Churn is four problems added together. Which one is actually costing you?

A failed credit card and a disappointed customer both show up as a cancellation, and only one of them is a marketing problem. Split the number, price each part, and the work orders itself. Nothing is collected, everything runs in your browser.

The mix

If you do not know the split, start from a benchmark and adjust. Getting this roughly right beats getting a churn rate exactly right.

Involuntary, the payment failed Expired cards, declines, fraud holds, stale billing addresses. They never decided to leave. 35% Never activated, they never reached the value Bought, then never got to first real value or the second order. 25% Completed, they got what they came for The need ended. Not a failure, and not fixable with a win-back email. 15% Dissatisfied, the value was not worth the price They experienced the product and returned a verdict. 25% How strong is your recovery capability today? Dunning and card updater coverage, onboarding intervention, service recovery. This scales the recoverable share of each group. Typical

Recovery rates are the defaults I work from before a business has its own numbers: roughly 45% of involuntary churn is recoverable with competent dunning and card-updater coverage, 15% of never-activated customers with a real onboarding intervention, 5% of completed customers through a genuinely different offer, and 10% of dissatisfied customers where the underlying cause gets fixed rather than discounted. The capability slider scales all four. Every one of these should be replaced with a measured number from a holdout as soon as you have one, because a recovery program without a control group reports the customers who were coming back anyway as saves.