How to run a premortem

Premortems help us identify and reduce risks before they happen. They are a great exercise to run before launching a product or feature.

What is a premortem

Product teams often use postmortems to understand what caused a problem and how to reduce the risk of it becoming a problem again. Premortems are the opposite. You do them at the beginning instead of the end, and you plan how to stop the worst from happening rather than try to understand why it happened.

When to use a premortem

Premortems are a great technique to use when:

• planning a product launch or feature
• starting a new project
• forming a new team

Benefits

Premortems help reduce the risk of failure by:

• early identification of risks. This means that the team can take steps to mitigate them before they happen
• better identification of risks. Asking people to imagine all the ways that something can go wrong is actually quite fun. Giving people license to think critically generates good ideas, and gives people who might not have otherwise spoken up about concerns encouragement to do so
• building shared understanding amongst team members and stakeholders about risks, their impact, likelihood and potential responses

How to run a premortem

To run a premortem, start with a question for the team to answer. The important thing here is that the question invites participants to think negatively. A question I used recently when planning a new product launch was: “What could go wrong when we launch the new version of our new product and redirect the old domain?”

Some people need time to think before they can generate their best ideas, so share the premortem question at least 48 hours in advance.

The premortem starts with everyone taking 5-10 minutes to write down all the things that could go wrong, and give them a score for:

• impact if the thing goes wrong
• likelihood that it will go wrong

The two ways I’ve done this before are:

1. mapping everything using post-its on a 2×2 grid (impact and likelihood as the x and y axis)
2. putting everything in a table in Notion / a spreadsheet, and asking people to select the correct value for impact and likelihood

I prefer option 1 if running the event in person, and option 2 if running it remotely.

Once everybody has had a chance to brainstorm, run through all the problems – starting with the highest impact /likelihood, and try to identify some actions you can take to reduce the impact/likelihood. Aim to spend no more than 5 minutes on each one to make sure that you can at least get through the top 10 within the hour. You probably won’t have time to discuss everything, but if you prioritise the highest impact/likelihood ones, then the ones that are left will be the less important ones.

Encourage the team to focus on identifying actions they can take to bring the impact/likelihood down to an acceptable level. Try to minimise the amount of time the team spends debating superfluous things like impact/likelihood scores, as the main priority here is to identify actions the team can agree on. As you go through the list, you will probably start speeding up as duplicates are identified, or you realise that actions which have already been discussed are relevant to more than one risk.

At the end of the premortem you should have:

• a list of scored risks
• an agreed list of actions to mitigate the risks the team are most worried about

The last thing to ask the team to do is to vote for the 5 most important actions of all the ones identified. This gives everyone a quick sense of which actions are the most important to prioritise.

At the end of the meeting add the actions to the backlog, so they can be prioritised alongside other work. If you have a risk register, you might want to add any new risks you’ve identified to that also.

Templates you can use