Dunning is the automated process a subscription or stored-card business runs after a scheduled charge fails: a short sequence of retries, customer-facing messages, and prompts to update the card on file. The goal is to collect the payment before the customer is involuntarily cancelled. The term comes from collections language but in modern commerce specifically describes subscription-platform retry flows.
These failures are involuntary churn — the customer did not choose to leave, the card did. The churn rate entry treats this slice separately because the recovery surface is different from voluntary cancellation: intent to stay is still there, only the payment broke. For many subscription brands it is the more recoverable of the two.
Recovery depends on failure mode. Soft declines — insufficient funds, issuer fraud heuristics — are retryable. Hard declines — closed accounts, expired cards — need a fresh card on file.
A modern subscription platform ships a reasonable default: a handful of retries across one to two weeks, sometimes ML-scheduled. The gap between default and tuned is where most of the recoverable revenue sits. The levers: Account Updater services (Visa’s VAU, Mastercard’s ABU) that replace re-issued numbers behind the scenes, pre-dunning messages before a known expiry, and the copy and friction of the failure-recovery page itself.
The headline metric is recovery rate — the share of failed-charge dollars eventually collected. Tuned programs recover a material portion, but the share moves enough across category, price point, and payment mix that a pinned range overstates how universal it is. Recovered revenue lands in LTV, so a few points of improvement compound across the active base.
The trap is treating “improve dunning” as “retry more times.” Aggressive retry trips issuer fraud heuristics, raises the soft-decline rate, and compounds the failure it was meant to fix.