Subscription Recovery Cohort

A service has $100,000 of subscription payments fail in January; by the end of February, $60,000 has settled, $10,000 remains in a permitted recovery process and $30,000 is unresolved. Reporting January as a 60% final recovery rate may be premature if the window continues.

Define cohort membership by the first failure date, and decide whether the unit is invoices, payment attempts, customers or currency value, since one invoice can have several retry attempts and counting attempts as separate failed subscriptions inflates the base. Specify the observation window, because a seven-day and a 60-day recovery rate answer different questions, and compare cohorts at the same elapsed age, not only as of the current dashboard date.

Stripe's revenue-recovery analytics distinguishes recovered, not-recovered and still-in-recovery payments, and its current-period rate can shift while retry windows remain open; the exact categories reflect that product and its reporting rules. Stripe's Smart Retries documentation describes automated payment attempts for eligible subscription invoices, but eligibility and processor limits vary, so a cohort report should not assume all failed amounts receive identical recovery treatment.

Confirm the original obligation, since a billing error, valid cancellation or duplicate attempt should not become a recoverable debt merely because the processor declined a charge, and state how voided invoices and refunds affect the cohort. Track settlement, not only a successful authorisation, because a charge can later reverse or be disputed, and define whether a recovered value must remain settled through the reporting date.

An illustrative 30-day recovered-value rate is valid failed invoice value collected within 30 days divided by valid initial failed invoice value, so if $60,000 of $100,000 is collected the rate is 60%, keeping the same currency and treatment of partial payments. Use a cohort table whose rows can be first-failure months and whose columns can show recovered share after 7, 30 and 60 days, with recent rows that have insufficient elapsed time visibly incomplete.

Segment where useful, since card declines, customer authentication requirements and manual-invoice lateness may have different resolution paths and should not be compared as though they received the same process. Separate methods without overstating causality: a payment may settle after a reminder but might also have cleared without it, so attribution to retries, email or account-manager help depends on available evidence.

Protect customer treatment, because a report should assess valid recovery, not incentivise repeated messages or charge attempts beyond terms and rules, and complaints and service loss belong alongside the financial rate. Watch customer mix, since a seasonal influx of trial conversions can change the denominator and a few large business accounts can dominate the value rate, so show count and value views together.

Reconcile to billing and cash, because an invoice marked paid may not yet have cleared to the bank and a refund can alter the final amount, and align the reporting basis with its purpose. Assess operational decisions carefully: if recent cohorts settle more slowly, investigate processor changes, payment method mix or billing errors before changing retry cadence, and note the campaign version, because a revised campaign may apply only to new failures while older invoices follow old rules.

Present uncertainty, since an open cohort can report recovery to date and remaining exposure, not a single final loss claim, and avoid extrapolating from one unusually small group. For an owner, recovery cohorts show whether payment problems are being fixed over comparable windows, making progress and unresolved value visible without turning a moving number into a false certainty.