The 37% Rule for Choosing a Partner

Imagine you’re an interviewer who has to hire a single outstanding candidate out of 100 applicants for a secretary position.

The 100 applicants line up in random order, and you interview them one at a time. Once you reject an applicant, you can never see them again. Conversely, the moment you hire someone, you lose the chance to see anyone still waiting in line behind them. Under these rules, what strategy gives you the best shot at hiring the very best applicant?

This is the famous optimal stopping problem widely known as the “secretary problem.”

It already has a known solution. The probabilistically optimal approach is to interview and reject the first 37% or so of applicants, take the best of those 37% as your benchmark, and then hire the very next applicant who scores even one point higher than that benchmark.

I’ll leave the proof to the clever folks and think about this from a different angle. There’s something about this problem that resembles dating. You’re (usually) with one partner at a time. When you break up, you (usually) never see them again. You meet the next person in sequence, and when you find the optimal match, you stop dating and end things with marriage (usually).

So does the 37% strategy work for love too?

For this strategy to work in real life, you’d need to know two things. First, you’d have to accurately predict your own n, that is, the number of romantic partners you’ll meet over a lifetime. Second, you’d have to accurately score the people you meet.

In the secretary problem, n is clear: there are 100 applicants. But in life, n is unknowable. How many people will I meet in my lifetime? I have no way of knowing whether the person I’m seeing now is the third, the seventh, or the last.

But we do have statistics. According to the ‘2026 Report on Perceptions of Love and Happiness’, a survey of 2,000 unmarried men and women across the country conducted by the matchmaking company Duo, the average number of romantic relationships among unmarried people was 3.2. If we assume that everyone stopped at the most rational point, just as in the optimal stopping problem, then the average pool of romantic candidates that an unmarried Korean will encounter over a lifetime works out to about 8.6 people, roughly 9.

Of course, this calculation is less a serious statistic than a mathematical joke. For one thing, the survey targets unmarried people, meaning people who haven’t yet reached the stopping point of marriage. So the average of 3.2 relationships isn’t the total number of relationships over a lifetime, it’s the number of relationships they’ve had so far. To put it in the language of the secretary problem, we don’t know how far along the pool of n candidates these people have gotten. Someone might still be at the 20% mark, while someone else might be at the 80% mark. Some may be about to stop, while others may have a much longer search ahead of them.

So the average number of relationships is not the same as my n. The number 3.2 is an interesting starting point, but it doesn’t tell me how many people I’ll come to love in my lifetime. Some people meet their lifelong partner after three relationships; others have had ten and still don’t really know themselves. Some marry their first love; others only learn what real love is after a long marriage. In real dating, n is not given in advance. We have to choose with whatever heart and experience we have in the moment, without knowing how many candidates there are in total, and without knowing where this particular person falls in the sequence.

The second problem is even harder. How do you score another person?

Matchmaking companies sort people in their own way. Age, education, occupation, income, assets, looks, family background, where you live, religion, views on marriage and the like all end up on the scorecard. Such conditions are relatively easy to turn into numbers. Salary is a number, height is a number, age is a number. Education and occupation can be ranked, too. So we fall into the illusion that even love can, to some degree, be reduced to a table.

But as everyone knows, the things that truly matter are not easy to convert into numbers.

Do I become a better person when I’m with this person? When we’re angry, can we avoid tearing each other apart? Do we keep from letting unspoken feelings fester for too long? When our paces differ, can we wait for each other? Does our time together drain me or restore me? When life gets hard, can we end up on the same side?

These are the things that are hard to put a score on.

Kindness, 87 points. Humor, 73. Conflict resolution, 91. It sounds plausible, but love mostly doesn’t work that way. A person’s kindness might look like a 50 most of the time, then become a 100 on the day I fall apart. A person’s humor may not make everyone laugh, but it keeps me alive. One person’s flaw might look small objectively yet be unbearable to me, while another person’s lack is somehow something I’d gladly hold close. The optimal partner may not be someone who shows up as a perfect 100 from the very start.

In the world of the secretary problem, the choice is simple. An applicant is an applicant, and an interviewer is an interviewer. Each applicant’s ability is fixed, and the interviewer just has to guess the ranking. But in the real world, both the one doing the evaluating and the one being evaluated keep changing. The person I feel is good for me today may not be the person who’s good for the me of five years from now. Conversely, someone who seemed ordinary at first can, over a long stretch of time, become the most important person in my life.

That’s why applying the 37% strategy to love as is would be dangerous. You can’t burn through your first 37% as mere practice, and you can’t marry someone on the spot just because they scored one point higher than everyone before. Above all, people are not applicants. Meeting someone is not the same as ticking boxes on a comparison chart.

Even so, the analogy isn’t completely useless.

What the 37% strategy teaches us is not “calculate who the best person is.” It’s closer to “don’t try to make the perfect choice from the very beginning.” We learn our standards through trial and error. Few people know exactly what kind of love they want from the start. Some breakups are not failures. Some regrets tell us what we are unable to tolerate. Some good memories leave behind a sense of what we should never give up. So there’s no need to feel too cheated by the loves that have passed. In the language of the secretary problem, we’re all passing through the stretch where we build our own standards.

The secretary problem also reminds us that at some point we have to stop comparing.

The thought that there might be someone better is always available to us. There really are a lot of people in the world, and the people I haven’t met yet will always outnumber the ones I have. But life, like the optimal stopping problem, doesn’t let us scan an infinite pool of candidates all the way to the end.

The secretary problem tells us the optimal moment to stop. But human affairs are far too complicated. Love involves timing, luck, family, money, health, and each person’s wounds. Even the best person can slip away if you meet them in a bad season, and an ordinary connection can deepen through mutual effort. Math is tidy, but life is messier than math. And it’s that very quality that makes life beautiful.

Love asks not when it pays most to stop, but with whom stopping would feel like no loss at all. Maybe good love isn’t about finding the best person, but about meeting someone who makes you feel it’s okay to stop searching endlessly. It’s choosing the person beside you right now, even while knowing full well that somewhere out there a better person might exist.