Purpose
Teams planning a new initiative are subject to optimism bias and groupthink. A pre-mortem exploits prospective hindsight: research shows that imagining an event has already occurred increases the ability to identify reasons for it by 30 percent. By assuming failure is a fact and asking "why did this fail?", the team bypasses the social pressure to stay positive and surfaces the real risks hiding beneath the plan.
When to Use
- Before committing significant resources to a new feature, product, or initiative.
- At the start of a design sprint to generate sprint questions (what do we need to learn this week?).
- After creating an assumption map, to stress-test which assumptions carry the most risk.
- When a team feels overconfident about a direction and you suspect important risks are being ignored.
Steps
-
Set the scene. Tell the team: "Imagine it is six months from now. We shipped this feature (or ran this experiment, or launched this product) and it failed completely. It is an unambiguous disaster."
-
Write independently. Give each participant five minutes of silent writing time to list every reason they can think of for why it failed. One reason per sticky note. Encourage both plausible and extreme scenarios. No discussion during this phase.
-
Share round-robin. Go around the group, each person reading one reason at a time, until all notes are posted. Do not debate or dismiss during this round; simply collect.
-
Cluster related risks. Group similar failure reasons together. Common clusters include technical feasibility, customer adoption, organizational alignment, competitive response, and market timing.
-
Assess likelihood and impact. For each cluster, discuss: How likely is this failure mode? How severe would it be? Use a simple high/medium/low rating if you want structure, but the conversation itself is more valuable than the rating.
-
Convert to testable assumptions. Rewrite the top risks as assumptions that can be tested. Example: "Customers won't adopt because they don't trust AI-generated analysis" becomes the assumption "Customers trust AI-generated analysis enough to act on it without manual verification."
-
Prioritize what to test first. Focus on assumptions that are both high-risk and currently supported by the least evidence. These become your most urgent research questions or experiment hypotheses.
-
Assign owners and timelines. Each high-priority assumption should have a clear owner and a plan to test it within the next one to two weeks.
Tips
- Independence is non-negotiable. The silent writing step must happen before any group discussion. If the most senior person shares first, everyone else anchors to their risks and the exercise loses most of its value.
- Celebrate scary risks. The most valuable output of a pre-mortem is the risk nobody wanted to say out loud. Thank people explicitly for raising uncomfortable possibilities; this reinforces psychological safety for future sessions.
- Revisit periodically. A pre-mortem is not a one-time event. As the project evolves and new information emerges, re-run the exercise to catch risks that were not visible at the start.
Source
Torres, T. Continuous Discovery Habits (assumption testing and pre-mortem for discovery). Knapp, J., Zeratsky, J., Kowitz, B. Sprint (pre-mortem as a technique for generating sprint questions on Monday).
