Hiring decisions improve when they're structured rather than intuitive

Bock argues that unstructured interviews, where interviewers ask whatever questions occur to them and rate candidates on gut feeling, are a remarkably weak predictor of actual job performance, despite being how most companies still hire. He advocates instead for structured interviews using consistent, predetermined questions scored against explicit rubrics, which research he cites shows correlates more reliably with how someone will actually perform once hired.

Google built extensive internal tracking correlating interview scores and interviewer identities with eventual job performance, discovering that certain interview formats and even certain interviewers were simply better predictors than others, information that let them refine the process based on outcomes rather than assumption.

He acknowledges that structured interviewing feels less personally satisfying to interviewers who enjoy relying on instinct, but insists the discomfort is worth it given the measurable gains in prediction accuracy and the reduction in interviewer-specific bias.

Takeaway: the interview process that feels most natural to conduct is often the one worst at actually predicting who will succeed in the job.