The interview with no questions

A manager hiring for a practical job stopped asking questions. Instead, candidates spent ninety minutes doing a small version of the actual work, with the same tools and the same kind of problem.

The results were uncomfortable for everyone who had relied on interviews. The ranking produced by the work sample bore almost no relation to the ranking produced by the conversation. Confident, articulate candidates did not reliably do the work well, and several quiet candidates who had interviewed poorly were among the strongest.

This matches a long line of research. Across many studies, unstructured interviews — a conversation, no fixed questions, judgement formed as you go — predict later performance barely better than chance. Work samples and structured interviews, where every candidate answers the same questions scored against the same scale, do substantially better.

The reason the conversation persists is not ignorance. It is that interviewing feels extremely informative. A manager leaves a good conversation with a strong impression and complete confidence, and the confidence is unconnected to the accuracy. Nothing in the experience signals that the judgement is weak.

For a candidate the practical implication is narrow. Ask early whether there is a practical element, and if there is not, find a way to show the work anyway — bring something you made, or describe one task in enough concrete detail that it functions as a small sample.