An algorithm that identifies agents who work independently of managers. Another that determines which agents are most likely to use co-listings. And a team of data scientists capable of culling information to parse the most appealing characteristics of a high-performing agent.
With 2 million transaction records ingested every quarter, Realogy, the largest real estate holding company in the United States, has a stark advantage over its competitors in the form of a robust data trove it uses to sharpen agent recruiting nationwide, company CEO Ryan Schneider said during a presentation Wednesday in New York City.
“We built a machine learning model that matches their characteristics to our characteristics,” Schneider said of algorithms his company deploys to match the compatibility of potential hires with Realogy’s unique corporate culture. “It’s all about trying to predict who’s going to be able to drive the most incremental productivity in the future.”
Because of its national scale, Realogy has brokerages and agents operating in every multiple listing service in the country, and over the past year it has leveraged data from those platforms to create a machine learning algorithm to drive recruiting, Schneider revealed during a presentation at the Evercore ISI Industrials/Housing & Building Products/Airlines Conference on Wednesday.
Realogy’s algorithm isn’t necessarily designed to find top-producing agents, or what Schneider calls “the whale,” because those professionals are easily identifiable in each market. It’s more aligned to identify agents who will grow comfortably in Realogy’s environment.
The high-tech strategy is reminiscent of the 2011 Brad Pitt film “Moneyball,” based on the book by Michael Lewis about the unconventional recruiting methods of Oakland A’s General Manager Billy Beane, who’s analytical, evidence-based approach led the baseball team to the playoffs in 2002.
“We take about 75,000 of the top people and put that out to local managers every quarter to really let them use that to prioritize their agent recruiting,” Schneider said. “It effectively helps them spend their resources.”
As Realogy ingests more data, the algorithm strengthens. Meanwhile, Realogy’s data science team has discovered that, out of the 10 most important characteristics identified, five are Realogy-specific, which Schneider believes hands them a competitive advantage.
Schneider gave two examples of that Realogy-specific analysis: The algorithm identified that agents who work autonomously may not be a good fit for Realogy, a company that operates in an environment with dedicated managers. Agents who are more inclined to do co-listings are more likely to succeed at Realogy than the average agent, the algorithm found.
Schneider has been with Realogy for slightly more than a year, coming over from Capital One prior to working in real estate. He said one of the things that surprised him in his first year was how little technology and data disruption there has been in the industry.
“This whole industry has a lot of opportunity to use technology and data better,” Schneider said. “I believe we’ve got the scale and the data assets to be the leader in driving that change.”
“I personally haven’t seen very much change at all in what the consumer and the agent experience have been in the last few years,” Schneider added.