A major platform launch is three months away. How would you run a pre-mortem, and why bother?

What the interviewer is testing

Whether you know a specific technique for surfacing risks that a risk register does not.

Why it works

A risk register asks people to speculate about what might go wrong, which invites hedging and optimism. A pre-mortem asserts that it has already failed and asks why — and that framing gives people permission to voice concerns they would otherwise soften.

The research on this is consistent: prospective hindsight substantially increases the number and specificity of causes identified. It also depersonalises the concern — you are explaining a failure, not criticising a colleague's design.

How to run it

Set the scene concretely. "It is six months after launch. The platform has failed badly. The programme has been publicly criticised and we are unwinding it." Specific and vivid, not "imagine things went wrong".

Everyone writes independently first, in silence, for ten minutes. This is essential — group discussion first anchors everyone on whoever speaks earliest and loses the divergent thinking that makes the technique work.

Then share round-robin, one cause each, no discussion or defence during collection. Defence kills disclosure immediately.

Cluster and prioritise by likelihood and impact.

Assign owners and actions to the top items, with dates. Without this it is a venting session.

Include the right people: engineering, operations, support, security, and someone from the business who will use the thing. The most valuable contributions frequently come from support and operations, who see the failure modes nobody else does.

What it typically surfaces

Operational readiness gaps, unowned dependencies, unrealistic assumptions about user behaviour, missing runbooks, capacity assumptions nobody validated, and the migration or rollback path — which is routinely the least designed part of any launch.

What a strong answer adds

Running it early enough to act — three months is workable, three weeks is theatre. And revisiting the top items as a standing checkpoint rather than filing the output, since a pre-mortem whose findings are recorded and not tracked has the same fate as a risk register.

Common weak answers

A risk workshop with a different name. Group discussion without independent silent writing.