Say Allo is based on compatibility. Depending on questions you answer and the app's ability to learn based on how you swipe, it's able to produce increasingly better matches for users.

Dating app uses machine learning to find your most compatible partner

DENVER — A new app born in Colorado is bringing machine learning to online dating apps.

Denver-based Say Allo marries the user interface of swipe-based dating apps that users have become accustomed to, like Tinder and Bumble, with the compatibility measurements of dating sites like OK Cupid, eHarmony and Match.com.

Zackary Lewis, founder of Say Allo, created the dating app after a relationship he had ended because of incompatibility.

But Say Allo is able to measure compatibility without requiring customers to submit long questionnaires, said founder Zackary Lewis.

Rather, users answer about 10 questions about themselves, instead of what they’re looking for. The questions let the algorithm Say Allo uses to start learning about the customer. As that person starts swiping through the app more, the algorithm learns about what they like and dislike. The algorithm can learn what factors are increasingly important to users, even going so far as to detect facial features a person prefers.

The result is the more a person uses Say Allo, the more the app shows matches that coincide with preferences. That’s different from apps like Tinder or Bumble, which shows the most popular, swiped-on profiles first and starts pulling less-popular profiles the more its used.

“We have a learning application where others don’t,” Lewis told BizWest. “Where other applications are casting a wide map of ‘here is the gender I’m looking for, here’s the proximity around me’ and once the application is fed that information, the results are based on popularity. It’s frontloaded with the most popular profiles. That’s not by definition a real algorithm.”

Say Allo got its start in 2015 from Lewis’ personal life. When his relationship ended, it was for compatibility issues: Lewis was undergoing the stress of his company at the time being acquired, his political views didn’t match with his partner, and he had a background of being a Marine, an entrepreneur and a single father that didn’t click with his then-partner.

Since then, he was able to partner with Steve Shaw, who helped built Amazon’s learning algorithm, Kris Collins, who previously developed apps for Audi and The North Face and with Board Member Brian Shaw, a co-developer of cognitive behavior therapy to develop the new dating app.

Say Allo says it has a more accurate product for long-term matches: “The message we’re trying to get out,” Lewis said, “is when people are done messing around with other apps, they can find compatibility with Say Allo.”

But the next hurdle the company will face is marketing.

Launched in the Denver/ Boulder area last month, the app already has 1,000 users. But Say Allo will have to compete with Bumble, which has 12 million users, and Tinder, which has more than 50 million. To do that, Lewis said the company is launching some aggressive marketing campaigns. The more users it has, the more data it collects, which means more for the algorithm to learn from and match better.

There are other things Say Allo is doing to set itself apart. Like Tinder or Bumble, it’s free to use, but customers can choose to upgrade for plans from $9.99 to $19.99 per month. Those plans let users see who is interested in them and video chat with a match, a feature which the other apps don’t yet have.

“For the safety factor alone, you can meet someone in a safe environment without giving personal information,” Lewis said. “We’re charging about the cost to go out on a date in person with someone. Sometimes all you need is 30 seconds in-person to decide if someone is compatible. With the video meetup, you can save considerable time.”

So far, there’s been about a 20 percent conversion rate of people who opt for the premium features.

Looking ahead, Lewis said the company plans to grow Say Allo in the Denver/ Boulder area and then in Colorado before launching nationwide. Success for the company would mean job growth in Montreal, where the app’s AI feature is built, and Denver, where the company’s business side is run.

“Our objective is to grow inside and outside the United States as quickly as possible,” Lewis said. “The growth for the industry is in the next two years, 672 million people are going to be using apps like these. Now, there are roughly 511 million. We want to be part of that space. Over the coming months we’re going to have fairly aggressive marketing campaigns, putting everything behind this to attract people to our platform. We’re going to be a fun company to watch.”