Premortem Before a High-Commitment Decision

Define failure concretely

Choose a time horizon and describe an observable failure for each option. Avoid vague statements such as it did not work.

Define failure at a specific review date using observable consequences. Missed launch, unsafe operation, budget overrun, or loss of a required service is clearer than saying the choice went badly. Use the same definition and horizon for both options. A precise failure statement helps participants find causes that can be checked or controlled.

Generate causes independently

Ask participants to write causes before discussion so seniority and the first answer do not narrow the list.

Have each person write causes alone before discussion. Ask for events and conditions, not blame or broad labels. What failed, what warning appeared, and what assumption was wrong? Independent writing protects less powerful voices and reduces the chance that the first confident answer becomes the group's entire risk list.

Rank by preventability and impact

Separate likely inconveniences from risks that threaten the purpose. Focus on causes that are both material and within your ability to detect or reduce.

Rate each cause by plausible impact, ability to prevent or detect it, and strength of evidence. Keep severe low-frequency risks visible when they need professional treatment, but do not fill the plan with every imaginable event. The priority list should be short enough that owners can act and reviewers can tell whether protection improved.

Convert concern into control

For each priority cause, assign a verification step, owner, early warning sign, and response. A risk list without action only creates anxiety.

Convert each priority cause into one of three actions: verify an assumption, reduce the chance of failure, or limit the harm. Add an owner, due date, warning signal, and response. If no useful control exists, compare whether the remaining exposure fits the decision maker's limit. A risk list without action should not be presented as preparation.

Run the session without turning it into theater

Give participants the same failure statement and several quiet minutes to write independently. Collect causes before open discussion, then group similar items. This reduces the influence of the first speaker and makes it easier for people close to the work to raise uncomfortable operational risks.

End with a short control register, not a long worry list. Record the cause, evidence to check, preventive action, warning sign, owner, and response date. Revisit only the priority items at agreed milestones. A control without an owner or trigger is only a hope, so remove it from the plan or make the responsibility explicit.

Run the exercise for both options, including the current arrangement. If only the proposed change receives a premortem, the status quo gains an unfair advantage because its failure paths remain unnamed. Use the same time horizon and definition of failure so the risk comparison remains balanced. Record which controls apply to either choice and which are option-specific.