The workbench closes with Gary Klein’s premortem, a planning tool that earns its place because it is simple, cheap, and hard to fake. Instead of asking a team what might go wrong, the facilitator asks them to imagine the project has already failed. The team then works backward to explain why.
That small shift matters. In an ordinary risk review, people often soften their doubts. Nobody wants to be the person slowing the launch, questioning the hire, or dampening the room after a confident executive pitch. A premortem changes the social permission. The failure is not an insult to the plan; it is the premise of the exercise. Everyone is invited to investigate the wreck.
The digest cites Klein’s Harvard Business Review article and notes Daniel Kahneman’s later popularization in Thinking, Fast and Slow. It also says studies show the technique increases identification of likely failure reasons by about 30% compared with standard risk reviews. The exact figure should be traced to the underlying research before becoming a universal claim, but the operational value is clear even without worshiping the statistic.
A useful premortem takes 15 to 20 minutes. First, state the decision or project plainly. Second, ask participants to imagine that six months or a year have passed and the effort has failed badly. Third, have each person write down reasons privately before discussion begins. Fourth, gather the reasons without debate. Fifth, identify the few risks that are both plausible and consequential. Sixth, adjust the plan.
Private writing is important. It prevents the first confident voice from anchoring the room. It also lets quieter people surface operational details that senior leaders may miss: brittle dependencies, unclear ownership, missing data, training gaps, migration timing, vendor fragility, legal review, customer confusion, or the simple fact that nobody has budgeted enough time.
The method is especially useful for AI projects. A team can imagine that the agent rollout failed because permissions were too broad, evaluation data was stale, users ignored the tool, costs ran past forecast, model behavior changed after an upgrade, or compliance teams were brought in too late. Each failure story becomes a design question before the damage is real.
The premortem should not become theater. If leaders collect the risks and change nothing, the room will learn that candor is decorative. The method works when it produces visible adjustments: smaller scope, clearer gates, better monitoring, named owners, fallback plans, or a decision not to proceed.
The frontier clerk’s version fits on a card: pretend the thing failed, write down why, protect the dissenters, and fix the plan while it is still cheap. Many bad projects do not need more optimism. They need an honest obituary before launch day.