Imagine you want to hire someone. In a meeting room, there are eight candidates waiting to be interviewed. You can only interview one of them at a time, and after each interview, you need to decide if you want to offer them the job straight away or if you want to send them away and continue interviewing. You can’t call candidates back.
Since you don’t know what the average candidate quality is like, you wouldn’t want to offer the position to the first one you talk to. Even the second one may be too risky. How many candidates should you interview to calibrate your expectations before you consider making an offer?
Astoundingly, there’s a precise answer to this question. If you have eight candidates, you should talk to three before making the switch. More generally, you should talk to 37% of your candidates.
This is also known as the Secretary Problem. The 37% rule is useful in scenarios where decisions must be made sequentially without the ability to return to previous options. My wife and used it when we were planning our strategy to buy a house. We knew that we wanted to make an offer in the next few weeks, but that we needed to calibrate our expectations. We didn’t make an offer for the first few houses we looked at because we didn’t know how they compared. Eventually, we switched to “offer” mode. Full disclosure: We ended up spending many more weeks than we had planned to find a place we were happy with, so our calibration period was probably more like 10% than 37%. Since in real-world scenarios, the strict requirements of the Secretary Problem are hardly ever met, its more fundamental lesson is to allow for a calibration period before making a decision.
Nehaveigur 