The Pre-Mortem: How to Run One Before Your Decision Fails

How to run a pre-mortem: Gary Klein's method for imagining a plan has already failed, with a copyable template, examples and what the evidence shows.

Colleagues in a meeting room beside a whiteboard listing four countries, with a laptop on the table
By Rob Griffiths · 13 min read
Most teams plan a decision by listing the reasons it will succeed. The pre-mortem turns that around: assume the project has already failed catastrophically, then ask why. The unstated worries that polite meetings suppress come out within minutes - and the plan that emerges is usually a meaningfully better one. The technique was popularised by cognitive psychologist Gary Klein in a 2007 Harvard Business Review article. Daniel Kahneman has since called it one of his favourite debiasing tools, citing it directly in Thinking, Fast and Slow. The reason it works has a name in the psychology literature: prospective hindsight.

What Is a Pre-Mortem?

Working backwards from imagined failure

A pre-mortem is a structured exercise run before a decision is finalised. The team is told to assume the project has already gone live and produced a disastrous outcome. Each person then privately writes a short story explaining what caused the failure. The lists are pooled and discussed. This sounds trivial. The interesting part is what it surfaces. In a normal planning meeting, dissenters often stay quiet - disagreeing with a decision the boss has clearly endorsed feels socially expensive. A pre-mortem inverts the social dynamic. The exercise asks for criticism. People who would have privately doubted the timeline, the supplier, or the assumed customer demand now have permission to voice it. Klein has described pre-mortems as a way to legitimise dissent. They treat scepticism as a contribution rather than an obstacle, which means the contribution actually arrives.

Where the Idea Comes From

Gary Klein, prospective hindsight, and Kahneman's endorsement

Klein is a cognitive psychologist best known for naturalistic decision-making research - how experts actually decide under time pressure, as set out in his 1998 book Sources of Power. He published the formal pre-mortem protocol in HBR in September 2007 under the title Performing a Project Premortem. The psychological mechanism Klein draws on is older. In 1989, Mitchell, Russo, and Pennington published Back to the future: temporal perspective in the explanation of events in the Journal of Behavioral Decision Making. They called the effect prospective hindsight: explaining a future event as if it had already happened. Their abstract reports that what mattered was treating the outcome as certain - people then gave longer, more episodic explanations - while past versus future framing on its own had little effect. You will often see the claim that prospective hindsight makes people "30% better at identifying risks". The 1989 abstract does not say that. It reports richer explanations, not more accurate forecasts, so treat the 30% figure as a popular simplification rather than a finding. Kahneman and Tversky's earlier work on the planning fallacy sets up why this matters. Teams systematically overestimate how quickly things will go well and underestimate how often things go wrong. A pre-mortem is a cheap way to push estimates back towards reality without asking people to override their natural optimism in the moment - and Klein's own experiment, covered below, suggests it reduces overconfidence more than ordinary critique does.

Why It Works

The psychology behind a 15-minute exercise

Three mechanisms do the work. Prospective hindsight. When you imagine an outcome has already happened, you stop forecasting and start explaining. Explanations are more concrete than predictions, so the failure modes you generate are more specific and more useful. "Marketing might be late" becomes "the agency missed the launch deadline because the brief wasn't signed off until week six". Reversed social pressure. In ordinary planning meetings, support for the leader's plan is the default. Pre-mortems flip the polarity: the entire room is now asked to argue for failure. Saying nothing in a pre-mortem is the conspicuous behaviour, not dissenting. Independent generation. Klein's protocol asks people to write their failure stories silently and individually before any group discussion. This blocks two well-documented group failures: anchoring on whatever the first person says, and groupthink suppressing minority concerns. By the time the lists are read out, the room already contains a diverse set of independently generated risks. One outcome is easier to explain than many. Kahneman's discussion of the method cites Karl Weick's observation that it is easier to imagine detailed causes for a single outcome than for a range of possible outcomes (Farnam Street). Fixing the outcome as "it failed" narrows the search and makes the stories concrete.

How to Run a Pre-Mortem

Klein's six-step protocol

  1. Gather the decision team

    Include everyone who has substantive context on the project - sponsors, leads, and execution staff. Klein recommends the same group that will execute the plan, including any sceptics.

  2. Set the scene

    Tell the team to assume it is a year (or whatever timeframe is appropriate) after the decision launched. The project has failed. The outcome was bad enough that everyone can agree on it.

  3. Each person writes silently

    Independently list every plausible reason the project failed. No discussion yet. Klein's own version gives about two minutes; for a complex plan, five to ten minutes produces fuller lists. The independence is the point - it stops the room anchoring on the first opinion.

  4. Round-robin reading

    Going around the table, each person reads one item from their list. Continue until all unique reasons are surfaced. Capture them on a shared board.

  5. Rate and group the failure modes

    Cluster related items, then rate each on the three axes Klein later added: how likely it is, how much damage it would do, and how easy it is to prevent. Challenge any that seem implausible.

  6. Build mitigations into the plan

    For each significant risk, decide whether it changes the plan, the milestones, the team, or the success criteria, and give it an owner and an early-warning sign to watch for. Some risks will be accepted. The discipline is making the acceptance explicit.

Does a Pre-Mortem Actually Work?

What the evidence does and does not show

The evidence is encouraging but thinner than the technique's reputation. Three pieces carry most of the weight. The 1989 prospective-hindsight study. Mitchell, Russo and Pennington found that treating an outcome as certain produced longer, more concrete explanations of it. That supports the mechanism - a pre-mortem generates more specific failure stories - but it is a study of explanation, not of forecasting accuracy. Klein's 2010 experiment. Klein, with Beth Veinott and Sterling Wiggins, compared the pre-mortem against other ways of critiquing a plan with 178 college students. Klein reports that the pre-mortem group showed the largest drop in overconfidence (Psychology Today). It is a single study on students, not professional teams, so treat it as a promising signal rather than proof. Practitioner endorsement. Kahneman has called the pre-mortem his most valuable decision-making technique, while crediting Klein, and notes that organisations tend to dislike pessimists, which is exactly the pressure the exercise is designed to relieve (Farnam Street). See our summary of Thinking, Fast and Slow for where it sits among his other debiasing ideas. The honest summary: the pre-mortem reliably makes risks easier to say out loud and more concrete. Whether that translates into better outcomes depends on what the team does with the list.

A Pre-Mortem Template You Can Copy

The prompts and the output to leave the room with

Read this framing aloud, word for word, so nobody softens it: "It is [date, six to eighteen months out]. We went ahead with [the plan]. It has failed - clearly enough that everyone in this room agrees. Take two minutes, on your own, and write down every reason you can think of for why it failed." Then work through these prompts in order:
  • Failure stories: what went wrong? One reason per line, as specific as possible.
  • Ratings: for each reason, how likely is it, how bad would it be, and how easy is it to prevent?
  • Mitigations: for the top risks, what changes in the plan now?
  • Owners: who is responsible for each mitigation?
  • Early-warning signs: what would we see first if this failure were starting to happen?
  • Kill criteria: what result would make us stop rather than push on? Agreeing this in advance is a guard against the sunk cost fallacy later.
The owners and early-warning signs are what turn a list of worries into a plan. A recent TechTarget feature on premortems for data modernisation programmes makes the same point, suggesting 8 to 12 cross-functional participants and ending with mitigations, owners and early-warning indicators rather than a bare risk list. Record the output alongside the decision in a decision journal, so the eventual post-mortem can check which risks were foreseen.

Three Worked Examples

How a pre-mortem looks in different contexts

Project management. A product team is about to commit to a six-month redesign. The pre-mortem prompt: "It's twelve months from now. The redesign launched, the bounce rate doubled, and the CEO has told us to roll it back. Why?" Plausible failure stories typically surface: the design was not tested with the segment that drives most revenue; the engineering team underestimated migration of legacy content; the launch coincided with a competitor's release; success metrics were not agreed before kickoff. Each story points to a concrete mitigation - a usability test before code, a content audit, a pre-defined rollback metric. Investing. An investor is sizing a position in a single mid-cap stock. The pre-mortem prompt: "It's three years from now. The position lost 60% of its value. Why?" Common failure modes: the thesis depended on a single regulatory outcome that didn't land; management changed and capital allocation deteriorated; a customer concentration risk that wasn't priced; macro conditions shifted in a way the base rate would have predicted but the optimistic case ignored. Several of these connect directly to expected-value thinking - they re-weight the downside scenario that the original analysis under-counted. Startup product launch. A founding team is preparing to ship a new SaaS product. The pre-mortem prompt: "Eighteen months in. We've burned through the round, ARR plateaued, and we're winding down. Why?" Failure stories tend to cluster around: the wedge use case was real but too narrow to grow; sales cycle was longer than the runway assumed; the founder spent most of their time on engineering instead of distribution; an enterprise feature shipped that no buyer asked for. The mitigations rarely change the headline strategy. They change which milestones the team commits to and how quickly capital is deployed.

Pre-Mortem vs Post-Mortem

Same questions, different timing

A post-mortem looks at a project that has actually failed and asks why. It is essential - but it has two limitations. First, it happens after the cost is already paid. Second, it is contaminated by hindsight bias: once you know the outcome, the causes feel obvious, even when they were genuinely unforeseeable in advance. A pre-mortem keeps the analytical structure of a post-mortem and moves it to a point where the findings can still change the plan. The team gets the diagnostic value without the irreversible loss. Pre-mortems and post-mortems are complementary: run the pre-mortem before launch, run the post-mortem after, and feed any lessons into the next pre-mortem template.

Common Mistakes

How a pre-mortem goes wrong

Treating it as a blame exercise. The room must understand that this is a hypothetical, not a referendum on the leader. If sponsors get defensive, the criticism that needs to come out won't. Skipping the silent writing phase. Once anyone speaks, everyone anchors. The independent generation step is what produces a diverse list of risks - collapsing it into open discussion early throws away the main benefit. Generating risks but not acting on them. A list of failure modes that doesn't change anything is worse than no list at all - it gives the team the feeling of having addressed risk while leaving the plan untouched. Each significant risk should map to a mitigation, an accepted trade-off, or an explicit kill criterion. Inviting only the inner circle. The most useful failure stories often come from people closest to execution: the engineer who knows the migration is brittle, the rep who knows the client base. Excluding them turns the pre-mortem into a managerial echo chamber.

When NOT to Run a Pre-Mortem

The technique has limits

Pre-mortems are valuable when a decision is large, partially reversible, and made under uncertainty. They are overkill - and can be obstructive - when the decision is small, easily reversed, or genuinely time-critical. They are also weaker for decisions where the bottleneck is information rather than judgement. If your team genuinely doesn't know whether the underlying market exists, a pre-mortem will surface the question but won't answer it. The right response there is a small, real test: a customer interview, a paid pilot, a prototype. The pre-mortem helps you decide which tests to run, not whether to run them.

Pre-Mortems and Probabilistic Thinking

Where the technique fits in a wider toolkit

A pre-mortem is one of several tools that push decision-makers from optimistic point estimates towards a more honest distribution of outcomes. Expected-value reasoning forces you to weight outcomes by probability. Second-order thinking forces you to consider downstream consequences. The pre-mortem complements both: it improves the inputs by surfacing failure paths that optimistic planning suppresses, and it acts as a structural defence against confirmation bias by forcing you to seek evidence against your own plan before you commit. Kahneman explicitly framed the pre-mortem as a debiasing tool against the planning fallacy and excessive optimism. Used regularly, it tends to make teams more calibrated about timelines, budgets, and risk-adjusted upside. It does not eliminate uncertainty - but it makes the uncertainty that matters legible enough to plan around.

Frequently Asked Questions

Is it pre-mortem, premortem or pre mortem?
All three mean the same thing. Gary Klein's original 2007 Harvard Business Review article spells it premortem; pre-mortem with a hyphen is the most common spelling in British usage.
How long does a pre-mortem take?
Klein says as little as 20 to 30 minutes, with about two minutes of silent writing at the start. Larger decisions with more people often take up to an hour, mostly spent on reading out, rating and turning the failure modes into mitigations.
Who should be in the room?
The decision team plus anyone with substantive context on the execution. Klein recommends including the people who will actually do the work, not only the sponsors. Diversity of perspective is what surfaces the most useful failure modes.
What's the difference between a pre-mortem and a risk register?
A risk register is a static document; a pre-mortem is a generative exercise. The pre-mortem produces the items that should go on the risk register, then forces the team to act on them. Most risk registers are too sanitised to capture the real concerns the pre-mortem surfaces.
Does a pre-mortem need a facilitator?
It helps. A neutral facilitator protects the silent writing phase from being interrupted and stops the discussion turning into a defence of the plan. For larger decisions, the facilitator should not be the project sponsor.
How often should a team run pre-mortems?
Whenever a decision is large, partially irreversible, and uncertain. For most teams that's the start of any significant initiative - a major hire, a product launch, a sizable investment, an acquisition. Routine decisions don't need one.
Can pre-mortems become a checkbox exercise?
Yes - and that's the most common failure mode. The defence is to insist that every meaningful risk is paired with either a mitigation, an explicit acceptance, or a kill criterion. If the pre-mortem doesn't change the plan, the plan was either perfect or the exercise was performative.
Is the pre-mortem evidence-based?
Partly. The mechanism, prospective hindsight, comes from Mitchell, Russo and Pennington (1989), who found that treating an outcome as certain produced richer explanations of it - not, as often claimed, a 30% gain in identifying risks. Klein's 2010 study with 178 students found the pre-mortem reduced overconfidence more than other critique methods. Kahneman discusses it favourably in Thinking, Fast and Slow.

Sharpen your decision-making toolkit

Pre-mortems work best alongside expected-value reasoning and an honest view of base rates. Start with our explainer.