As of 2022, Tinder has 75 million active users, had a total of 30 billion matches and makes possible 1.5 millions dates per week. It’s, hands down, the most popular dating app.

Here’s how Tinder works:

Tinder can't show you all the registered profiles (over 75 million), so it's in their best interest to show you in priority the profiles that might interest you (and be interested in you) first. The more profiles you correspond to, the more matches you have, the more you use the application.

The algorithm they use to rank profiles is described in their patent US20190179516A1. This article explains how it works and which factors are used. References made to the patent will be accompanied by the number of the relevant section, in square brackets.

Note that the patent does not describe how the algorithm works, but how it “could work", so this article is largely speculative. Some features described in it are not even in the app (like a possibility for users to rate other users’ attractiveness from 1 to 9 [69]).

However, the patent gives a fairly clear idea of the current operation and the consequent number of factors used.

What is the algorithm doing?

Tinder calculates a “match potential” score between each user based on a variety of factor. People with the highest score will appear first in the profiles presented to the user.

This score is hard to compute, indeed:

There is statistical and empirical evidence to suggest that successful on-line dating entails far more than simply a matter of matching supply and demand. [4]

Previously, Tinder used an Elo Score, like in chess games. Most liked users had the highest score and were proposed profiles with a similar score. They stopped for ethical reasons, and because attractiveness is more complex than a game of chess; the same profile can be seen as very attractive by one person and as less attractive by another.

They declared now using mostly usage (such as who you liked, your time spent on the application, how quickly you respond to your matches, etc.) and preferences (location, age and gender preference). They certainly use these factors to a large extent, but that’s not all, as we’re going to see.

How does it works?

Tinder gives a high “match potential” to similar people:

People having similar and/or compatible character traits and values should be matched together. [5]

To create the list of profiles proposed to a user, Tinder:

  1. Filters profiles based on explicit preferences (a male profile should not be proposed to a straight male user). Blocked profiles or profiles that have already expressed their disinterests into users should not be presented as well. [86][87]
  2. Ranks remaining users, based on a score built from specific factors of profiles and current user. [95]

The factors used by the algorithm tell us a lot about how it works, and what it considers important. Let's dig into this topic.