There are already 120 potential game-changing machine learning applications in 12 industries — and the evolution is accelerating. Join our latest VB Live interactive event for a deep dive into how machine learning can directly impact your business results.

Register right here for free.

What machine learning does best: apply vast and intricate sets of data to complex problems, and come up with solutions no human could have.

At REX, a real estate service platform, the complex problem seems deceptively simple at first glance.

“What we’re doing at REX is trying to take a very very complicated consumer transaction — the selling of a house and the buying of a house — and we’re trying to use data and algorithms to go way beyond what a traditional real estate agent can do,” says Andy Barkett, VP of AI and machine learning at REX.

One of the biggest issues they tackle is one of the most challenging all realtors knock into: lead qualifying. Seriousness, commitment, and timeframe vary substantially when sellers start thinking about putting their home on the market, and sifting through the detritus has been always been a stumbling block that can lead to wasted time and lost sales.

Data to the rescue, Barkett says.

“When someone creates a profile on our site or signs up indicating they may have interest in selling their house, we get about 1300 pieces of data about that person,” Barkett explains.

They get the info from traditional consumer marketing companies, which means they’re capturing an extraordinary variety of information. It’s everything from number of kids and demographic data to likelihood of buying home office furniture.

That data gets poured into what Barkett calls classic supervised machine learning, which means the algorithm is actually tuned to reach a specific outcome, like signing up to list their house or actually picking up when someone from REX calls.

But though they don’t actually know in advance which information might actually correlate to the outcomes they’re looking for, it’s not just throwing a vat of jello salad against the wall and seeing what sticks.

“We let the machine figure out which of those variables are actually predictive, and we continually adjust that over time,” explains Barkett. That’s because while, for example, being in a certain income bracket is predictive one month, the market will change, liquidity will dip, and a month later, that variable might not be as predictive.

“So rather than try to speculate about what’s predictive, and rather than try to do a simple, one-time static analysis of variables to figure out which is interesting, we just continuously run a machine learning process and let the machine decide which variables are predictive of those outcomes,” Barkett says.

But the need to consistently iterate, test, and iterate some more requires setting expectations across the country, he’s found. His sales executives were initially under the impression that it was a one-and-done process.

“The expectation was essentially you run the machine learning process, and then a model comes out of it,” he says. “And then once that model has been built, they get better information about which leads are good ones. The reality is that we actually ran the process, and the first time we ran it,I’m not sure it gave us anything useful.”

It’s a process, he emphasizes.

“As we continually try things, every time we try a new experiment or add a new piece of data or tweak something, the model gets one percent better,” Barkett says. “It took something like 30 iterations before we got to a place I think we were genuinely adding value for them.”

To learn more about how machine learning is continually disrupting industries across the spectrum by actually producing results — and how your company can get in on the action — don’t miss this VB Live event.

Don’t miss out!

Register here for free.

In this webinar you’ll:

  • Learn how cognitive technologies scale across mobile devices (including cars)
  • Evaluate the value of a machine learning product to your organization
  • Tailor your data structure to optimize for future machine learning initiatives


  • Andy Barkett, VP of AI and Machine Learning, REX
  • Daniel Lizio-Katzen, Senior VP Product and Revenue, Fareportal
  • Stewart Rogers, Director of Marketing Technology, VentureBeat
  • Wendy Schuchart, Moderator, VentureBeat