Today’s tool is old enough to have earned its keep. The pre-mortem, associated with cognitive psychologist Gary Klein, asks a team to imagine that a project has already failed. Not “what might go wrong” in the abstract, but “it is one year from now, this effort is a ruin, and here is why.” That small change in tense does real work.

Ordinary planning meetings often reward confidence. People soften doubts, avoid looking obstructive, and assume that raising a risk means volunteering to solve it. A pre-mortem changes the social permission. The failure is already imagined, so the team is no longer attacking the plan; it is investigating the wreckage.

The exercise is compact. Gather the people close enough to know where the plan is brittle. State the objective. Ask everyone to write down reasons the project failed. Collect the reasons without debate. Cluster them into themes: unclear customer need, missing owner, integration risk, weak distribution, regulatory friction, budget pressure, dependency on one vendor, or a timeline built from hope. Then convert the most serious risks into mitigations, decision gates, or explicit kill criteria.

The method is especially useful around AI projects because demos create false certainty. A prototype can look alive while the production system lacks evaluation, permission boundaries, cost controls, data rights, and support workflow. A pre-mortem forces the team to ask where the magic show breaks once real users, real data, and real invoices arrive.

The frontier newspaper version fits on an index card: “The project failed. Name the cause before pride edits the record.” Run it before the kickoff hardens, before the budget is publicly blessed, and before the team mistakes momentum for evidence.