premortem
Premortem
Overview
Before committing to a plan, the team imagines that plan has already failed catastrophically, then works backward to enumerate causes. The retrospective frame ("it failed — what caused it?") surfaces risks the prospective frame ("what could go wrong?") systematically misses. Gary Klein operationalized this in HBR (2007), grounded in Mitchell-Russo-Pennington (1989) showing prospective hindsight increases correct failure-mode identification by 30%.
Composes with inversion (premortem is inversion made operational), confirmation-bias (structural counter), hindsight-bias (leveraged as a feature), and critical-thinking.
When to Use
- Before any high-stakes, hard-to-reverse decision (launch, major hire, contract, capital allocation)
- When the team has converged quickly on a single plan with little visible dissent
- When a previous similar effort failed and the team is about to repeat the pattern
- At project milestones to identify emerging failure modes
- Before committing to an AI product launch, model migration, or AI capex/growth spend where model commoditization, inference unit economics, a safety incident, or stretched AI valuations could break the plan within a year
Not when: small reversible decision; equivalent rigorous risk analysis already done; Chaotic domain (action before analysis); time-critical where premortem delays response.