Pre-mortem analysis: spot project risks before launch

Pre-mortem analysis is a short team meeting where everyone assumes the project has already failed and explains why. You run it before launch, while there is still time to fix things. It takes about 30 minutes and needs only a whiteboard and a calm room.

The method was described by psychologist Gary Klein in the Harvard Business Review in September 2007. It is simple, and that is why managers keep using it.

Sources: Gary Klein, "Performing a Project Premortem," Harvard Business Review, September 2007.

What a pre-mortem is, and why it works

Most project reviews ask what might go wrong. That question feels hypothetical, so people give polite, safe answers. A pre-mortem changes the question. You tell the team the project failed. Now they only have to explain what happened.

This shift matters because of an older idea called prospective hindsight. When people imagine an event has already happened, they find it easier to come up with reasons for it. Researchers Mitchell, Russo and Pennington wrote about this in 1989.

A pre-mortem also helps quiet people speak up. In many teams, someone has a doubt and keeps it to themselves because the plan is popular or the boss is keen. Klein argued that giving doubters a safe, expected job to do lets those worries reach the table. In a pre-mortem, naming a risk is the task, so nobody is being negative.

Sources: Gary Klein, "Performing a Project Premortem," Harvard Business Review, September 2007; Wikipedia, "Pre-mortem" (summary of prospective hindsight and the 1989 research by Mitchell, Russo and Pennington).

Before the meeting: set it up

Pick a project that has a plan and a launch date but has not launched yet. Invite the people who will do the work, not only the managers. Include someone from each function that touches the launch, such as sales, operations and customer support.

Brief everyone on the plan first. Spend five minutes on it. The team must understand the goal, the timeline and who owns what. Without that, the failure stories will be vague.

Choose a facilitator who is not the project owner if you can. The owner may find it hard to hear criticism of their own plan. Ask the owner to listen, take notes and not defend the plan during the session.

Sources: Gary Klein, "Performing a Project Premortem," Harvard Business Review, September 2007. The roles and timings in this guide are the editors' practical suggestions.

A 30 minute agenda

Here is one way to split the time. It is a suggestion, not a fixed rule. Adjust it to your team.

Minutes 0 to 5: Review the plan. State the launch goal and date.
Minutes 5 to 7: Say the key line: "Imagine it is six months after launch. The project has failed badly. Why?"
Minutes 7 to 15: Everyone writes down reasons for the failure, alone and in silence. Aim for five or more each.
Minutes 15 to 25: Go around the room. Each person reads out one reason at a time until all are shared. Group similar ones on the board.
Minutes 25 to 30: Vote on the top three risks and name an owner for each.

Writing alone first is important. It stops the loudest voice from setting the direction. It also gives every person a chance to add something new.

Sources: Gary Klein, "Performing a Project Premortem," Harvard Business Review, September 2007 (the core steps: brief the team, imagine failure, generate reasons). The agenda and timings are the editors' own.

Prompts a manager can use

Some teams freeze when asked to imagine failure. Prompts help them get started. Try a few of these.

"The launch date passed and we were not ready. What did we miss?"
"Customers tried it once and never came back. Why?"
"We ran out of budget halfway. Where did the money go?"
"A key person left in the last month. What stopped us?"
"A supplier or partner let us down. What did we rely on without checking?"
"Something we all assumed turned out to be wrong. What was it?"

Keep the tone light and curious. If someone offers a worry, thank them and write it down. Do not argue about whether it is likely. That comes later.

Sources: prompts are the editors' own, built on the premortem approach in Gary Klein, "Performing a Project Premortem," Harvard Business Review, September 2007.

After the meeting: turn risks into action

A list of fears is not a plan. Go through the top risks and ask two questions for each. How likely is it? How bad would it be? Then pick a response.

Prevent it: change the plan so the risk cannot happen.
Reduce it: make it less likely or less costly.
Watch it: choose a warning sign and check it every week.
Accept it: decide it is small enough to live with, and write that down.

Give each action an owner and a date. Add the changes to the project plan. Then share a short summary with the whole team so people see their input was used.

Sources: Wikipedia, "Pre-mortem" (weighing threats by impact and probability, then taking preventive steps). The four response types are a common risk practice, presented here in the editors' own words.

Common mistakes to avoid

First, do not hold it too late. A pre-mortem after the launch is just a post-mortem. Run it while changes are still cheap.

Second, do not let the project owner or most senior person speak first. Their view can shape everyone else's. Let people write before anyone talks.

Third, do not stop at a list. If no risk gets an owner, nothing will change.

Fourth, remember the limits. A pre-mortem starts from an imagined failure, so it can raise worries that never turn out to be real. Treat the list as a set of questions to check, not as facts. Used well, it is a cheap way to find weak spots early, and a good habit for any team before a big launch.

Sources: Wikipedia, "Pre-mortem" (caveat that the method can flag threats that are not real). Other points are the editors' own practical advice.

Sources

Gary Klein, "Performing a Project Premortem," Harvard Business Review, September 2007. https://hbr.org/2007/09/performing-a-project-premortem
Gary Klein, premortem method overview. https://www.gary-klein.com/premortem
Wikipedia, "Pre-mortem." https://en.wikipedia.org/wiki/Pre-mortem