How do you run incident reviews and write postmortems?

Why run incident reviews in your team chat

Postmortems tend to get pushed back, and a week later nobody remembers when the release was rolled back or why the tests missed it. The review gets written from memory in a wiki and discussed in DMs, and the action items end up as bullet points with no owner. In Pachca, the review comes together from the thread while the outage is still fresh, and it lives right next to your alerts.

• The timeline is already written. The alert time, the on-call engineer’s first moves and the recovery are all in the thread, down to the minute. The agent carries them into the review, so nobody has to reconstruct events from memory.
• The draft is ready while the outage is fresh. The agent builds it from your template right after recovery. All the team has to do is check the facts and answer the open questions.
• Blameless by design. The template asks what in the process let the outage happen. The causes list a release, a timeout and a stub on staging, not names, so people can talk openly about what they did.
• Every action item has an owner and a due date. The agent sets reminders for the action items, and they reach the owners on the due date. A week later, you can see what got done.
• All reviews in one chat. Postmortems are listed in a feed in the Postmortems chat, and you can find them by searching for a service or alert name. Each one is discussed in its own thread, where you can loop in a developer from another team.
• Team leads see the big picture. How many outages there were this month, which ones keep coming back and which action items are stuck: it’s all visible in the chat, no separate report needed.