What a simulation actually measures

Every hiring process is a measurement problem wearing a costume. You have an hour, a stranger, and a decision worth several hundred thousand pounds. The interview is the instrument almost everybody reaches for, and it measures one thing well: how fluently a person can narrate work they have already done.

That is a real signal. It is not the signal most roles are hired on.

The gap between describing and doing

A candidate who has spent six years running operations can tell you, credibly and in detail, how they would handle a supplier failing mid-quarter. What they cannot tell you, because nobody can narrate this, is what they would actually notice first, which of the four competing problems they would leave alone, and how their judgement holds up when the second thing goes wrong forty minutes later.

A simulation puts them in that morning. Not a case study about that morning: the morning itself, with the inbox, the half-finished spreadsheet, the colleague who is confidently wrong, and no interviewer nodding along to tell them they are on the right track.

What comes out of it

The output is not a score. Scores compress exactly the thing you were trying to see. What comes out is a record of the decisions somebody made when nobody was steering them, and an assessment written against the work itself: what they prioritised, what they left, what they asked for, and what they shipped.

• Where they went first, and what that says about what they think the job is.
• What they chose not to do, which is usually more revealing than what they did.
• How they handled being wrong, which every simulation eventually arranges.

You end up comparing candidates on the same morning rather than on the quality of their anecdotes. That is the whole idea.