Why job interviews get it wrong

I've worked in recruitment for eleven years, and I've interviewed hundreds of people. For most of that time I was sure I could spot a good candidate within five minutes. I was wrong more often than I'd like to admit. Hiring the wrong person is expensive. You lose months finding someone new, and one bad hire can drag down a whole team.

The problem with a normal interview is that it rewards the wrong things. A relaxed chat favours people who are confident, friendly and good at talking about themselves. Those are nice qualities, but they're not the same as being good at the job. I also used to make up my mind in the first few minutes. After that, I spent the rest of the meeting looking for proof that I was right. One candidate arrived late because of a flooded road, and I never quite got over it, however good her answers were.

Research on hiring backs this up. Unstructured interviews are a weak guide to how well someone will actually do the job. Bias creeps in easily, too. Interviewers tend to prefer people who remind them of themselves, people who went to the same university or share their hobbies. Do that for a few years and you end up with a team that looks and thinks alike.

So we changed how we hire. Every candidate now answers the same questions in the same order, and a small panel scores each answer against criteria we agree in advance. We also give people a short task based on the real job, such as replying to an angry customer's email or planning a simple budget. It's less fun for the interviewer. The results, though, have been far more reliable.

Personality still matters. A team needs people who get along with others and stay calm under pressure. You can test that fairly as well, by asking candidates to describe a real situation they handled in the past. If you're job hunting, prepare three or four real examples, not just a confident smile. And if you're the one hiring, don't trust your gut as much as I used to.