Adam Badar

The failed-payment case is the one I would push hardest on. Involuntary churn and voluntary churn look identical in a Stripe export but they need completely different responses: one is a dunning and card-update problem you can usually recover, the other is a value conversation that started weeks earlier. Splitting those two apart is worth more than any blended churn number.

The scheduled-cancellation signal is the other one I would watch closely. The window between someone clicking cancel and the period actually ending is the only time you get a real conversation, and most teams never see it until the subscription is already gone.

Curious whether you surface how long a customer sat in an at-risk state before they churned. That lag is usually where the playbook lives.