What does it take to write an equation for love?
According to Mashable[*1] and Match.com, it takes 16 years of data, a crack team of Ph.D.s to interpret it, and a handful of brilliant engineers to create the equation itself. Several years ago, when Match decided to overhaul the algorithm that guides its matchmaking process, they brought on current VP of strategy analytics Amarnath Thombre to revolutionize their system.
And revolutionize he did: the number of "yes" matches on the site has doubled since Thombre's changes took effect. Here's what you need to know about how the Match.com algorithm works:
- The first thing the algorithm evaluates is what you say about yourself. This information - like height, weight, location, religious affiliation, and what you're looking for in a partner - is all gathered from your profile. Match's suggestions, at their most basic level, are made by pairing people whose self-descriptions match.
- After examining how you describe yourself and what you're looking for, Match's algorithm compares that data to the information it receives from studying your actions on the site. Knowing that there is often a huge discrepancy between what we say are interested in and who we actually make an effort to pursue, the Match team developed a highly-intelligent algorithm that is capable of aggregating and interpreting all the information it receives about a member, whether it is explicitly stated or not. If, for example, you say you are only interested in meeting members who belong to the same political party as you, but you frequently send messages to members belonging to the opposite party, Match will know that you are willing to date outside of your stated preferences and will suggest potential matches for you based on that knowledge.
- The next component of the Match algorithm is an evaluation of what other members who are like you do. The algorithm takes a look at the actions of members whose behavior mirrors your own, like a tendency to communicate with the same people, and uses their experiences to learn more about preferences you likely have, but did not explicitly state.
- Finally, the Match algorithm makes use of the 16 years of data that have been collected over the site's lifetime. Using the correlations that have been found in that data, like "a woman who smokes daily is less likely to email a man who doesn't smoke," the algorithm predicts who you will be interested in before you have even rated your first match.
Match's new approach to online matchmaking has so far been a success, and the algorithm will only get more detailed and accurate as time goes on. In Thombre's words: "Instead of trying to create the perfect algorithm, we try to create the perfect algorithm for you."
For more information on this popular dating service, please read our review of Match.com.