Most operators can describe their failed payments in percentages and still not know whether fixing them is worth paying for. Eight percent failed sounds alarming. Three percent sounds manageable. Neither figure tells you what a recovered renewal puts back, what it costs to get it back, or how long the customer who was saved is likely to stay. Before you compare tools, demos, or promises, run one month of your own billing through a plain calculation. The arithmetic is unglamorous. That is why it works: it forces every hopeful claim to meet your invoice values first.
Start with one real month, not an average year
Take last month, because it is complete and because memory has not smoothed it yet. Count renewals attempted, not subscribers. Note the average invoice, excluding one-off setup fees and annual plans if they behave differently. Suppose you attempted 2,000 renewals at $49. Eight percent failed on the first try: 160 invoices, or $7,840 that should have arrived and did not. That $7,840 is not your opportunity. It is the ceiling. Some of those customers wanted to leave, some cards were closed, and some attempts would have recovered with a single ordinary retry you already run.
Separate recoverable failures from final ones
Now sort the 160 by reason, as far as your processor lets you. Insufficient funds, temporary bank holds, and expired or replaced cards are usually worth work. Stolen cards, closed accounts, and hard fraud blocks usually are not. If, say, 120 of the 160 were plausibly recoverable, your workable pool is $5,880, not $7,840. Apply a sober recovery rate to that pool, not a best-case rate to everything. At 40 percent, you recover 48 invoices. Gross recovered revenue for the month is 48 times $49: $2,352. Write that number down before anyone mentions lifetime value. It is the only figure the rest of the argument is allowed to use.
Subtract the fee the way the invoice subtracts it
Fee models change the answer more than brochures admit. A provider charging 20 percent of recovered revenue would take about $470 from that $2,352, leaving $1,882 before your own time. A flat fee of $179 would leave $2,173. A per-recovery charge needs the recovered count, not the revenue: at $12 per save, 48 saves cost $576, leaving $1,776. None of these is automatically cheapest. At a $19 price point the flat fee needs many more saves to clear; at $199 the percentage fee grows with every success. This is the practical core of working out what payment recovery actually returns — the model, not the headline rate, decides the return.
Work out your break-even rate backwards
Turn the calculation around. With a flat $179 fee and a $49 invoice, you need four recovered invoices in a month just to cover the fee, ignoring staff time. With a 20 percent model you cover the fee on the first dollar, but you keep giving away a fifth of every later save, including saves your existing retry might have won anyway. That last point matters. If your current process already recovers 15 of those 48, the new spend should be judged only on the other 33. Incremental gross is $1,617, not $2,352. Vendors who count every recovered invoice as theirs are counting some of your money twice.
Add retained months once, and label them as a forecast
A recovered subscriber often pays again. If those 48 customers stay an average of three further months, that is another $7,056 in future invoices, before ordinary churn and before the cost of serving them. It is legitimate to show that figure in a business case. It is not legitimate to add it silently to this month's cash, or to assume every recovered customer behaves like your happiest cohort. Put cash recovered now in one column, forecast renewals in another, and costs in a third. A decision that still looks good using only the first column is robust. A decision that needs the second column may still be right, but you should know that is the bet you are making.
A short worksheet before any purchase
Ask five questions. How many first-attempt failures did we have last month, in invoices and dollars? What share were soft, plausibly recoverable failures? What do we already recover without help? What would each fee model take from our sober estimate, not a vendor's? And how many incremental saves must appear each month for the spend to clear its cost with room to spare? If a seller can help you answer those with your data, keep talking. If the conversation keeps returning to industry averages, pause. Your renewal file is a better source than anyone else's benchmark, and it is already paid for. Resources that start from that file, including Swipe Again's homepage, are at least asking the question in the right order: count first, buy second.
Top comments (0)