The tool worth filing today is the pre-mortem: a half-hour meeting that asks the team to imagine the project has already failed. That small inversion changes the room. Instead of asking people to sound negative before work begins, it gives them permission to describe a future fact pattern. The project failed. Why?

The method is useful because normal planning meetings reward confidence. People want the launch to work. Sponsors want momentum. Operators do not want to look obstructive. Junior staff may see the risk first and say it last, if at all. By the time the evidence is obvious, budget, ego, and calendar have already hardened around the plan.

A good pre-mortem is intentionally simple. Gather the people closest to the work before the initiative starts. State that it is six months in the future and the project failed in a way everyone recognizes as serious. Give each person a few quiet minutes to write down causes independently. Then collect the reasons before discussion begins. Independence matters because the first confident speaker can otherwise steer the whole ledger.

The output should not be theatre. Turn the strongest failure reasons into owners, mitigations, and tripwires. If the risk is “data migration corrupts customer records,” assign a migration rehearsal and rollback test. If the risk is “sales sells a feature that does not exist,” assign launch-language controls. If the risk is “model costs erase margin,” assign unit-economics thresholds before scaling. Every useful pre-mortem ends with changed work, not merely wiser expressions.

The technique fits today’s broader edition. Markets are running pre-mortems on oil shocks and AI valuations. Governments are running them on defense capacity and model governance. Startups should do the same before betting a quarter, a product line, or a hiring plan on assumptions that have not been made explicit.

Use the method when the project touches more than a trivial share of revenue, users, staff time, or reputation. Keep it short, written, and specific. Optimism is still allowed. The point is to make optimism earn its keep before the post-mortem writes itself.