decision-making

Pre-Mortem Practice

Also known as:

The pre-mortem — imagining that a decision has already failed and working backwards to understand why — is one of the most effective debiasing techniques for overconfident plans. This pattern covers how to conduct a genuine pre-mortem: separating the imagining from the planning, surfacing dissenting views that politeness suppresses, and using the identified failure modes to strengthen the plan before commitment.

Imagine that a decision has already failed catastrophically, then work backward to identify the specific causes — before commitment locks the path.

[!NOTE] Confidence Rating: ★★★ (Established) This pattern draws on Gary Klein / Decision-Making.

Section 1: Context

Teams and organizations face a recurrent pressure: decisions must be made and committed quickly, yet the confidence with which plans are announced often far exceeds the actual robustness of the reasoning beneath them. In corporate strategy, a new market entry is greenlighted based on market research that felt comprehensive at the time. In public policy, a regulation is drafted to address a perceived crisis, but unintended consequences ripple through communities months later. In activist collectives, a campaign strategy is locked in before dissenting voices have space to surface their doubts. In product teams, a feature roadmap ships with high conviction, then fails to solve the user problem it was meant to address.

The common thread: the system is already committed (or nearly so) before the hardest thinking happens. Overconfidence is baked into the timeline itself. Politeness, hierarchy, and the momentum of planning suppress the knowledge that actually exists within the team — the doubts, the edge cases, the failure patterns people have seen in similar situations. This pattern emerges in systems that need to make decisions under uncertainty but want to preserve the quality of reasoning before stakes are locked in.

Section 2: Problem

The core conflict is Pre vs. Practice.

The tension runs between two necessary acts: imagining what could go wrong (pre) and committing to what will be done (practice).

When teams skip straight to practice — moving from idea to implementation — they inherit all the blind spots of the initial framing. Overconfidence blooms. Known risks go unnamed because naming them feels like admitting doubt. Dissenting views stay quiet because the decision has already been socially sealed; speaking up now feels like undermining the group. The person who saw a similar failure happen elsewhere stays silent because the momentum is forward.

When teams linger too long in imagining — endless scenario-planning, perpetual devil’s advocating — nothing gets decided. The system becomes paralyzed, and decision authority diffuses into cynicism. Imagination without a gate becomes procrastination.

The core wound: a team can be simultaneously overconfident about what will work AND unable to voice what they suspect will break. Both are true. The decision feels locked in just as the critical doubts are arriving. By the time dissent surfaces, it’s too late to reshape the plan without losing face.

Pre-mortem practice creates a structured permission structure to imagine failure before the decision hardens, and to use that imagining to strengthen the actual plan. It separates the act of imagining from the act of committing, making space for the knowledge that politeness and hierarchy suppress.

Section 3: Solution

Therefore, before committing to a major decision, conduct a structured pre-mortem: ask the team to imagine the decision has already been implemented and has failed spectacularly, then work backward to identify the specific failure modes — without immediately fixing them or defending the plan.

The mechanism works because it flips the cognitive burden. Instead of asking “What could go wrong?” (which invites defensive reasoning and abstract worry), you ask “What did go wrong?” in the past tense. This small shift unlocks realism. The past tense makes failure feel inevitable and specific, not hypothetical and vague. It moves the conversation from abstract risk to concrete breakdown.

Crucially, the pre-mortem separates imagining from planning. In the imagination phase, dissent is safe. People are not defending the decision; they are diagnosing a failure that has already happened. The plan is not yet alive; it is already dead. This creates psychological permission to voice doubts that would be suppressed in a normal decision meeting.

The pattern draws on Gary Klein’s research showing that people who have direct experience with failure are better at predicting failure in advance. A pre-mortem borrows that experience-based realism by treating the future failure as already real. The imagining phase surfaces the tacit knowledge — the edge cases, the patterns from past projects, the relationships that might break — that exists scattered across the team but never gets named in a forward-looking meeting.

Once the failure modes are named, the team moves to the planning phase: examining each identified failure, understanding its root cause, and deciding whether to redesign the decision, accept the risk, or mitigate it deliberately. This is where Pre and Practice reconcile. The team commits to the decision with genuine understanding of what can break, and with intentional choices about which risks to carry and which to redesign away.

The vitality the pattern creates is in clarity before commitment. The system keeps its adaptive capacity by refusing to lock in blind confidence.

Section 4: Implementation

Prepare the container. Gather the core decision-makers and stakeholders — the people whose voices will shape what happens next. Announce the pre-mortem explicitly: “We’re going to imagine this decision has failed. We’re not defending the plan right now. We’re diagnosing.” The frame matters. Without it, people will still be defensive.

Create the imagined failure. Spend 10–15 minutes in silence. Ask each person to write down: “Imagine it’s 18 months from now. We committed to this decision. And it has failed completely. What happened? Why did it break?” Silence is essential. Writing is essential. It prevents groupthink and surfaces individual knowledge.

Surface and organize. Go around the room. Each person reads their failure modes. Write them on a visible surface without editing, arguing, or defending. Cluster them into themes as they emerge: stakeholder resistance, technical miscalculation, market shift, relationship breakdown, resource scarcity, misaligned incentives. Do not move to solution yet.

Examine root causes, not symptoms. For each cluster, ask: “If this failed this way, what had to be true before we even decided?” This moves past surface symptoms. A feature launch that “flopped due to low adoption” becomes “we never verified that users actually wanted to switch from their existing workflow.” A policy that “faced backlash” becomes “we designed it without the communities it would affect.”

In corporate strategy: Conduct the pre-mortem before final board sign-off. Name market shifts, talent departures, competitive moves, and relationship damage with specific executives or partners. Use the failure modes to reshape the success metrics and contingency triggers. Decision-makers hear not abstract risk but concrete breakdown.

In government policy: Run the pre-mortem with frontline practitioners — the people who will implement the policy, not just the people who designed it. They will name unintended consequences and enforcement gaps that policy teams miss. Use the failures to write implementation guidance and stakeholder communication before the policy launches.

In activist collectives: Pre-mortem in a circle, with space for quiet voices to contribute fully. Name burnout, internal conflict, mission drift, and co-optation. The failures often expose underlying tensions about power and decision-making that the campaign strategy was masking. Use them to strengthen the collective’s resilience, not just the campaign.

In product teams: Run the pre-mortem with frontline engineers, customer support, and a sample of users or customers if possible. The imagined failures will often name technical debt, user behavior patterns, and competitive responses that product managers missed. Redesign the roadmap sequencing or the success criteria based on what you learn.

Decide what to do. After all failure modes are named, move into deliberate planning. For each major cluster:

  • Redesign the decision to remove the root cause.
  • Accept the risk explicitly and name the trigger that would force a course correction.
  • Add a specific mitigation that reduces (but does not eliminate) the risk.

The team should not move back into defensive reasoning. The decision might shift substantially. That is the point.

Document and track. Write down the failure modes and the decisions about each one. In 6–12 months, return to them. Which ones materialized? Which were wrong? Which did you mitigate effectively? This loop closes the learning cycle and trains the team in realism.

Section 5: Consequences

What flourishes: Teams that practice pre-mortem develop a more mature relationship to uncertainty. They stop confusing confidence with clarity. They discover that dissenting views strengthen rather than undermine decisions — a shift in psychological safety that ripples into future collaboration. The pattern also surfaces distributed knowledge: the engineer who saw a similar architecture fail, the community member who knows resistance patterns, the veteran who has lived through policy backlash. Once surfaced, this knowledge can reshape the plan before it hardens. Decision-makers also develop better calibration: they commit to plans with genuine understanding of what can break, which means they are more realistic about timelines, resource needs, and contingencies. The plan itself becomes more robust because it was designed to handle its own failure modes rather than pretending they don’t exist.

What risks emerge: If pre-mortem becomes routine without genuine imagination, it decays into theater. Teams perform the exercise without actually surfacing doubt. This happens when leaders dismiss the identified failures or when the exercise becomes a compliance box rather than a real gate in the decision process. The pattern can also enable what we might call “risk amnesia”: teams identify risks, document them, then implement the original plan anyway without genuine mitigation. Six months later, the named risks materialize and people say “we knew that was possible” — but the knowledge did not change behavior. Watch especially for: the pre-mortem that surfaces real concerns but the decision-maker overrules them without explanation; the exercise that becomes a venting session but never reaches the planning phase; the identified risks that are written down but never revisited during implementation. Because this pattern sustains existing health rather than generating new adaptive capacity, it can become routine and hollow. The vitality reasoning surfaces this directly: overuse without renewal risks rigidity.

Section 6: Known Uses

NASA’s Mission Planning (1980s–present): NASA embedded pre-mortem-style “failure mode analysis” into mission planning after early failures in the Apollo program. Engineers were explicitly asked to imagine a mission had failed and to work backward to its causes. This practice surfaced the issues that led to the redesign of the Apollo 13 oxygen system and the structural weaknesses that the Challenger investigation later examined. The pattern is now woven into NASA’s culture: no mission proceeds without explicit identification and mitigation of failure modes. The practice created a shared language around risk and prevented overconfidence from locking in fatal assumptions.

The Veterans Health Administration’s Policy Implementation (2010s): When the VA rolled out major IT system changes and policy shifts, they faced recurring failures where frontline clinics and patients experienced unintended consequences that central planners had not anticipated. A VA leader trained in Gary Klein’s work introduced pre-mortems into policy development. Frontline staff — nurses, administrators, billing clerks — were asked to imagine a new policy had failed and to diagnose why. They named workflow bottlenecks, communication gaps, and patient safety risks that policy designers had missed. The policy was redesigned to account for these realities before rollout. Subsequent implementations had far fewer disruptions. The practice shifted who was included in the imagining phase, not just the planning phase.

Airbnb’s Product Decision Framework (2015): As Airbnb scaled, product teams faced the recurring risk of shipping features that sounded good in planning meetings but created host friction or guest confusion. A product leader introduced pre-mortems into the decision framework. Before greenlight, the team imagined a feature had shipped and failed — low adoption, support volume spikes, host complaints. The imagining surfaced use cases the team had not considered and assumptions about user behavior that turned out to be wrong. One pre-mortem for a trust-and-safety feature identified that the implementation would create a two-tiered system that would alienate new hosts. The redesign addressed this upstream, preventing a launch that would have generated significant churn.

Section 7: Cognitive Era

In an age of distributed intelligence and AI-assisted decision-making, pre-mortem practice faces new pressures and new possibilities.

New pressures: AI systems can project scenarios and identify risks at scale, which can create false confidence that the “hard thinking” about failure has already been done by algorithm. Teams may skip the pre-mortem because they have run a thousand Monte Carlo simulations. But simulation cannot replace the tacit knowledge of people who have lived through related failures. AI can surface statistical patterns; it cannot yet surface the social and psychological breakdown modes that emerge from how humans actually respond to change. The pre-mortem becomes more necessary in an AI-augmented environment, not less, because the gap between what systems can calculate and what teams can imagine keeps widening.

New leverage: AI can accelerate the imagining phase. Language models can generate plausible failure scenarios based on historical data, then the team can evaluate and refine them. Tools can help track failure modes and their mitigations across a portfolio of decisions, creating a learning loop that would be too burdensome to maintain manually. Product teams can test imagined failures with rapid prototyping and user simulation before committing to implementation.

New risks: If pre-mortem becomes delegated to an AI system, it loses its primary value: surfacing the distributed tacit knowledge and building psychological safety within the team. An AI-generated list of failure modes is not the same as a team member naming the failure mode they saw happen at their previous company. The relational work — the alignment, the trust-building, the permission to dissent — gets bypassed. Organizations might also use AI-generated failure scenarios as a substitute for dialogue, reinforcing hierarchy rather than surfacing dissent.

The pattern’s value in the cognitive era rests on keeping it human-centered: use AI to augment and accelerate the imagining phase, but keep the dialogue real. The pre-mortem remains a practice of collective sense-making, not a technical exercise in risk quantification.

Section 8: Vitality

Signs of life:

  • Dissenting views surface before the decision is locked in, not after. People name their doubts explicitly and find they are heard.
  • The team redesigns the plan based on identified failure modes. At least one significant assumption or sequence changes as a result of the pre-mortem.
  • Six months into implementation, people reference the pre-mortem. “We knew this risk might happen — here’s the mitigation we designed for it.” The imagined failures inform real decisions.
  • Psychological safety increases. People feel it is safe to voice doubt and uncertainty, not just confidence and forward momentum.

Signs of decay:

  • The pre-mortem becomes a checkbox. The team completes it, writes down generic risks, then implements the original plan unchanged. The exercise becomes theater without consequence.
  • Identified risks are dismissed or overruled by leadership without explanation. People learn their voice doesn’t matter and stop bringing real concerns.
  • The pre-mortem surfaces legitimate concerns, but the team feels trapped between “we have to decide” and “we know this will break.” Resignation replaces realism.
  • The practice becomes routine without renewal. The same failure modes are named each time because nothing was actually learned from the last implementation cycle.

When to replant: If you notice decay — the pre-mortem becoming hollow, or the imagined failures not changing behavior — pause the practice for one full decision cycle. Instead, conduct a “post-mortem” on a recent decision: look back at what actually happened, compare it to what the pre-mortem imagined, and ask why the gap. Did the pre-mortem miss something? Did the team know the risk but not mitigate it? Did leadership override the imagined failure? Let the post-mortem teach you how to redesign the pre-mortem itself. Replant with fresh framing, different participants, or a clearer gate for how identified failures actually shape the decision.