Payment Recovery KPI Template: Metrics & Targets

A recovery program without a scorecard is a hope with an invoice attached. Retries fire, dunning emails send, a card-update page gets traffic — and nobody can say whether any of it is working, getting better, or quietly getting worse. This template fixes that. It defines the seven KPIs that actually describe a failed-payment recovery program, gives the formula for each one, tells you where the number lives, and — just as important — tells you which direction is healthy, because a recovery rate moving the wrong way for two months is a problem you want to catch in month two, not quarter two.

One discipline before the table: do not borrow someone else's "healthy range." Recovery numbers depend on your price point, your customer mix, and your decline mix, so a range copied from another business is a target built on sand. Set your baseline in month one, then judge every later month against your own baseline and your own trend. The only benchmark figures in this template are the attributed industry stats in the context section below — context, not targets.

The seven recovery KPIs

KPI	Definition	Formula	Where to find it	Healthy direction
Payment failure rate	The share of scheduled renewal charges that fail on the first attempt, before any recovery work begins.	Failed first attempts ÷ total renewal charges attempted × 100	Your billing platform's payment or invoice report, filtered to first attempts only — retries must be excluded or the rate is understated.	Lower is better. A rising failure rate is an upstream problem (card mix, price changes, a new acquisition channel) that no amount of dunning fully compensates for.
Recovery rate	The share of failed payments that are eventually collected within your recovery window, by retries, card updates, or any other recovery action.	Failed payments eventually collected ÷ total failed payments × 100	Your recovery tool's reporting, or your billing platform's invoice history comparing failed-then-paid invoices against all failed invoices in the cohort.	Higher is better. Judge it on closed cohorts only — a month whose recovery window is still open will always look worse than it finishes.
Recovered revenue	The dollars collected from payments that initially failed. This is the number the whole program exists to produce.	Sum of the collected amounts on all recovered payments in the period	Your recovery tool's dashboard, cross-checked against payouts or settled invoices in your billing platform.	Higher is better, but read it next to failed volume: recovered revenue can rise simply because more payments failed. Pair it with recovery rate every time.
Save rate	The share of subscribers who entered the cancel flow and stayed — accepted an offer or abandoned the cancellation — instead of completing the cancel.	Subscribers who started cancelling and remained active ÷ all subscribers who started cancelling × 100	Your cancel-flow tool's reporting; if deflection runs inside your billing platform, its cancellation or retention report.	Higher is better, with a caveat: watch the 60–90 day retention of saved subscribers. A save that churns next cycle was a delay, not a save.
Time to recover	How long, on average, a recovered payment takes between the first failure and successful collection.	Sum of days from first failure to collection across recovered payments ÷ number of recovered payments	Your recovery tool's attempt log, or invoice timestamps in your billing platform (first failure date vs. paid date).	Lower is better. Faster recovery means less time in a past-due state, fewer support contacts, and — for physical goods — a better chance of shipping on schedule.
Cost per recovered dollar	What the program costs you for each dollar it brings back. The honest efficiency number, and the one percentage-of-recovery pricing makes hardest to see.	Total recovery program cost for the period (software + any per-recovery fees) ÷ recovered revenue	Your invoices from recovery vendors, divided by the recovered revenue figure above. With flat pricing the numerator is fixed; with percentage pricing it scales with success.	Lower is better. Recalculate it whenever pricing, volume, or recovery rate changes — it drifts quietly in both directions.
Involuntary churn rate	The share of subscribers lost in the period specifically because a payment failed and was never recovered — not because they chose to leave.	Subscriptions terminated for non-payment ÷ active subscribers at the start of the period × 100	Your billing platform's churn or cancellation reporting, filtered to non-payment terminations; keep it separate from voluntary cancels or the number is meaningless.	Lower is better. This is the lagging indicator the other six KPIs feed: if recovery rate rises and this does not fall, something in the handoff is broken.

Context: the attributed numbers worth knowing

Three published figures frame why these KPIs matter. They are context for sizing the problem — not targets for your scorecard.

• The problem is large. Involuntary churn is an estimated $129B problem for 2025, and 20–40% of subscriber churn is involuntary (Recurly research). If involuntary churn is a fifth to two-fifths of all churn, it deserves its own line on the scorecard — which is exactly why it is KPI seven above.
• The failure mix is concentrated. Across 5.4M failed payments, insufficient funds is the largest decline category at 40.6% (Churnkey's published analysis of Stripe data). A failure rate driven by insufficient funds behaves differently — timing retries around paydays matters more — than one driven by expired cards, where updater services and pre-dunning do the work.
• Layers add up. Stripe Smart Retries alone recover ~51% of failed payments, and layered precision retries reach ~55% (Churnkey analysis of Stripe data). Each additional dunning email touch adds roughly 1–2 points of recovery (same source). Small gaps between layers are where recovered revenue hides.

Your monthly scorecard

Copy this table into a spreadsheet and fill it in on the same day each month, using closed cohorts. The example row is hypothetical — it illustrates the format, not a benchmark, and your month one row becomes your baseline.

Month	Failure rate %	Recovery rate %	Recovered revenue $	Save rate %	Avg. days to recover	Cost per recovered $	Involuntary churn %	Notes / changes made
Example (hypothetical)	8.2	47	6,340	22	6.5	0.09	1.8	Added a second dunning email; baseline month
Month 1 (your baseline)
Month 2
Month 3

The notes column is not decoration. When a KPI moves, the first question is always "what changed?" — a new email, a retry-schedule change, a price increase, a new acquisition channel. If the change is not written next to the month, you will be guessing at the cause three months later.

How to read the scorecard

• Failure rate up, everything else flat: an acquisition or card-mix problem, not a recovery problem. Look at what changed upstream before touching dunning.
• Recovery rate up, involuntary churn flat: recovered payments are not turning into retained subscribers. Check what happens after collection — reactivation, access restoration, and the next renewal.
• Time to recover creeping up: your sequence is leaning on late attempts. Early, well-timed retries and an early card-update ask usually pull it back down.
• Cost per recovered dollar rising: either recovery is slipping or your pricing shape scales against you. This is the row where flat pricing and percentage-of-recovery pricing diverge most over time.
• Save rate up, involuntary churn up: you are saving voluntary cancellers while losing people to failed cards. Different problems, different owners — the scorecard keeps them from masking each other.

For the definitions behind the recovered-revenue and recovery-rate figures — what counts as recovered, when the clock starts, and how attribution is stamped — see how we count. For the method behind reading these numbers as a program rather than a pile of reports, see how to measure dunning performance. And if you want to sanity-check what a point or two of recovery rate is worth in dollars before you invest in moving it, run your own failed volume through the recovery calculator.

The bottom line

Seven numbers, one page, once a month. Define each KPI once, pull it from the same place every time, set your baseline in month one, and let direction and trend — not borrowed benchmarks — tell you whether the program is working. The operators who recover the most are rarely the ones with the cleverest sequences; they are the ones who notice a two-point slide the month it happens.

Frequently asked questions

How often should I fill in the recovery scorecard?

Monthly, on the same day, using closed cohorts whose recovery window has fully run. Weekly checks are fine for spotting a broken integration or a sudden failure spike, but judging recovery rate on an open cohort understates it, because the late recoveries have not landed yet.

What is a good recovery rate?

There is no universal number worth targeting, because recovery rate depends on your decline mix, price point, and customer base. Published analysis of Stripe data puts Stripe Smart Retries alone at roughly 51% of failed payments recovered, with layered precision retries reaching roughly 55% — useful context, but your own month-one baseline and the trend after it are the standard that matters.

Should recovery rate count payments or dollars?

Track both, and label which one you are quoting. A payment-count recovery rate and a dollar-weighted recovery rate can differ when high-value subscriptions fail at different rates than low-value ones. This template uses payment count for recovery rate and tracks dollars separately as recovered revenue, so the two never get conflated.

Where do I find these numbers if I only use my billing platform's native tools?

Most billing platforms report failed invoices, paid-after-failure invoices, and non-payment terminations, which covers failure rate, a workable recovery rate, recovered revenue, and involuntary churn. Time to recover comes from invoice timestamps. Save rate requires a cancel flow that records starts and completions; if yours does not, that gap is itself a finding.

Why separate involuntary churn from overall churn?

Because the fixes are completely different. Voluntary churn responds to product, pricing, and cancel-flow offers. Involuntary churn responds to retries, card updates, and dunning. Blended into one number, each can hide the other's movement for months.

Measurement without an operating rhythm decays — the recovery SOP template turns these metrics into a weekly procedure with named owners.