Our users export data 3 days before they cancel.

We analyzed export events for 1,900 canceled accounts in the 30 days before cancellation.

Exported data within 30 days of canceling: 61%
Exported within 7 days: 38%
Exported within 72 hours: 29%
Exported on the same day as cancellation: 11%

Export was our earliest reliable churn signal, and we had been treating it as a neutral feature. Users were not exporting to migrate. Many were exporting to justify the decision to themselves, or to preserve work they were not sure they could recreate. The export was part of leaving, not a precursor to staying.

We added a "before you go" export summary that showed what they had built and offered a one-click archive.

Cancellation completion dropped 6 points among exporters, not because we blocked them, but because seeing the summary reminded some of them what they had invested.

What user action are you treating as neutral that is actually the first step out the door?