Pre-mortem

A pre-mortem is a forward-looking, risk-identification activity used by teams to imagine the failure of a project before it starts.

Goal

The goal of the pre-mortem is to enhance project planning and execution by preemptively identifying what could go wrong, thereby increasing the chances of success. It encourages open and creative thinking about potential problems, fostering a culture of transparency and preparedness.

Context

We have rose-tinted glasses for how great our products and features will be. By forcing people to imagine that the product has failed it helps to identify risks and assumptions that need to be validated.

How to run a pre-mortem

A pre-mortem session typically takes 30–60 minutes and runs in four steps:

1. Set the frame. "Imagine it's twelve months from now. We launched this solution and it failed badly. We're now in a meeting trying to figure out what went wrong."
2. Silent generation. Each team member writes down as many specific failure modes as they can think of, individually, on stickies. Five minutes minimum.
3. Share and group. Each person reads out their stickies one at a time, and the group clusters similar ones. Encourage piggybacking when someone hears a related failure mode.
4. Convert to assumptions. For each cluster, ask "what would we have to believe for this not to happen?" That belief is an assumption to validate.

The exercise works because imagining a specific failure produces concrete, named risks that can be tested. Asking "what could go wrong?" in the abstract usually gets you nothing.

Never underestimate people gaming the system

Pre-mortems consistently surface a particular class of failure: customers and partners behaving in ways the team didn't intend, because the new offer creates a new incentive. The middle-seat-purchased-by-a-partner-system case is one example. Others include discount codes used by people who would have paid full price, referral programs gamed by self-referral, free tiers cannibalising paid tiers because the conversion path was too easy to avoid.

Spend a deliberate five minutes during the pre-mortem asking "how would someone game this?" The team will catch failure modes that don't show up in the desirability or usability dimensions.

Inputs

Outputs