Every recovery report has a quiet choice buried in it: what counts as the base. Put every failed payment in the denominator and the recovery rate looks modest and honest. Put only the failures the tool actually touched in the denominator and the same work suddenly looks heroic. Neither number is a lie, exactly. They are answers to different questions, and the trouble starts when a team compares one vendor's touched-base rate against another vendor's all-failures rate and calls it an evaluation.
Run a denominator audit before you believe any ROI figure, including your own. Start with a single month and write down four numbers in order. First, total failed renewal value: every dollar that was attempted and declined. Second, the workable pool: that total minus the failures no sequence could have saved, such as accounts the customer closed outright or cards reported stolen with no replacement issued. Third, recovered value: money that actually collected, however it landed, counted once. Fourth, the cost of recovering it: tool fees, message costs, and the staff hours the queue still demanded. Most inflated reports collapse somewhere between steps one and two, where the unworkable quietly disappears and the rate climbs.
The counting rules matter as much as the arithmetic. A payment that succeeds on a retry the platform ran natively is still recovered revenue for the business, and any honest business-level ROI includes it. A payment that succeeds because the customer updated a card after a message is recovered revenue too. What neither case licenses is crediting the same dollar twice, once to the retry engine and once to the email sequence, because the dashboard total then exceeds the bank total. Following the recovery return calculation that counts only money that actually landed keeps the ledger reconciled to deposits, which is the only reconciliation a finance lead will accept at quarter end.
Timing is the second quiet distortion. Recovery that lands ninety days after the failure is worth less than recovery that lands in the same billing cycle, because the late dollar often arrives after the account has already been written off, refunded a prorated amount, or replaced by a new signup that marketing paid to acquire. A serious ROI read therefore buckets recoveries by age: same cycle, thirty days, ninety days. A tool that looks average on the headline rate and strong on same-cycle recovery is usually the better buy, and the headline alone would never tell you.
There is also a cost nobody puts in the spreadsheet: the recoveries that were never attempted. If a fifth of failures sit in a status nobody monitors, the effective ROI of the whole operation is dragged down no matter how elegant the attempted fifth performs. The audit should end with a coverage question, not a rate question. Of everything that failed last month, what share entered any recovery path at all? Raising coverage from sixty percent to ninety percent routinely beats squeezing two extra points out of the sequence that handles the covered share.
Do the audit quarterly, on the same definitions, and keep the worksheet. The definitions are the asset. Vendors change, platforms change their native retry behaviour, and card networks shift their decline mix, but a stable denominator lets you see which change actually moved the money. Without it, every quarter is a fresh argument about whose number felt right.
That discipline is the working habit behind Swipe Again: recovered dollars counted once, on collected money, against the full workable pool. An ROI figure that survives a denominator audit may be smaller than the one in the sales deck. It will also still be true in the board meeting, which is the only place the number really has to work.
Top comments (0)