Pre-Mortem

A pre-mortem is the inverse of a post-mortem. Before a project begins (or at a major milestone), the team imagines the project has already failed — catastrophically — and works backward to identify what went wrong. The framework exploits a well-documented psychological effect: it is easier for humans to generate reasons for a fictional past failure than to speculate about a hypothetical future failure. The result is a rich, honest list of risks that planning meetings rarely surface.

Origin

Gary Klein — a cognitive psychologist whose research on expert decision-making influenced Daniel Kahneman's work — introduced the pre-mortem in the early 2000s. He published the canonical description in Harvard Business Review in 2007 ("Performing a Project Premortem"). Klein's insight came from studying how experts reason about risk: when asked to speculate about failure, people hedge and speak vaguely. When asked to explain a failure that has already happened (even hypothetically), the same people produce specific, concrete, and honest accounts. The framing change unlocks the cognition.

The Framework

1. Frame the failure:  "Imagine it is [end date]. The project has
                       failed catastrophically. It is a disaster."

2. Silent writing:     Each person writes, individually, every
                       reason they can imagine for the failure.

3. Share and cluster:  Surface all failure modes. Cluster by theme.

4. Prioritize:         Rank failure modes by likelihood x impact.

5. Mitigate:           For each top risk, agree on a mitigation,
                       an owner, and a trigger ("we act when X").

6. Integrate:          Fold the mitigations into the plan itself.

The Specific Framing

The framing is the whole move. Compare:

Weak framing (standard risk ideation):
  "What are the risks with this project?"
  Result: generic, cautious responses. "Scope might creep." "We might
          have technical issues."

Strong framing (pre-mortem):
  "It is six months from now. The project has failed in spectacular
   fashion. The news is bad. We got dragged into the all-hands to
   explain what went wrong. Write down, privately, every reason you
   can imagine."

  Result: specific, concrete responses. "The data migration produced
          silent corruption we didn't detect for 3 months." "The PM
          left and nobody knew the customer requirements." "We
          underestimated the legal review by 8 weeks."

The imagined past framing unlocks candor that the speculative future framing does not. This is the entire claim of the pre-mortem, and it is well supported by psychological research on prospective hindsight.

How to Use It

When to Run It

- At project kickoff (before the plan is locked)
- At major milestones (before committing to the next phase)
- Before high-risk deploys or launches
- Before a reorg
- Before a major hire (especially senior leadership)
- Before a strategic pivot or investment

Preparation

1. Scope the project clearly. "The Q4 pricing migration" — not
   "our pricing work."
2. Set a specific end date for the imaginary failure. "It is
   February 1st, the migration shipped in January, and it's a
   disaster."
3. Invite 5-15 people who have diverse views of the project:
   engineering, product, design, legal, ops, customer-facing.
4. Set up silent writing (shared doc or stickies) for the data
   gathering.

Running the Pre-Mortem

1. Framing (5 min):
   "It's [date]. The project has failed catastrophically. Take 10
    minutes, silently, and write down every reason you can imagine
    for why it failed. Be specific. No reason is too small."

2. Silent writing (10 min):
   Every participant writes their list. No discussion.

3. Round-robin share (15-20 min):
   Each person shares one item at a time, rotating. Others can
   add. Facilitator captures on a visible board.

4. Cluster (10 min):
   Group similar items. Name each cluster.

5. Prioritize (15 min):
   Dot-vote or rate likelihood x impact for each cluster.

6. Mitigate (20 min):
   For top 3-5 clusters, define:
   - Preventive actions: what we do now to reduce likelihood
   - Detection: how we'd know it's happening
   - Response: what we do if it does happen
   - Owner and trigger date

7. Integrate (5 min):
   Agree how mitigations go into the project plan.

A thorough pre-mortem is 60-90 minutes. Shorter is possible for smaller projects (20-30 min), but the silent writing and round-robin share steps should not be skipped.

Tech & Company Example

A fintech company is about to launch a major pricing migration — moving from a flat-rate plan to a usage-based plan, affecting all 40,000 customers. Pre-mortem:

Framing: "It is 90 days post-launch. The migration has failed.
         Churn is 3x baseline. Press is bad. The CFO is furious.
         Write, individually, every reason you can imagine."

Silent writing produces (after clustering):

Cluster 1 — Customer experience (8 items)
  - Surprise bills from usage spikes that customers didn't model
  - Unclear migration timeline; customers caught off guard
  - Self-service tools didn't actually let customers predict costs
  - Support team overwhelmed; response times >72 hours

Cluster 2 — Data / Billing integrity (6 items)
  - Usage metering was off by 15% silently for 6 weeks
  - Double-billing on migration day for some customers
  - Refund process was manual and overwhelmed support
  - Historical invoices did not match new system in some cases

Cluster 3 — Stakeholder comms (5 items)
  - Sales team learned about the migration from customers
  - Account managers had no answer when customers asked about cost
  - Investor relations team was not briefed on the model change
  - Press statement was reactive, not proactive

Cluster 4 — Legal/compliance (4 items)
  - Terms of service change required explicit consent in EU;
    we didn't have flow
  - Enterprise contracts had pricing locks we overlooked
  - Tax calculation changes for some jurisdictions not validated

Cluster 5 — Technical (3 items)
  - Pricing calculation code had edge cases around prorated
    contracts
  - Metering pipeline was not load-tested at 10x current volume
  - Migration rollback path was not actually tested end-to-end

Prioritized mitigations:

1. Customer experience:
   - Owner: CS lead. Build a cost-calculator tool, make it free
     for 60 days pre-migration. Trigger: CS can test it by D-45.

2. Data integrity:
   - Owner: Data eng. Run metering reconciliation for 90 days
     pre-launch. Launch if error <0.1%. Trigger: reconciliation
     pass by D-30.

3. Sales/stakeholder comms:
   - Owner: GTM lead. Sales training D-60, account manager playbook
     D-45, investor briefing D-30. Trigger: dates.

4. Legal:
   - Owner: Legal. Consent flow for EU + validate enterprise contract
     overrides. Trigger: legal sign-off D-21.

5. Technical:
   - Owner: Eng lead. Load test 10x, rollback test end-to-end.
     Trigger: both pass by D-30.

Without the pre-mortem, many of these risks — especially the stakeholder comms and legal ones — are routinely missed until they are already problems. The pre-mortem surfaced them weeks before launch.

When It Works

Complex projects with multiple failure modes
High-stakes launches, deploys, migrations
Strategic decisions with significant reversibility cost
Projects with cross-functional dependencies
Anywhere teams tend toward over-optimism

When It Does Not Work

Routine work — Daily work does not benefit from pre-mortem overhead.
Teams with low psychological safety — If people cannot honestly name risks, the pre-mortem produces sanitized, useless output.
Purely experimental work — If the goal is to learn and fail cheaply, pre-mortem framing is misaligned; you want the failure data.
Leader-dominated culture — If the leader's view of risk dominates, silent writing fails to surface divergent views.

Common Failure Modes

Framing

Vague failure scenario — "The project doesn't go well." Not specific enough to unlock cognition. Must be concrete: "It failed. We're in an emergency all-hands. The CFO is angry."
Not making it personal — "Imagine risks" is prospective thinking; the magic is in the past-tense framing of a specific catastrophe.

Process

Skipping silent writing — Going straight to group discussion produces anchoring, groupthink, and dominant voices.
Round-robin monopolies — Letting extroverted participants dominate the share phase. Facilitator must enforce rotation.
No prioritization — Surfacing 40 risks and treating them all equally. Mitigation budget is finite; prioritize.

Action

Orphan Mitigations — Identifying risks without assigning owners or triggers. Mitigations only happen when they are someone's job.
Over-mitigation — Trying to mitigate every risk, turning the project into risk theater. The test: would the mitigation survive a cost-benefit review?
Paper Mitigations — "We will be vigilant about X" is not a mitigation. Mitigations must be specific actions.

Cultural

Pre-mortem as Theatre — Going through the motions because "we always do a pre-mortem" without actually engaging with the findings.
The Silenced Risk — A risk is raised but quietly removed from the list because it is politically uncomfortable (often: leadership risk, resourcing risk, or strategic-fit risk). The risks most commonly silenced are the ones most worth hearing.
Retroactive I-Told-You-So — Using the pre-mortem list post-failure to blame people. Destroys future pre-mortems. The list is not an audit.

Pre-parade (pre-mortem's twin) — Imagine the project succeeded spectacularly. What made it succeed? Surfaces key success factors. Often paired with pre-mortem.
Red Teaming — Military-origin practice of assigning a team to actively attack a plan. Deeper than pre-mortem; more expensive.
Devil's Advocate Assignment — Formally assigning one person to argue against the plan. Lightweight cousin of red teaming.
Scenario Planning — Exploring multiple plausible futures rather than just the failure case. Used at strategic horizon.
Bazerman & Watkins — Predictable Surprises — Framework for the class of risks that organizations systematically miss; pre-mortem is one tool to surface them.
Inversion (Munger) — Charlie Munger's general heuristic: "Invert, always invert." Pre-mortem is a specific application to project risk.