SaaS Revenue Recovery Benchmarks: What Is Actually Published
The recovery numbers that have a source behind them, and the widely-quoted ones that do not. Every figure here is linked to the study it comes from.
PaidGuard is new and has no customer dataset to publish. What follows is other people's published work, with the source next to each number — including the awkward parts, like two respected sources that disagree with each other by a factor of thirty on the same question.
The benchmark breakdown
Recovery rates that have a published source
| Median dunning recovery Recurly, 2024 — 2,200+ merchants | 49.0% |
| Stripe Smart Retries, on their own Stripe, average across businesses | 57% |
| What an optimised strategy targets Recurly — a target, not a measured quartile | 53% → 71% |
No quartile table appears here because nobody publishes one. Recurly shows percentiles inside its own product without publishing the values or the sample size, and the P25/P75 tables circulating on marketing blogs have no study behind them. An earlier version of this page had a quartile table. It was invented.
Two of those three numbers are vendor data — Recurly and Stripe both sell the thing they are measuring. That does not make them wrong, but it is the label they deserve, and you should read them the way you would read any figure published by an interested party.
Time-to-recovery matters more than you think
There is no published curve of recovery rate against elapsed time. There is one published fact about timing, and it is a useful one:
The one timing figure with a source
90%
of recovered transactions happen within the first 10 days after the failure (Recurly). That is the whole published record on timing. Anything more precise — hour-by-hour tables, "88% within one hour" — comes from nowhere, including from an earlier version of this page.
The defensible reading is narrower than the one usually sold: recovery is front-loaded, so a sequence that is still running on day 20 is mostly running for nothing. It does not follow that an email at minute 59 beats one at minute 61, and nobody has published anything that would let you claim it does.
What separates top-quartile from the rest
We have no companies to look at. What the published sources support, and nothing beyond it:
- Act early. 90% of recoveries land in the first 10 days (Recurly). Front-load the sequence.
- Expect diminishing returns from each extra email. Open rates fall from 55.8% on the first dunning email to 20.6% on the fifth, and each additional email adds roughly 1–2 points of recovery (Churnkey with Stripe, 5.4M failed payments, 2025).
- Turn on everything Stripe gives you first. Smart Retries, failed-payment emails, automatic card updates and recovery analytics are all included and cost nothing extra (Stripe docs).
- Respect the network limits. Visa allows 15 retries and Mastercard 35 over 30 days (Churnkey). Beyond that you are not persistent, you are non-compliant.
- Do not assume a branded page beats Stripe's. We looked for a published A/B test on that and found none.
What failure reasons appear most
Two credible sources answer this question and they disagree sharply. Both are shown, because picking the more convenient one is how this article got into trouble in the first place:
Why payments fail — two sources, two answers
| Churnkey × Stripe 5.4M failures, 2025 | Recurly 2024 report | |
| Insufficient funds | 40.5% | 32.2% |
| Generic decline | — | 39.3% |
| Expired card | 1.1% | — |
The expired-card figure is the interesting one. Everyone's mental model says expired cards drive involuntary churn; the largest published dataset puts them at roughly one failure in ninety. The likeliest explanation is that the card networks now update expiring cards automatically before they ever fail — which is a Stripe feature you already have. No source publishes a per-reason recoverability rate, so this table does not have that column.
The honest summary: if insufficient funds is the leading cause, then most failed payments are a timing problem rather than a dead-card problem — the money arrives later in the month. That argues for patience and a sequence that spans days, not for hammering the same card on the same afternoon.
Before benchmarking yourself against anyone, find out what failed payments have actually cost you. Connect Stripe read-only — no card, nothing charged — and see your own last 90 days.
See my real number, free →