advance, boring in the best sense, and the vendor's incentive is to keep you subscribed rather than to meter your bad luck. Neither model is automatically right, but the arithmetic belongs in the evaluation, run against your loudest historical month rather than your average one.
Third, watch how the tool counts. Every vendor reports a recovery rate; almost none volunteer the denominator unprompted. Recovered out of all failures is one number. Recovered out of failures the tool touched is a larger number. Recovered out of failures still open when the report ran is a larger one again — and all three will be described in meetings as "the recovery rate" by people who are not lying, just repeating whichever figure reached them first. Ask which population the headline rate uses, and ask whether platform retries, subscriber self-fixes and your own manual saves are inside it. A vendor who answers precisely is telling you something about the whole company.
Finally, test the exit. Recovery tooling touches your subscribers at their most annoyed moment, which makes it a strange thing to rent without an off-ramp. If you cancel, what happens to sequences in flight — do they finish, halt, or vanish? Can you export the case history, including which subscribers were contacted and when? Businesses switch tools; the ones that switch painlessly are the ones that asked these questions while the demo was still pretty.
None of this makes demo day useless. It makes it the first filter rather than the verdict. The verdict comes from your failure file, your loudest month, the denominator question and the exit terms — four exhibits no demo can stage, and the four that decide what the tool is worth on the ordinary Tuesdays when renewals actually fail.
Published by Swipe Again, the flat-fee failed payment recovery service for subscription businesses.
Top comments (0)