How Tinder is using data to make your right swipes count

To combat overzealous swipers, Tinder has turned to a new tool for analyzing data.

Mobile dating app Tinder is not always thought of as a wholesome place to find a date, but even it has some unwritten rules. And when users break them, Tinder takes action—sometimes by analyzing data.

For example, some users swipe right (that’s Tinder-speak for “liking” someone) too often in order to maximize their chances of finding a match. The carpet bomb approach might seem like a smart tactic if you’re desperate, but Tinder doesn’t like it. It “decreases the value of the swipe,” explained Tinder Vice President of Technology Dan Gould in a recent interview with Fortune. And as goes the value of the swipe so goes Tinder.

Earlier this year, to combat this collection of overzealous swipers, the company set a limit on the number of right swipes users could make daily. Then it rounded up profiles of the most prolific right-swipers and watched what happened. Would they get upset and leave? Would they stop using for a while and lick their wounds? Or would they realize the error in their ways and be a little more discriminating in suggesting possible hookups?

“We saw those users actually converge [with the rest of Tinder users] in behavior,” Gould said. “They started using Tinder as expected. They learned the value of the right swipe.”

That Tinder is using data to improve the user experience should not be surprising. Data analysis informs everything from product design to marketing campaigns at most successful digital companies, but the technology Tinder is using might come as a surprise. Especially considering Gould, who joined the company in January, describes himself as being much more inclined to build software rather than buy it. Tinder is using software from a startup called Interana, which premiered in 2014 and counts two former Facebook FB 1.00% engineers among its founders.

The Facebook connection was actually crucial in this case. While they were at Facebook, Interana co-founders Lior Abraham and Bobby Johnson created a tool called Scuba, which was designed to let Facebook employees run really fast queries over all sorts of data. Gould had learned about (and was impressed with) Scuba from some friends who worked there, and when he heard that Interana was in many ways a commercialized version of Scuba, he decided to give it a try.

Anything would have been better than the status quo at Tinder, which had been using legacy analytics software that was completely overwhelmed by the company’s meteoric growth. But Interana’s technology proved particularly adept. Queries that were taking hours were cut down to less than a second, Gould said, and that changed the way employees think about data. Now, Interana is everywhere inside Tinder, troubleshooting network connectivity issues and measuring the effectiveness of social media partnerships, among other things.

But the Interana story is much bigger than Tinder, and it points to a future where a growing amount of enterprise software has its roots in large web companies. They often build tools designed to handle data or user traffic in volumes much greater than what mainstream users might encounter. So when those technologies are commercialized and implemented by other companies, they’re more than up for the job. Often, these are open source technologies, but even paid enterprise versions or strictly proprietary software can attract customers if they’re worth the money.

For example, I first learned that Microsoft was using Interana to analyze behavioral data on Bing not from the Interana press release, but by overhearing a Bing engineer singing its praises at my airport gate. That’s saying something.

In fact, Interana co-founder and CEO Ann Johnson said the company is winning over quite a few folks like Tinder’s Gould and the Bing team at Microsoft, who, like Facebook, have the inclination and the engineers to just build what they need rather than pay for software licenses. In the tech industry, especially if you’re selling to engineers, that can be a difficult thing to do.

“A lot of [our users],” Johnson said, “have never really bought a piece of software.”