Pre-mortem: How to Spot Project Risks Before They Happen
By the end of this lesson you'll be able to:
- Explain what a pre-mortem is and why it surfaces risks a normal planning meeting misses
- Run a pre-mortem in five steps and rank the results by likelihood and impact
- Turn the top risks into prevention steps with named owners
Quick answer: what is a pre-mortem?
A pre-mortem is a short team exercise held before a project starts: everyone imagines that the project has already failed and writes down why, so the team can act on the biggest risks while there is still time. In this Kanbanchi lesson you run one and study Denver airport and the Mars Climate Orbiter.
How to run a pre-mortem, in 8 slides
One idea per slide. Use the buttons, the dots or your arrow keys. The full text of every slide is right below the player.
Lesson 8 slides
8 slides
The full lesson: the pre-mortem, step by step
What is a pre-mortem?
A pre-mortem is a short team exercise held before a project starts: everyone imagines that the project has already failed and writes down why, so the team can act on the biggest risks while there is still time. A post-mortem asks why a project failed. A pre-mortem asks the same question while you can still change the answer. The idea is generally credited to the psychologist Gary Klein, and it fits into step 4 of the project management process, when you list the main risks.
A project risk is an uncertain event or condition that, if it happens, affects at least one project goal, such as scope, time or cost. A pre-mortem is one of the fastest ways to find them, because it uses the knowledge that is already in the room.
Why it works
Saying "it has already failed" makes it safe to name problems. People stop defending the plan and start explaining what went wrong. Quiet worries that nobody would raise in a planning meeting come out as reasons. In a normal planning meeting the question is "what could go wrong?", and optimism usually wins. In a pre-mortem the failure is a given, so the only question left is why.
Step 1: Set the scene
Before the work starts, bring the team and key stakeholders together for about an hour. Recap the goal, the plan and the deadline. Invite the people who will do the work, not only the managers: they usually know where the plan is weakest.
Step 2: Imagine it has failed
Say: "It is a year from now. The project has failed, badly. What happened?" Use a real date and make the failure concrete: the launch slipped by three months, the budget ran out, or the customer refused to sign off.
Step 3: List the reasons
Everyone writes reasons alone for a few minutes, then shares one at a time until the list is complete. Writing alone first stops the loudest voice from setting the agenda. Good reasons are specific: "the vendor delivered the data two months late" is useful, "tech problems" is not.
Step 4: Rank them on a project risk matrix
A project risk matrix is a grid that scores each risk by its likelihood and its impact, so the team can see at a glance which risks to act on first. Score each reason for likelihood and impact from 1 to 3 and multiply. The highest scores come first. A score of 6 or 9 means act now, 3 or 4 means plan a response, and 1 or 2 means keep an eye on it. A risk matrix does not make the decision for you, but it makes the conversation concrete: the team argues about one number at a time instead of about the whole plan.
A simple project risk matrix: score equals likelihood times impact.
Step 5: Act early
For each top risk, agree a prevention step, one named owner and a warning sign to watch for. Then put them into the plan. A prevention step should be an action someone can do, such as "run a trial migration in week 2", not a wish such as "be careful with the data". A risk you decide to accept still needs an owner and a fallback.
Pre-mortem vs risk register vs retrospective
A pre-mortem finds risks before the project starts. The risk list tracks them while it runs. A retrospective looks back at what actually happened. The three work together: the pre-mortem fills the risk list, the risk list is reviewed in monitoring and controlling, and the retrospective at closing checks which risks were missed. Once you know your top risks, decide what to do first with Lesson 7: How to prioritize tasks, and put the prevention steps on the schedule from Lesson 6.
Case studies: Denver airport and the Mars Climate Orbiter, failures a pre-mortem could have named
Denver International Airport is a project where the reasons for failure were written down before the contract was signed. All facts below come from Kanbanchi's Project Failures page on the Denver airport baggage system, which is based on the US General Accounting Office (GAO) chronology.
Project Failures case study 1
Denver airport baggage system, 1990 to 1995
Real story · United States · Hidden dependencies
The warnings were on paper before the contract
In 1990, two consultants hired by the City of Denver raised doubts. In April, one said there was not enough time to design, install and test an airport-wide automated system by October 1993. In August, the second, Breier Neidle Patrone Associates, said it did not believe such a system could be built within the schedule. On that advice, the city chose a conventional tug-and-cart system.
In July 1991, the city reversed course and decided to build an integrated, airport-wide automated system after all. In November 1991, the airlines' baggage subcommittee wrote that, according to the master project schedule, the baggage system would not be finished until mid-1994. In February 1992, several airlines wrote again, questioning whether an integrated system was feasible. In May 1992, the city signed a final contract of $195.6 million with BAE Automated Systems.
What happened next
The city issued numerous change orders, including relocating baggage carousels and adding conveyors. The opening moved from October 1993 to late December 1993, then to March 1994, then to May 1994, and in May 1994 it was postponed indefinitely. The first large-scale test, in February 1994, had major failures. A conventional backup system was contracted only in September 1994. The airport opened in February 1995, and the baggage system's cost had risen from about $195 million to more than $290 million.
The cost of waiting
Debt payments on the new airport began on Jan 1, 1994, and from then on the airport system ran a deficit of $18 million to $19 million a month. GAO estimated the total cost of the delay at about $360 million if the airport opened at the end of February 1995.
What a pre-mortem would have done with these warnings
- Imagine the airport opening late in 1993 and ask why: the consultants' and airlines' warnings answer the question before anyone signs a contract.
- Rank them: an automated system the whole airport depends on, built on a schedule experts doubted, scores high on likelihood and on impact.
- Act early: Kanbanchi's page notes that the risk was accepted without a named owner, and that the fallback came last. A pre-mortem would have given the risk an owner and started the backup plan from day one.
Source: Denver airport baggage system: project failure lessons, Kanbanchi Project Failures (with the GAO sources listed on that page).
The Mars Climate Orbiter was lost to a units mismatch between two teams that nobody caught over a nine-month cruise. All facts below come from Kanbanchi's Project Failures page on the Mars Climate Orbiter, based on NASA's investigation board.
Project Failures case study 2
Mars Climate Orbiter, 1998 to 1999
Real story · NASA · Specs lost in hand-off
A hand-off between two teams
NASA designed the Mars Climate Orbiter to be the first interplanetary weather satellite. It launched on Dec 11, 1998. The work was split across organizations: Lockheed Martin Astronautics designed and developed the spacecraft, and NASA's Jet Propulsion Laboratory (JPL) kept overall project management and navigation design. Software called SM_FORCES wrote the results of small thruster firings to a file that the trajectory modelers used, and the written interface specification for that file said metric: newton-seconds.
The units mismatch nobody caught
The data came in pound-force seconds instead of the specified newton-seconds, so the navigation software underestimated the effect of each firing by a factor of 4.45. The firings happened 10 to 14 times more often than the navigation team expected, and each one added a little error. Kanbanchi's page puts the problem in project terms: the specification was right, but the question "does this file actually match the spec?" belonged to nobody in particular.
Warnings that stayed informal
Throughout the spring and summer of 1999, there were concerns at the working level about discrepancies between navigation solutions, but they were reported only informally. When conflicts in the data were found, the team relied on e-mail to solve problems rather than the formal anomaly process. The operations navigation team was small, oversubscribed and had no independent peer review of its work.
The result
On Sep 23, 1999, the spacecraft began its Mars orbit insertion burn and its signal never came back. Its trajectory was approximately 170 kilometers lower than planned. Board chairman Arthur Stephenson said the root cause was "the failed translation of English units into metric units in a segment of ground-based, navigation-related mission software."
What a pre-mortem would have done
- Imagine the orbiter lost on arrival and ask why: "a data file passed between two teams did not match its spec" is exactly the kind of specific reason a pre-mortem brings out.
- Rank it: likelihood may look low, but the impact is the whole mission, so it belongs near the top of the risk matrix.
- Act early: Kanbanchi's page suggests putting "Units match the interface spec" on the hand-off as a checklist item with a named person, and the warning sign, navigation discrepancies, needed a formal report with an owner, not an e-mail thread.
Source: Mars Climate Orbiter failure: a PM case study, Kanbanchi Project Failures (with the NASA sources listed on that page).
Exercise: sort risks from issues, then run a pre-mortem
Two steps. First a quick sort, then a pre-mortem worksheet that starts with an example and gives feedback on every reason.
Step 1: risk or issue?
0 of 6 right
A pre-mortem looks for risks, things that might happen. Sort each item. You get feedback on every answer.
Worked example
"The supplier could go out of business before delivery" is a Risk: it might happen. "The supplier has gone out of business" is an Issue: it already has. Now sort the 6 items below.
1. The vendor might deliver the hardware late.
Answer: Risk. It has not happened yet, so it is a risk: you can still prevent it or plan for it.
2. The lead designer resigned yesterday.
Answer: Issue. It has already happened, so it is an issue that needs action now.
3. Users may not like the new layout.
Answer: Risk. An uncertain event that could affect the goal is a risk. Test the layout early to reduce it.
4. The test server is down and nobody can deploy.
Answer: Issue. It is happening now, so it is an issue, not a risk.
5. A key stakeholder could be away during the approval week.
Answer: Risk. It might happen, so it is a risk. Agree a stand-in approver now.
6. The budget for the second phase has already been cut by 20%.
Answer: Issue. It has happened, so it is an issue: re-plan the scope or ask for a change decision.
Step 2: run a pre-mortem on a sample project
The project: launch a new customer portal by March 1. The prompt: it is now April, the launch has failed. Why? Four reasons are filled in as an example. Edit them, add your own (up to 8), score each one, then rank the list to get feedback.
Worked example
Reason 1 is ready to use: it is specific ("the data migration took twice as long as planned"), it scores 3 x 3 = 9, it has a concrete prevention step ("run a trial migration in week 2") and one owner. Reason 4, "tech problems", is too vague to act on: what exactly breaks, and why? Rank the list to see feedback on all four.
Knowledge check: Lesson 8
Five questions on this lesson. Get at least 4 right to complete Lesson 8 and move your course progress forward. Each answer comes with an explanation.
Key takeaways
- A pre-mortem imagines the project has already failed and asks why, before the work starts.
- Write reasons alone first, then share, so every worry reaches the list.
- Rank by likelihood times impact and act on the top risks first.
- Every top risk needs a prevention step, one named owner and a warning sign.
Try it on a real board: a risk list in Kanbanchi
Keep the results of your pre-mortem where the work happens. In Kanbanchi, the task and project management app that works inside Google Workspace:
- Add a "Risks" list to your project board.
- Create a card for each top reason from your pre-mortem, with the prevention step in the description.
- Assign each card to its owner and set the priority to match its score.
- Give the prevention step a due date, so it shows on the Gantt chart before the risky work starts.
- Review the Risks list at every status meeting and archive cards when the risk has passed.
This step is optional. The course works without an account, and your progress does not depend on it.
Pre-mortem: FAQ
What is the difference between a pre-mortem and a post-mortem?
A post-mortem, or retrospective, happens after a project and asks why it went wrong. A pre-mortem happens before the project starts and imagines the failure in advance, so the team can prevent it.
How long does a pre-mortem take?
About an hour for most projects: a few minutes to set the scene, a few minutes for everyone to write reasons alone, then sharing, ranking and agreeing actions for the top risks.
Who should attend a pre-mortem?
The project team, the sponsor and anyone with a stake in the result. Include the people who will do the work: they usually know where the plan is weakest.
What do you do with the results of a pre-mortem?
Rank the reasons by likelihood and impact, then give each top risk a prevention step, a named owner and a warning sign. Add them to the plan and the risk list, and review them as the project runs.