How can you get a sense for how productive an engineer will be in an interview?

I doubt it. You can ask them what projects they worked on before, and how they planned the projects and whether the deliverables came in on time, and if not, why not, and what they did about it. You can get a sense of the honesty of the person when they tell you these things. If it’s all positive, then that’s a strike against them. No one is perfect all the time. You get a sense for who is sincere and passionate, and who is just phoning it in.

But the only real way to test out someone is to have them work on real work. Some interview situations set up a test where the person has to face a problem and describe how they would work on it. But really, there is no instant judging. There’s no magic bullet. People are complex. They have good days and bad ones. They may be inarticulate or downcast in an interview and yet be really good workers.

Ask them to tell you stories—with details, about past projects. If the details feel real, then I think you can begin to trust their perceptions of themselves. On your side, be genuinely curious and interested in what they have worked on. You can draw people out more this way. Many interviewers are amateur, and ask dumb questions like “tell me about your strengths and weaknesses.” If someone asked me that in an interview, I’d get up and walk out right then, because that would tell me that they have no idea what they are doing. I don’t want to work for someone like that.

Anyway, the more you get away from what you think an interview should be, and you get towards talking as you would to someone from work or life that you are really interested in, the better the information you get will be. Work is real. Interviews should be, too.

No. All the studies show that interviews are awful predictors of who will work well in a company and how productive they will be. The best predictor is, of course, a work trial (internships often serve this function).

Really the only way is to give them a task and have them perform it. A take-home task.

That, and a trial period.

I think you can get to know somebody by asking questions which have a wide variety of answers and then comparing the answers across a pool of applicants. Microsoft does this. A friend of mine reported that one of the questions was “How would you design an oven so that it was completely safe?”

He asked for clarification on completely safe, first of all. Then they said “An oven that a small child or an elderly person living alone could use without posing a danger to themselves or to their home.”

The genius behind this question is that you can take an administrative approach, an engineering approach, a construction approach, a policy approach, and even a legal approach. It’s completely open-ended. But by seeing how the people choose to think about the problem you can infer whether they are suitable for the job you are hiring them for.

When I interviewed people to work for my house cleaning startup, I put a ¾” empty square on the application and asked them to fill it in. All they had was a blue pen. The more completely they filled in the box, the more attention I thought they paid to detail. some people just scratched across the box very lightly.

As a final example, I was on a hiring committee for an IT position. We broke a computer (in multiple ways) and had the interviewees fix the problems in front of us. This told us not only if they knew something simple about computers but also revealed their problem-solving style in a pressure situation. This was important because we would ask the technicians to fix computers in front of classes and faculty during classes if they went down suddenly. They needed to be able to think clearly while we were talking to them.

So for engineering it gets tougher. But I think a little forethought can bring some interesting details to light. Maybe a guideline is: Don’t expect the question/test to tell you who will be really good, but do design it to tell you who will be really bad.

But there is a lot of self-delusion going on when you try to use tricks like these. Still…even if they gain you no real information, they are fun to do. :)

And never ask anybody what their plan is in 5 years. In my experience, nobody does what they said they were going to do…so don’t cloud your decision-making process with information you already know is probably falsified.