Gary Klein's Premortem: How to Run One in 20 Minutes

In the late 1980s Klein ran a research company. Most of its projects went well and a few failed. After each failure the team held an after-action review and found what they should have done differently. Then Klein asked why they waited until the end. A physician's post-mortem helps the doctor and the family and does nothing for the patient, so he moved the review to the kickoff meeting.

His change is in how the question is put. The usual kickoff ends with "Does anybody see any problems?" and Klein says nobody answers. The team has just spent an hour on the plan, everyone is impatient to start, and naming a flaw can cost you. A premortem takes the hedge away. The facilitator looks into an imaginary crystal ball, six months or a year out, and reports that the project has been a disaster. The ball is never wrong, though it does not say why the project failed.

A risk review asks what could go wrong, which invites a polite guess. A premortem says the project failed and asks for the explanation, so each person's experience goes to work on a story. In a 2010 study by Veinott, Klein and Wiggins, 178 students critiqued a university's plan for a flu outbreak. The group that ran a premortem lost the most overconfidence, more than the groups that critiqued the plan or listed pros and cons.

This is why the leader goes first. When a leader gets cold feet and says they just want to hear what everyone else has, Klein says the room turns dangerous and only the leader is protected. It is also why he prefers a premortem to appointing a devil's advocate. Once one person is the official critic, everyone else stops criticizing, and nobody can tell whether the critic means it. In a premortem the critics are the people who will carry out the plan, so their concerns are real.

The same list can become your stopping rules. Annie Duke's kill criteria take the warning signs from it and attach an action to each one before launch, so the team knows in advance what will make it stop.