Same incident 3x. Fix was in postmortem

Been collecting data on the gap between postmortems and actual follow-through.

Average postmortem generates 3-7 action items. Fewer than 40% get completed within 90 days. 60% never get completed at all. "Twenty action items means zero action items. Teams will complete two and forget the rest."

The failure mode is always the same: incident happens, postmortem written, action item says "add regression test for X." Item sits in the backlog. Next sprint starts. Feature work wins. Test never gets written. Same incident 3 months later. Rinse repeat.

Ben Treynor Sloss (Google): "To our users, a postmortem without subsequent action is indistinguishable from no postmortem."

The hard part isn't knowing what broke. It's that "add a test" always loses to feature work in sprint planning. Always.

What if the regression test was created as part of incident response, not a backlog item? Root cause identified, test describing the failure scenario created right then, merged before the incident is closed. Not a ticket for next sprint. Part of the resolution.

How does your team handle this? Does "add a regression test" actually get done, or does it rot in the backlog?