METHODS / RISK, CHANGE, AND IMPROVEMENT / PROJECT PREMORTEM

METHOD

Project Premortem

A project premortem asks a team to assume that a planned initiative has already failed and explain what caused the failure. People generate reasons independently before the group discusses them. The team then strengthens the plan, assigns controls, and identifies signals that should trigger action.

WHERE IT COMES FROM Gary Klein in Harvard Business Review, 2007

WHAT IT HELPS WITH

A leadership offsite. A board cycle or major investor process. An executive transition. A launch, an acquisition integration, or any plan with several teams and unfamiliar risks, still early enough to change.

TYPICAL SCOPE Team
SETUP EFFORT Light
ONE MATERIAL LIMITATION

The exercise creates value when priority risks lead to a change in the plan, a preventive action, an owner, and a review date.

WHAT IT IS

Why the sequence matters.

Gary Klein developed and popularized the project premortem, drawing on research into prospective hindsight. Asking people to imagine that an outcome has already occurred can change how they identify and explain possible causes.

Participants write independently before group discussion. The group then collects causes, identifies warning signals, prioritizes material risks, and changes the plan. Independent writing helps preserve perspectives that might otherwise be shaped by the first or most senior speaker.

Silent first, pooled second, and every surviving risk leaves with an owner.

FIT

When a project premortem is useful.

Reach for it when

  • You are planning an executive or leadership offsite.
  • You are preparing a board cycle or major investor process.
  • You are managing an executive transition or organization change.
  • You are launching a new operating cadence, tool, or workflow.
  • You are committing to a project with several teams, dependencies, or unfamiliar risks.

Choose another approach when

  • The work is routine and stable, and an SOP or checklist handles it better.
  • The idea has not become a real plan yet.
  • The incident is already happening.
  • The matter is legal, HR, security, or safety and the right specialist is not in the room.
  • The technical risk needs testing or quantitative analysis beyond the group discussion.
  • The group cannot raise concerns safely. Address that condition before running the exercise.

HOW TO RUN IT

Move from independent thinking to action.

01
Choose a real plan.
Use a plan that is developed enough to examine and early enough to change.
02
Bring the right perspectives.
Include the sponsor, people doing the work, affected functions, and someone who can challenge shared assumptions.
03
Make dissent safe.
Explain the purpose, confidentiality boundaries, respectful conduct expectations, and how concerns will be used.
04
State the future failure.
Use a specific prompt such as: "It is three months after launch. The project failed badly. What happened?"
05
Write silently.
Give each person several minutes to generate causes alone before anyone shares.
06
Share without early debate.
Collect one cause at a time around the group before debating them. This gives several perspectives a chance to surface.
07
Group and prioritize.
Group similar causes, identify early warning signals, and prioritize the risks the team can influence.
08
Change the plan.
Give each priority risk a preventive action, a contingency, an owner, and a review date.
09
Escalate residual risk.
Bring risks outside the team's authority to the person or governing body that can accept or address them.
10
Revisit at milestones.
At agreed milestones, check whether warning signals appeared and whether the controls still fit the plan.

THE WORKSHEET

What to record for each priority risk.

EXAMPLE

Strengthening a leadership offsite.

The team imagines it is three months after a leadership offsite and nothing improved. Independently written causes include unclear decision rights, a late pre-read, CEO changes to the agenda, no safe route for dissent, decisions with no owners, and a failed hybrid setup.

Before the event, leadership names decision owners, freezes the pre-read date, opens an anonymous risk channel, tests a backup hybrid setup, creates a decision-log structure, and schedules a seven-day owner confirmation. The fictional exercise produces specific changes to the plan and owners for the remaining risks.

KEEP IT USEFUL

Common implementation and participation considerations.

Common implementation issues

  • Using a vague failure scenario.
  • Letting the senior leader speak first.
  • Priming the group with the facilitator's own risks.
  • Generating a large list with no mitigation or owner.
  • Turning the exercise into catastrophizing.
  • Punishing the person who raises a political risk.
  • Treating the exercise as a replacement for evidence or testing.
  • Filing the output without changing the plan.

Participation and hierarchy

  • Hierarchy, group similarity, and a leader's attachment to the plan narrow what gets said.
  • Use silent or anonymous input and neutral facilitation.
  • Invite external evidence where the group is assuming things about customers, technology, law, or safety.

AI AND JUDGMENT

Use AI after the independent risk round. People test the risks and change the plan.

AI can help

  • Group similar causes after participants complete their independent work.
  • Compare the list with approved lessons, incidents, or risks from similar projects.
  • Flag risk categories the group may not have discussed.
  • Draft the risk and action record with proposed owners and review dates.

People remain responsible for

  • Generating the first round of risks independently.
  • Judging which concerns are credible and material.
  • Choosing controls, contingencies, owners, and escalation routes.
  • Deciding which risks to accept and what must change in the plan.

Do not show AI-generated risks before the independent human round. Early examples can narrow the group's thinking, and AI suggestions remain prompts until the team verifies them.

SOURCES

What the research supports.

Independent evidence

Research on prospective hindsight suggests that imagining an outcome as already having happened can help people generate reasons for it. Later research and practitioner guidance apply that idea to project-risk discussions. A premortem surfaces possibilities; the team still needs to verify material risks and change the plan.

View sources (5)OpenClose
Mitchell, Russo, and Pennington, the prospective-hindsight research, 1989. Source, checked August 29, 2026.
Gary Klein, Performing a Project Premortem, Harvard Business Review, 2007. Source, accessed August 29, 2026.
Veinott and colleagues, the plan-confidence experiment, 2010. Source, accessed August 29, 2026. Examines what the exercise does to a team's confidence in its own plan.
AHRQ, the Premortem Tool in clinical quality work. Source, accessed August 29, 2026. An applied worksheet version for clinical quality work.
Premortem use in game-development teams, 2023. Source, checked August 29, 2026.

WHERE IT FITS

What this pairs with.

LAST REVIEWED: AUGUST 30, 2026 Every claim is checked against the source it names, and where the evidence does not support a claim we say so. Spotted a problem? Tell us and we correct it in the open.