A pre-mortem analysis asks your team to imagine the project has already failed — a year from now, in vivid detail — and then work backward to explain why, before a single dollar is spent.
Coined by psychologist Gary Klein, the pre-mortem analysis exploits a documented cognitive shift called prospective hindsight: people are dramatically better at identifying risks when asked to explain a failure that’s already “happened” than when asked to predict risks that might happen. Framing the exercise as certainty rather than speculation gets past the optimism bias that suppresses dissent in normal planning meetings.
Where Pre-Mortem Analysis Comes From
Klein first laid out the technique in a 2007 Harvard Business Review article, building on 1989 research by Deborah Mitchell of Wharton, Jay Russo of Cornell, and Nancy Pennington of the University of Colorado. Their study found that imagining a future outcome as if it had already occurred — rather than trying to predict it — increased people’s ability to correctly identify the reasons behind it by roughly 30%. That gap is the entire case for running a pre-mortem analysis instead of a standard risk review: the technique doesn’t ask people to be smarter, it asks them to answer a different, easier question.
How to Run a Pre-Mortem Analysis
- Before greenlighting, gather the team and state flatly: “It’s eighteen months from now. This has failed completely.”
- Give everyone silent time to write down every reason they believe it failed — independently, before discussion, to avoid anchoring on the first voice in the room.
- Go around and collect every reason, no matter how unlikely it sounds in the moment.
- Sort by which failure causes are both plausible and currently unaddressed in the plan.
- Build mitigations into the plan before launch, not into a retro after the fact.
People are far better at explaining a failure that’s already “happened” than predicting one that might.
Why Pre-Mortem Analysis Beats a Standard Risk Register
A standard risk register asks “what could go wrong,” which invites hedged, generic answers — “market risk,” “execution risk.” A pre-mortem analysis asks people to justify a failure that’s already certain, and certainty produces specificity. It also does something a risk register structurally can’t: it makes dissent safe. Klein’s original insight was as much organizational as cognitive — junior team members who’d never contradict a leader’s optimistic timeline in an open discussion will happily list “the timeline was unrealistic” once the failure is treated as a foregone conclusion everyone is now just explaining.
A risk register asks what could go wrong. A pre-mortem asks you to justify why it already did — and certainty produces specificity.
Pre-Mortem Analysis in Practice
The method scales from a single product launch to board-level decisions. Before committing to an acquisition, a leadership team runs a pre-mortem and surfaces that three separate people privately doubt the target’s retention numbers — a concern nobody voiced in the actual diligence meetings. Before deploying an autonomous system into production, a pre-mortem forces the question “if this agent makes a costly decision with no human in the loop, who explains that in a courtroom?” — precisely the kind of accountability gap that only becomes visible once you assume it’s already happened.
Common Mistakes When Running a Pre-Mortem Analysis
The exercise fails when it’s rushed into a five-minute agenda slot, when the most senior person in the room speaks first and anchors everyone else’s answers, or when the failure causes collected are never actually routed back into the plan. A pre-mortem analysis that doesn’t change a single decision was just an interesting meeting, not a risk-management practice.
When to Use Pre-Mortem Analysis
Use it when committing budget or headcount to any initiative with real downside — product launches, M&A, hiring plans, and, increasingly, deployments of autonomous systems where nobody has mapped what failure actually looks like until after it happens in production.
Build the mitigation into the plan before launch. Don’t wait for the retro to discover what everyone secretly suspected.
Also read: OODA Loop · Cynefin Framework · AI Agent Accountability