The postmortem template that turned action items into machinery.
Most postmortems produce promises. Ours produce mechanical rules. The template is the difference.
Every meaningful incident gets a postmortem. That sentence is easy to say and almost useless without the second half: the postmortem has a template, and the template has teeth.
Promises don't compound. Rules do.
The failure mode of the average postmortem is the action item that says "be more careful next time." That's a promise made by a tired person to a future tired person. It compounds at exactly zero.
Our template ends every postmortem with action items that ship as mechanical rules — gates, hooks, checks. Not "remember to verify," but "the pre-flight gate now runs this check." The postmortem isn't closed when the document is written. It's closed when the action items are shipped or explicitly accepted.
The loop
The reason this works is that the postmortem isn't a standalone ritual. It's one third of a loop: the sprint contract pre-registers the definition of done, the pre-flight gates enforce it mechanically, and the postmortem evaluates the structure — not the person — when something breaks. Each postmortem strengthens the original contract. That's the compounding.
One rule did more for follow-through than any amount of discipline: a future incident in the same domain can't open until the prior postmortem in that domain is closed.
What the template can't do
It can't make an incident meaningful. Trivial incidents get trivial postmortems, and the template will happily process garbage. The judgment call — was this meaningful — stays human. The template handles everything after that call. The full template and the worked examples are in the lab's postmortem discipline lesson.