How many candidates are enough?
One more interview might uncover the best person. It also costs time, and at some point you have to decide. A mathematical model called optimal stopping offers a way to think about that trade-off.
If the pool is fixed and candidates arrive in random order, the model says to spend roughly the first 37% learning what good looks like. Then choose the first candidate who beats everyone you have seen so far. It improves your chances of picking the best; it does not guarantee it.
- Set the criteria before interviews start.
- Learn enough to recognise a strong candidate.
- Decide when to stop searching. Use the rule as a guide, not an interview quota.