Decision-making & prioritisation
Pre-mortem
Before committing to a plan, imagine it is a year on and the plan has failed outright, then have each person independently write down why. Prospective hindsight legitimises the doubts a kickoff meeting suppresses, and it costs an hour.
Also known as Premortem, Project pre-mortem, Prospective hindsight exercise. First set out by Gary Klein in 2007; the primary source is cited in full below.
- Format
- Checklist / audit
- Level
- Team · Business unit · Product
- Best for
- Assess risk · Plan execution
- Decision stage
- Decide · Plan
- Difficulty
- Introductory
- Time to apply
- 60 to 90 minutes for the session; allow another hour to fold the findings back into the plan.
Plate · The model
The components
Assume the plan has failed
The framing move that distinguishes a pre-mortem from a risk brainstorm: the failure is stipulated as certain and already in the past, which switches participants from defending the plan to explaining an outcome. The research basis is prospective hindsight, where imagining an event as having occurred increases the ability to generate reasons for it, by around 30 per cent in the original study.
Signals of strength
The facilitator states failure as fact, not as possibility · Participants speak in the past tense about the failure · Nobody is defending the plan during the exercise
Generate the reasons independently
Silent, individual writing of failure explanations before any discussion. This is where the exercise mines what the room privately knows: independent generation defeats anchoring, groupthink and the deference that keeps juniors quiet in open discussion.
Signals of strength
Several minutes of genuine silence with everyone writing · Reasons include internal and political causes, not just external events · The quietest people produce items nobody else had
Consolidate and rank
Pooling every reason without early evaluation, clustering duplicates, then ranking by plausibility and damage. The step turns raw candour into a prioritised weakness list short enough to act on.
Signals of strength
Every reason recorded before any is debated · Clusters named as failure modes rather than complaints · A top handful agreed rather than a forty-item wallpaper list
Strengthen the plan
The payoff: concrete revisions, mitigations with owners, and early-warning tripwires for the ranked failure modes, with the list revisited as execution proceeds. Without this step the exercise was theatre, and the team knows it.
Signals of strength
The plan document actually changes · Each top failure mode has an owner and a tripwire · The list resurfaces at later reviews rather than dying in the minutes
When it earns its keep
- A significant plan is about to be approved, the room has converged, and dissent has gone quiet. That silence is precisely the condition the exercise is built to break.
- The team is deep in the plan and can no longer see it from outside. People invested in a plan suffer from overconfidence in it, and the pre-mortem gives scepticism official cover.
- Junior or peripheral people know things about the plan's weaknesses that they will not volunteer against a senior sponsor. Independent written contributions flatten the hierarchy for twenty minutes.
- You want risk identification that is cheap, fast and needs no software, standing agenda slot or consultant.
And when it doesn't
- The decision is already irreversible. A pre-mortem on a signed contract produces regret, not improvement; run it while the plan can still change.
- There is no plan yet. The exercise stress-tests a specific, stated plan; run against a vague intention it generates vague failures.
- The real need is quantified risk exposure for governance, likelihoods, impacts, owners, mitigations tracked over time. Feed pre-mortem output into a risk matrix; the exercise itself does not manage anything.
- Leadership intends to ignore the output. Soliciting failure stories and then filing them teaches the team that candour is decorative, which is worse than not asking.
How to run it
Before starting, gather the inputs the analysis depends on:
- A concrete plan, written down and circulated, that participants genuinely understand: objective, approach, timeline, resources.
- The right people in the room: those who will execute the plan, plus at least some who know the terrain but did not write the plan.
- A facilitator willing to hold the frame, the plan has failed, it is certain, explain why, and to stop early drift into fixing before the failure list is complete.
- Psychological licence from the sponsor: an explicit statement that the exercise exists to find weaknesses and that naming them is a contribution, not disloyalty.
- 1
Brief the plan, then declare it dead
Ensure everyone understands the plan, then set the frame exactly: imagine we are a year or two out, the plan has been carried out, and it has failed, completely and embarrassingly. The certainty matters. Mitchell, Russo and Pennington found that treating the outcome as having already happened, rather than as merely possible, is what unlocks the richer explanations; Klein's design rides on that effect.
- 2
Generate the reasons independently
Each person writes their own list of reasons for the failure, silently, for several minutes, before anyone speaks. Independence is load-bearing: it prevents anchoring on the first voice and protects the quiet expert's contribution. Expect the plausible-but-unspoken reasons, the sponsor loses interest, the integration partner underdelivers, we staffed it with whoever was free.
- 3
Consolidate and rank
Go around the room, one reason per person per pass, recording everything before evaluating anything. Then cluster duplicates and rank by the combination of how damaging and how plausible each failure story is. The ranking converts a cathartic list into an agenda.
- 4
Strengthen the plan
For the top-ranked failures, decide what changes: revise the plan, add mitigations and owners, define tripwires that would signal the failure beginning, or in the honest extreme, abandon the plan. Klein also suggests periodically revisiting the list during execution to keep the team sensitised to early signs.
Reading the result
A ranked list of plausible failure modes for a specific plan, generated before commitment, together with the resulting plan revisions, mitigations, owners and early-warning tripwires, and a calibrated, usually lower, level of collective confidence.
- Weight failure modes the team can influence over external catastrophes. 'The sponsor moved on' is actionable; 'a recession hit' mostly is not, though it may still deserve a tripwire.
- Reduced confidence is a success condition, not a side effect. The measured effect of the exercise in controlled comparisons is precisely the deflation of overconfidence in the plan.
- Notice what the list says about the organisation. If half the failure stories are political, the plan's biggest risk is not in the Gantt chart.
A worked example
A museum pre-mortems a major touring exhibition
A regional UK museum is planning its most ambitious project in a decade: a nine-month international touring exhibition with £1.2m of costs, borrowed objects from three overseas lenders, a corporate sponsor covering a third of the budget, and a visitor target of 180,000. Two weeks before trustees sign off, the director runs a 90-minute pre-mortem with curators, operations, fundraising, marketing and front-of-house.
- Assume the plan has failed
- The director sets the frame: it is two years on, the exhibition has closed, and it was a failure people mention carefully at conferences: 70,000 visitors, a six-figure deficit, a strained lender relationship. Stating it as accomplished fact visibly changes the room; the marketing lead, cheerleader-in-chief for months, starts writing first.
- Generate the reasons independently
- Ten minutes of silent writing produces 41 reasons. Alongside the expected ones (shipping delays, indemnity problems) come items only insiders held: the visitor target was copied from a metropolitan venue with triple the catchment; the sponsor's contract lets them withdraw at rebrand; ticket pricing assumed families the programming does not actually attract; front-of-house cannot staff the late openings the business case quietly relies on; the registrar, on whom all three loans depend, is one person with no cover.
- Consolidate and rank
- Clustering yields nine failure modes. Ranked by plausibility times damage, the top four are: the overcooked visitor projection, single-point dependence on the registrar, sponsor withdrawal, and late object arrival compressing installation. Notably, the risks the original plan documented at length, fire, theft, damage, rank near the bottom; they were the insurable, thinkable ones.
- Strengthen the plan
- The plan changes before sign-off: the visitor target is rebased to 120,000 with a break-even case at 95,000; a freelance registrar is retained on call; the sponsorship contract gains a staged-payment schedule and a replacement-sponsor clause; installation gains a two-week buffer and a reduced-show fallback layout. Tripwires with owners: advance ticket sales at 12 weeks, loan paperwork milestones, sponsor payment dates.
The read. Trustees approve a plan that is £150,000 more conservative and considerably more honest than the one that entered the room. The exercise cannot claim to have prevented anything yet, and that is the correct reading: its demonstrated value was deflating a projection everyone privately doubted and surfacing a staffing dependency no risk register held. Whether the tripwires get watched is now a management question the pre-mortem cannot answer.
Pitfalls
- Softening the frame to 'imagine things went wrong'. Without certain, stipulated failure the exercise collapses into an ordinary risk discussion and loses the prospective-hindsight effect it exists to exploit.
- Skipping silent independent writing and going straight to discussion, which restores the anchoring and hierarchy effects the method is designed to defeat.
- Letting the room start solving after the first juicy failure mode. Generation and evaluation must stay separated or the list stops at three items.
- Running it after commitment, as a compliance ritual. The exercise only pays when the plan can still change, and teams can smell the difference.
- Producing the list and never revisiting it. The follow-through, changed plans, owners, tripwires, is the deliverable; the workshop is just how you get it.
- Using it to kill a plan someone already opposed. A pre-mortem stacked with opponents is politics wearing a method.
What the critics say
The direct empirical base is thinner than the technique's popularity suggests. The founding studies established prospective hindsight in individual lab tasks, and the most-cited direct test of the pre-mortem itself measured reduced plan confidence among students, not improved real project outcomes. Controlled field evidence that pre-mortems change project results remains scarce.
Veinott, B., Klein, G. and Wiggins, S. (2010) 'Evaluating the Effectiveness of the PreMortem Technique on Plan Confidence', Proceedings of the 7th International ISCRAM Conference, Seattle.
Outcomes are heavily facilitation-dependent. The exercise's protections, stipulated failure, silent generation, deferred evaluation, are exactly the parts unskilled facilitators drop, and in low-trust teams a candour exercise can surface blame rather than information. The literature offers little guidance on these boundary conditions.
Klein, G., Koller, T. and Lovallo, D. (2019) 'Bias busters: Premortems: Being smart at the start', McKinsey Quarterly, April 2019.
Debiasing effects of this family of interventions are situational rather than guaranteed: research on related consider-the-opposite techniques shows effects that vary with task and can backfire, for instance when generating many reasons proves difficult and the difficulty itself is read as evidence the plan is sound.
Sanna, L. J., Schwarz, N. and Stocker, S. L. (2002) 'When debiasing backfires: Accessible content and accessibility experiences in debiasing hindsight', Journal of Experimental Psychology: Learning, Memory, and Cognition, 28(3), pp. 497–502.
Sources and further reading
- Klein, G. (2007) 'Performing a Project Premortem', Harvard Business Review, 85(9), September 2007. ↗
- Mitchell, D. J., Russo, J. E. and Pennington, N. (1989) 'Back to the future: Temporal perspective in the explanation of events', Journal of Behavioral Decision Making, 2(1), pp. 25–38.
- Veinott, B., Klein, G. and Wiggins, S. (2010) 'Evaluating the Effectiveness of the PreMortem Technique on Plan Confidence', Proceedings of the 7th International ISCRAM Conference, Seattle. ↗
- Klein, G., Koller, T. and Lovallo, D. (2019) 'Bias busters: Premortems: Being smart at the start', McKinsey Quarterly, April 2019. ↗