Look-Then-Leap Rule

The look-then-leap rule is an optimal stopping strategy: spend a defined phase gathering information (look), then commit when a threshold is reached (leap). It comes from the secretary problem and related optimal search literature. The idea is to resist committing too early — before you've seen enough options or signals — and to avoid searching forever. There is a right moment to stop looking and act.

In the classic secretary problem, you see n candidates one by one, must accept or reject each immediately, and want to maximise the chance of picking the best. The optimal policy is to reject the first n/e candidates (the "look" phase), then choose the first candidate who is better than all previous ones (the "leap"). That policy gives a probability of 1/e ≈ 37% of selecting the best — far better than random, and provably optimal under the problem's constraints. The rule generalises: when you face a sequence of options and must decide in real time with no recall, reserve a fraction of the "budget" for looking, then leap when you see something that clears a threshold.

In practice, look-then-leap appears in hiring (interview many, then hire when you see someone above a calibrated bar), in dating and partnership, in venture (see many deals before writing cheques), and in product (explore before doubling down). The mistake is leaping too soon (first acceptable option) or looking too long (never committing). The discipline is setting the look phase and the leap threshold in advance, then following them.

The rule also clarifies when not to use it. If you can batch options and compare them (e.g. receive 50 CVs and rank them), you're not in a sequential, no-recall setting — use comparison, not look-then-leap. If the cost of rejecting a candidate is low (you can re-open the search), the optimal policy may be more conservative (look longer). The model is most valuable when options arrive over time, you must accept or reject each, and you cannot go back.