Framework
Risk

Pre-mortem

Imagine the failure first

Best for
surfacing risks before committing
Time
30–60 min
Difficulty
Beginner

Fill in your own Pre-mortem right here.

Free · no sign-up · saves in your browser · export a PDF

Pre-mortem diagramSchematic of the Pre-mortem framework: Imagine the failure first.!Imagine it failed
The Pre-mortem diagram — Imagine the failure first.
Example

Q2 pricing experiment — surfacing failure modes before launch

Imagined failure modes
  1. Customers churn when the new tier price hits annual renewals
  2. Existing power users feel grandfathered out of the best features
  3. CFO blocks discount approval — marketing has no fallback
  4. Annual contracts have legal carve-outs that void the new tier
  5. Support volume spikes; the team isn't trained on the new SKUs

What it is

A pre-mortem is a meeting you hold before a project begins, in which the team is asked to imagine that the project has already failed — spectacularly — and to brainstorm the reasons why. The framework comes from research psychologist Gary Klein, who published it in his 2004 book The Power of Intuition and then wrote it up for Harvard Business Review in September 2007 — the two-page column that nearly everyone now cites as the origin. The technique is a deliberate counter to optimism bias: people are much better at criticizing a plan when they're told to assume it's already broken than when asked to flag risks in the abstract.

It's prospective hindsight — borrowing the clarity that comes after a failure and applying it before launch, while you can still change things.

The Premortem Provenance Chain

Search results for this technique almost always compress to one line: "Gary Klein, Harvard Business Review, 2007." That line gets two things wrong — the date, and whose research the famous statistic belongs to. Each step below is separately documented.

DateWhoWhat they actually did
1989Deborah J. Mitchell (Wharton), J. Edward Russo (Cornell), Nancy Pennington (Colorado)Published "Back to the future: Temporal perspective in the explanation of events," Journal of Behavioral Decision Making 2, 25–38. Found that describing an event as certain rather than possible produced longer explanations containing more episodic reasons. This is the evidence base — and it is not Klein's.
1998Gary KleinSources of Power: How People Make Decisions (MIT Press) establishes the naturalistic decision-making research programme the premortem grows out of.
2004Gary KleinThe Power of Intuition (Doubleday) carries the first published description of the premortem exercise — three years before the article that gets the credit. Klein's own HBR author bio lists this book.
September 2007Gary Klein"Performing a Project Premortem," Harvard Business Review, pp. 18–19 — a two-page item in the Forethought/GRIST column, reprint F0709A. Not a study, not a paper. This is the canonical citation.
2011Daniel KahnemanThinking, Fast and Slow singles the premortem out as a practical remedy for overconfidence. This endorsement, not the HBR column, is what moved the technique into mainstream management practice.

The 30% Gap

The single most-quoted sentence about the premortem is a misquote — and the person who introduced the error was Klein himself. His HBR column says the 1989 research found prospective hindsight:

"increases the ability to correctly identify reasons for future outcomes by 30%."

Mitchell, Russo and Pennington measured how many reasons people produced, not whether those reasons were any good. Their design contains no accuracy check at all. So the honest version of the claim is:

What Klein's column saysWhat the 1989 study measured
Ability to correctly identify reasons, +30%Number of reasons generated, ≈ +30%
Implies better risk detectionImplies greater fluency — quality unassessed

Why this matters for how you run one. A premortem reliably gets more risks onto the table; nothing in its evidence base says those risks are the right ones. That makes step 6 — voting and ranking — the load-bearing step, not the brainstorm everyone focuses on. Treat the output as a generated list to be triaged, not a validated forecast, and budget real time for the triage. The academic literature here is thin: essentially one 1989 study and a handful of follow-ups, which is worth knowing before you cite the 30% in a stakeholder deck. There is also a class of failure triage cannot rescue, because the brainstorm never generates it: the 1989 study found prospective hindsight produces more episodic — story-shaped — reasons, and an already-false unit economic is not a story. Worked against a real seven-year shutdown in the Pre-mortem Example: The Failure a Premortem Cannot Catch, which introduces the Day-One Test.

Klein's own column is more useful than its abstract suggests, because it contains two real cases. In one, a team building computer algorithms for military air-campaign planners heard a member — silent through the entire kickoff — volunteer that one algorithm wouldn't fit on the field laptops and would take hours to run. The developers had a shortcut they'd been "reluctant to mention." It was substituted, and the project succeeded. That is the mechanism working exactly as advertised: not better analysis, but permission to speak.

When to use it

A pre-mortem is most valuable on commitments large enough that failure would hurt, but small enough that you can still alter course. Reach for it when:

  • Kicking off a quarter-plus initiative with multiple teams
  • Greenlighting a product launch with a non-trivial budget
  • Signing a long-term vendor or partnership contract
  • Replatforming or migrating a core system
  • Hiring for a senior role where a bad outcome takes a year to unwind

Skip it for small, reversible decisions — the overhead isn't worth it.

How to run it

  1. Get the full project team plus 1–2 outside skeptics in a room (or on a call) for 60–90 minutes.
  2. Set the scene: "It's 12 months from now. The project has been a complete failure. The launch missed every target. Take three minutes and write down, individually, every reason you can think of for why."
  3. Have each person write silently. Silent writing prevents the loudest voice from anchoring the room and gives introverts equal weight.
  4. Round-robin the reasons. Capture every one — don't debate yet.
  5. Cluster duplicates and group by theme (product, team, market, dependencies, execution).
  6. Vote on the top 5–7 most likely failure modes.
  7. For each, assign a specific mitigation, an owner, and a date to revisit.
  8. Save the list. Review it at the project's midpoint — a pre-mortem that's never reread is a pre-mortem that didn't happen.

Common pitfalls

The most common failure mode is running it as a checkbox. The team gathers, dutifully lists three generic risks ("scope creep," "communication issues," "resource constraints"), and goes back to work. A real pre-mortem produces specific failure stories with named actors: "the data engineer leaves in Q2 and nobody can debug the pipeline," not "key person risk."

Second pitfall: holding the pre-mortem too late. Once people have publicly committed to a plan, they'll defend it. Schedule the pre-mortem before the project gets a name and a kickoff deck.

Third: skipping the mitigation step. The point isn't catharsis; it's changing the plan. If the list of failure modes doesn't produce at least three concrete edits to the project plan, you've held a venting session.

Variations

A close relative is the Murder Board — a more adversarial review where outside reviewers actively try to dismantle the plan rather than imagining its eventual failure. Use a Murder Board when you have a polished proposal and want it stress-tested; use a Pre-mortem when the plan is still forming. For the difference between a premortem and a postmortem (the meeting you hold after failure has occurred), see Premortem vs Postmortem. Another adjacent technique is Red Teaming, common in security and defense, which assigns a sub-team to play the adversary continuously rather than in a single meeting. For most product and ops contexts, the one-hour pre-mortem is the right cost-to-value point — light enough to actually run, structured enough to produce changes.

Playbooks by industry

Sources

Frequently asked questions

What is a premortem?

A premortem is a structured exercise in which a team imagines that a project has already failed and then works backwards to identify why. Psychologist Gary Klein formalized the technique in a 2007 Harvard Business Review article. The reframing leverages a cognitive bias — people generate more specific failure scenarios when asked 'why did this fail?' than when asked 'what could go wrong?'. The output is a concrete, addressable risk list before work has started.

How is a premortem different from a risk register?

A risk register lists abstract risks ('integration delays', 'scope creep') that teams have heard before and tune out. A premortem produces specific imagined failures ('we shipped on time but no one used the feature because onboarding looked like an internal tool') concrete enough to act on. The reframing — 'it failed; why?' — is what reliably produces the difference in output quality between the two formats.

When should you run a premortem?

Run a premortem before committing to any plan where the cost of failure is hard to reverse — major launches, acquisitions, hires, organizational changes. Skip premortems for low-stakes reversible decisions where the 60–90 minute investment isn't justified. The technique is also useful as a periodic check-in on long-running projects, not just at kickoff.

Who should facilitate a premortem session?

Not the project owner — they have too much investment in the plan to surface risks honestly. Choose a senior peer from a different team or a designated devil's advocate. The facilitator enforces silent writing, prevents early debate, and keeps the imagined-failure framing alive. Rotating the facilitator role across projects develops the skill organization-wide and prevents the practice from collapsing into a checkbox.

Pre-mortem applied to real companies

All examples →

Related frameworks

Ready to try it? The Pre-mortem worksheet is free and needs no account.
Open the free worksheet →