How to run a pre-mortem that people actually tell the truth in
The technique is simple and almost every team runs it badly. The failure is nearly always in the setup, not the exercise.
Amara Okafor4 July 2026 · 2-minute read
A pre-mortem asks a team to imagine a project has already failed and explain why. It surfaces risks that a normal risk review never gets near, because it removes the social cost of being the person who raised a doubt.
It also fails often, and it fails in a predictable way: the setup lets people stay polite.
The setup that works
State the failure as fact, in the past tense. Not "what could go wrong" but "it is twelve months from now and this failed badly".
Everyone writes alone for ten minutes before anyone speaks.
The most senior person in the room contributes last, or does not contribute at all.
Collect everything before evaluating anything.
Reading the output
Sort the causes into two piles. Things that would have been visible in advance, and things that would not. The second pile is genuine risk and you manage it with buffers. The first pile is not risk at all. It is a list of decisions you have not made yet, and it is almost always longer than people expect.
Run it at the point where the plan is real enough to critique and not so far along that changing it feels like an admission. In practice that is earlier than it feels comfortable.
Written by
