The 37% Rule: Why the Optimal Strategy for Finding the Best Option Requires You to Reject the First Third of All Candidates

A hiring manager staring at a stack of 100 resumes, a homebuyer touring open houses in a tight market, or a renter looking for an apartment all face the exact same mechanical constraint: they must evaluate candidates one at a time, they can only either accept or reject the option in front of them, and once they pass on a candidate, that option is permanently gone. The decision-maker cannot look ahead to see who is next, nor can they rewind to claim a choice they already discarded.[4]

This is not merely a psychological dilemma; it is a rigid mathematical structure known as the "Secretary Problem" or the game of optimal stopping. The objective is uncompromising: maximize the probability of selecting the single best candidate in the entire pool. Second best counts as a failure.[1][2]

The mathematical consensus on how to solve this is absolute, and it requires a strategy that feels deeply counterintuitive to human risk aversion. According to the foundational theorem of optimal stopping, the decision-maker must unconditionally reject the first 37 percent of the applicant pool, using them solely to establish a baseline standard of quality.[1][2]

After that initial 37 percent is discarded, the rule dictates that the searcher must immediately hire the very next candidate who is better than everyone seen so far. This strategy—derived from the mathematical constant e (approximately 2.718)—yields a 37 percent probability of securing the absolute best option in the entire set.[1]

"The beauty of the 37 percent rule is that it provides a mathematically rigorous answer to a fundamentally human problem," notes the American Mathematical Society in a 2018 analysis of sequential decision-making. "It tells you exactly when to stop gathering information and start taking action."

The evidence for this threshold is rooted in probability theory dating back to 1960, when Martin Gardner first popularized the problem in Scientific American. As outlined in the UCLA Department of Mathematics' 2007 text Optimal Stopping and Applications, the solution relies on the fact that as the number of candidates n grows large, the optimal stopping point converges to n/e, which equals approximately 0.368, or 37 percent.[1][2]

The data shows that deviating from this threshold severely penalizes the decision-maker. If a hiring manager stops too early—say, after evaluating only 10 percent of the pool—they lack the necessary information to know what a "good" candidate actually looks like, risking a premature commitment to mediocrity.[2][4]

Conversely, if they wait too long and evaluate 60 percent of the pool before making a move, the math shows they will likely pass right by the best candidate, who is statistically probable to appear earlier in the sequence. The 37 percent mark represents the exact mathematical peak of the probability curve.[1][4]

However, the application of this rule in real-world scenarios carries significant limitations. A 2017 paper published on arXiv, On a Class of Optimal Stopping Problems with Applications to Real Option Theory, highlights that the classic Secretary Problem assumes the decision-maker has zero prior knowledge about the distribution of candidate quality.[3]

"In most practical applications, such as real estate or financial options, the searcher actually possesses some historical data about the market," the authors write. When prior information is introduced, the strict 37 percent threshold shifts, often allowing the searcher to stop earlier if an exceptionally high-value option appears.[3]

Furthermore, the rule assumes that a rejected candidate can never be recalled. In a 2014 analysis, the Harvard Business Review pointed out that modern hiring does not always follow this strict sequential elimination. "If you can keep a candidate warm while interviewing others, the mathematical penalty for continuing your search drops significantly," the publication noted.

AI

Yet, even with these caveats, the 37 percent rule remains the most robust baseline for decision-making under uncertainty. It forces a structural discipline onto processes that are typically derailed by emotional fatigue or the fear of missing out.[4]

The rule also scales infinitely. Whether a person is evaluating 10 apartments or 1,000, the optimal strategy remains exactly the same: spend the first 37 percent of the search purely gathering data, and then strike on the next record-breaker.[1][2]

The math does not guarantee success—a 37 percent chance of finding the absolute best candidate still means a 63 percent chance of failure. But in a universe of unknown variables, it provides the highest possible floor for making a choice. The next time a decision-maker faces a sequence of options, the most rational move is to let the first third pass them by.[4]