There is no good evidence that interviews reliably predict how someone will perform in a job. That is uncomfortable for anyone who prides themselves on being a good judge of people, myself included.
The problem is simple. In an interview you can only see a candidate’s observable behaviours. What makes the difference over the long run are their innate behaviours, and those only come out with time and skill. Then we add our own biases on top.
Confirmation bias
We make a snap judgement and then spend the rest of the meeting, consciously or not, trying to justify it. The questions drift towards whatever will support the first impression. We want our instincts to be right.
Around 60% of interviewers decide whether a candidate is suitable within 15 minutes of meeting them. The danger is obvious: you could be passing over great hires for no real reason at all.
The halo effect
One factor, such as a shared hobby, a sport or a mutual friend, ends up carrying far too much weight. We like people who are like us. You can hear it happening in leading, positive questions: “And you probably achieved that on your own?”
This is one reason I would always have more than one person in the interview, ideally peers who can be honest with each other afterwards.
Predictable questions, rehearsed answers
Standard interview questions get standard, rehearsed answers. Candidates who seem to lack authenticity are often just nervous, performing in a situation that will never be repeated in the actual job. None of it tells you much about the behaviours you will be living with.
Too little data, too late
This is the biggest weakness of most hiring decisions. References, work and personal, are usually taken at the end, after they can still change anything. Cognitive tests, such as reasoning and working memory, are better predictors of job success than interviews, yet they are rarely used.
We are only human
None of this means you shouldn’t like the person you hire. Ideally you should; you will be working closely with them. The point is to stop liking them from being the whole decision. There are practical ways to reduce these biases and still hire people you want to work with, and they sit at the heart of how I run a search and how I assess fit.