RecoverThe 90-day backtest
Reflects the code as of 28 July 2026. If the product changes and this page does not, that is a bug worth emailing about.
The moment you connect Stripe, a one-off scan reads your recent history and shows you what already failed and what a recovery process might have won back. It is the fastest honest answer to "is this worth anything to me".
What it reads
- 90 days of history.
- Up to 200 invoices. The cap exists so a large account cannot stall the job or exhaust Stripe rate limits. If your account is bigger than that, the scan is a sample and not a complete audit, and the result should be read that way.
- Failed invoices only, with their decline codes.
It is read-only. The backtest never retries anything, never emails anyone, and never changes a subscription. It is looking at the past, not acting on it.
How the estimate is built
Each failed invoice is classified by decline code, and each classification carries a recovery range rather than a single figure:
| Classification | Assumed recovery range |
|---|---|
| Soft decline | 30% to 50% |
| Unrecognised code | 15% to 30% |
| Hard decline | 0% to 5% |
What those percentages are, and are not
They are assumptions. They are not measured results from RecoverFlow customers, because there is not yet a body of customer results to measure. They are set deliberately low so that a real account is more likely to beat its backtest than miss it.
You should treat the output as an order-of-magnitude answer to "is there enough failed revenue here to be worth anyone's time", and not as a forecast. Anyone showing you a single confident recovery percentage for your account before running your account is guessing too, with more confidence and less disclosure.
Fixed on 28 July 2026: six decline codes were previously misclassified as retryable, which inflated the estimate for accounts holding those failures. They are now classified correctly. If you connected before that date, your backtest range was optimistic for those invoices, and re-running the scan will give you a lower and more accurate figure.
Common questions
What does the RecoverFlow backtest actually do?
It reads up to 200 failed invoices from the last 90 days of your Stripe history, classifies each by decline code, and applies a recovery range per classification to estimate what a recovery process might have won back. It is read-only and never retries or emails anyone.
Are the backtest recovery percentages based on real results?
No. They are deliberately conservative assumptions, set low so a real account is more likely to beat its backtest than miss it. RecoverFlow does not yet have a body of customer results to measure against, and says so rather than publishing a number it cannot support.
Why is the backtest limited to 200 invoices?
So a large account cannot stall the scan or exhaust Stripe API rate limits. If you have more failed invoices than that in 90 days, the result is a sample rather than a complete audit.
Does connecting Stripe for a backtest commit me to anything?
No. The scan is read-only, there is no card required to start, and you revoke access from your own Stripe dashboard whenever you like.