How Match.com Makes A Match (Part II)

As an employee at i2, a ­supply-chain management company based in Dallas, Amarnath Thombre was tasked with finding the most efficient ways to transport products around the country. Now, as a key engineer for Match.com, Thombre is revolutionizing the online dating industry.

Thombre, a native of Pune, India, attended the Indian Institute of Technology in Bombay before taking an advanced degree in chemical engineering at the University of Arizona. Though his experience hadn't specifically prepared him to enter the world of online dating, Mandy Ginsberg, his former co-worker at i2 and current president of Match.com US, saw potential in Thombre when she needed to overhaul the algorithm behind Match's pairing process.

"I brought over a bunch of people who I thought could help solve one of the most difficult problems out there, which is how to model human attraction," Ginsberg told David Gelles for FT Magazine. Her choice proved to be a wise one, and Thombre's expertise, along with a team of 9 math whizzes, proved to be exactly what the company needed to update the Match algorithm.

The Match algorithm, also known as "Synapse," learns about its users in a way that is similar to the way the human brain learns, Thombre says. "When you give it stimuli, it forms neural pathways," he told Gelles. "If you stop liking something, those shut off. It's learning as you go." This innovative idea is similar to the groundbreaking technology used by sites like Amazon, Neflix, and Pandora to recommend new products, movies, or songs based on a user's preferences. But when it comes to matching people based on their potential love and attraction, however, things get significantly more complex.

"With Netflix, people are constantly rating movies," Thombre explains. "But there's only one The Godfather, and you rate it once. Even if you like The Godfather, The Godfather doesn't have to like you back. The whole problem of mutual matching makes the problem 10 times more complicated."

To solve this problem, the Match algorithm gives variables different weights, according to how users behave. "For example," Gelles writes, "if conservative users were actually looking at profiles of liberals, the algorithm would learn from that and recommend more liberal users to them."

So far, Ginsberg and Thombre's changes seem to be working wonders. Since the implementation of the updated algorithm, "Yes" ratings on the site's Daily5 feature have increased over 100%, and more than half of e-mails sent on Match originate from matches recommended by Synapse. The future looks bright for the Match algorithm.

Read the original FT Magazine article here. For part 1 on this story you can read How Match.com Makes A Match.

For more information on this online dating site you can check out our Match.com review.