How Do Dating App Algorithms Work?

A dating app algorithm decides whose page appears in front of whom. The apps say they use how active you are, who you like and skip, how far away you are, your stated preferences and who tends to like you back. None of them publishes the full formula. What follows is what each app says in its own help pages, plus what experience adds.

Tinder says the most important factor is how actively you use the app. Its matching favors active users, and people who are active at the same time.

It also says it considers distance, interests and lifestyle details, anonymized photo cues similar to photos you have liked, and your likes and nopes.

Two things follow. A man who opens the app once a week is competing against men who open it daily, and the app has less recent information about him. And every like or nope teaches the app something about your taste.

Hinge says its proprietary matching algorithm uses age, gender, location, preferences, and who you like, skip or match with. It uses these to recommend compatible people to you, and to recommend you to others.

Its help center advises sending quality likes over quantity because the algorithm learns from your activity. That advice is worth following. Hinge is telling you directly that a thoughtful like teaches it more than a dozen careless ones.

OkCupid works differently, because it is built around questions. Its Match percentage is built from what you want, what the other person wants and how you both answered the same questions.

It also has a rule you can use. If someone likes you, you are more likely to see them sooner. If you like someone first, they are more likely to see you. A like on OkCupid is partly a way of getting in front of her.

The apps share inputs, not weights. Nobody outside the company knows how much distance counts against activity, or how much any single signal counts at all.

Be wary of anyone who claims to know the exact formula. They are guessing. a dating coach says the same: it is a proprietary system running behind the curtain, so you work from experience and logic. His logic is that every dating app is a business, and the app rewards what keeps its users happy and on the app.

a dating coach's view, drawn from years on the apps, is that the algorithm punishes certain behavior. He lists three.

Swiping right on everyone. He used to auto-like and it worked, but over time it stopped. His reasoning is that an app learning that you like every woman shown has nothing to go on, so it may show you women who get fewer likes and fewer of the ones everyone wants. He has told students who auto-liked to stop, and says they did better with fewer, more selective likes.

Getting reported or unmatched often. Harsh or polarizing behavior with women, getting unmatched a lot and getting reported can all count against you. The app wants women to stay happy, so it has a reason to show you less.

Dull pages that get no response. If women pass on you, the app learns that too.

The flip side is a gain. Better photos get more likes, and a good page that starts conversations tells the app you keep women engaged. Being good at the visible part helps you twice, once with the women directly and once with the app.

You cannot touch the formula. You can change the inputs the apps say they use.

Use the app often enough to be active, and in short sessions rather than one marathon. Like only the women you would actually meet. Fill in the details the apps ask for, since interests and answers are inputs. Send quality likes, as Hinge advises.

Answer the women who write to you, and keep conversations going, because a match that goes nowhere teaches the app nothing good about you. Keep your behavior clean. Say nothing to a woman that you would not say to her face.

Tricks built on a guess about the formula do not last, because the apps can change how they weigh things without telling you. Auto-liking is the clearest case: it worked for a dating coach once and then stopped.

The sensible approach is the dull one: strong photos, a page that says something specific, selective likes and real replies. The algorithm is trying to do what you are trying to do, which is to put you in front of women who will like you. Help it, and it works for you.