Imagine it already failed.
Pre-Mortem
Project six months forward where the project has failed. Ask: what went wrong? Surface hidden risks before they surface you.
A post-mortem happens after a project fails. You analyse what went wrong and commit to doing better next time. A pre-mortem moves that exercise to before the project starts — when you can actually do something about it.
The method
Gather the team. Say: 'Imagine it is six months from now. The project has failed — completely, catastrophically. Not partially failed — it is dead. Now write down, individually, every reason you think it failed.' Collect the answers. Prioritise by likelihood and severity. Address the top items before you start.
Why individual writing matters
The pre-mortem must start with silent individual writing, not group discussion. In a group, the loudest person anchors the conversation. Status dynamics suppress dissent. The engineer who knows the biggest technical risk might not say it in front of the exec. Silent writing surfaces the distributed knowledge in the room — including the uncomfortable parts.
What it surfaces
Typically: unclear ownership at critical moments, dependencies you assumed were solved but weren't, underestimated complexity in a specific area, interpersonal tensions that slow decisions, and external factors nobody wanted to name out loud. These are rarely surprises — they're known unknowns people were politely ignoring.
In engineering leadership
Run a pre-mortem before any major infrastructure change, product launch, or organisational restructure. The most common failure mode in senior eng: 'we thought we'd have time to address X after launch' — the pre-mortem surfaces X before the decision is locked in.
In practice
Before launching a zero-downtime database migration: 'We migrated, it is three weeks later, and we have a P1. What happened?' You'll surface: the rollback procedure was untested, two engineers who understood the migration left the team, a latent data integrity issue wasn't caught by staging tests. Fix these. Then migrate.
TL;DR
Assume failure, ask why. Silent individual writing. Surface what optimism suppresses.