Making money in real estate requires big bets to be made. Knowing what property to buy at what time can be the difference between a cash cow and a money pit. For Adam Brueckner of Silverstone Partners, when he found that between 50 to 80 percent of excess returns over a five-year period are driven by market selection, it forced him to rethink the role of analysis for his organization. “Deploying data science and elementary machine learning becomes a valuable tool for building a strong market-selection framework,” says Brueckner. “It’s fundamentally about understanding what matters most and what doesn’t matter as much.”
As easy as it might sound to “use data” in making big real estate decisions, there are plenty of barriers to doing so. Some of these barriers are external: data can be hard to find or inaccurate and there is little to no standardization between providers. But other barriers are internal. Switching a company over to a new method of doing just about everything requires rethinking the entire organizational structure and can aggravate senior leadership who prefer to keep their current processes.
Among the leaders interviewed about their data science initiatives, change management was the biggest topic of all. Changing the way an organization allocates their resources and augmenting investment decision making with data science is a considerable challenge. Each company must find its own way through the journey of adopting data science. There is a spectrum of ways to approach the change, each with its own challenges, here are examples of a few companies that did so and succeeded.
Making internal adoption work – Marc Rutzen at Walker & Dunlop
Three years ago, Marc Rutzen was among the early-wave PropTech entrepreneurs racing to scale a venture out of a small office in Chicago. Now he serves as Chief Product Officer for Walker & Dunlop, one of the nation’s leading capital providers for the multifamily industry.
His previous company, Enodo, parsed rent rolls and operating statements, then meshed that with other data to analyze rents, expenses and ROI on multifamily investment properties. But the volume of available data is often a limitation for small companies. “When Enodo was a startup, users ran documents through Enodo, but there was only so much we could do at that volume level,” said Rutzen. “But after we joined Walker & Dunlop, we had another 40,000 rent rolls and operating statements coming in each year, significantly accelerating our predictive capabilities.”
More opportunities for analytical advancements have opened up for Rutzen now that his team has launched multiple internal products at Walker & Dunlop. His client base may consist of internal departments, but as Rutzen explains, a captive market doesn’t necessarily guarantee success.
Among the challenges that Rutzen points out is that most large firms generate a lot of data, but much of it is not ready to deploy for predictive analytics. “Much of our first three years with Walker & Dunlop were spent collecting and structuring data,” said Rutzen. “Preparing internal data for analysis isn’t as simple as putting it all in one place.”
Rutzen explains, “We designed software to collect the data that people generate in their day-to-day work. To motivate users to use our products, we had to work with the business to rethink some processes before building tech around them to ensure the most efficient automations.”
As a simpler example of that process, moving users off Excel is a herculean task in the commercial real estate field. But Rutzen’s team realized that for some of the data they needed to capture, the best approach was to meet people where they work. Rutzen’s team built an add-in to push and pull data from Excel models to the cloud where the data could be secured and analyzed. This assured that data previously trapped in Excel could be used for broader business intelligence, and also made it possible to lock the models so they couldn’t be used outside Walker & Dunlop. By moving data and business logic to the cloud, Excel became more of a way to securely view and analyze data than a place to store it.
In talking through the issues of building successful data science platforms, Rutzen also points out the highest-priority challenge for most companies. “Adoption is tough,” he says. Even with ‘captive’ customers already within the organization, advancing the work requires thinking about customer adoption every bit as aggressively as an independent company serving external clients.
“We have a full customer success team, chat support, explainer videos, constant usage monitoring for usage, and detailed feedback collection processes,” said Rutzen. “We treat our internal stakeholders like customers…really close customers.”
All this effort has enabled the rollout of several major products to deploy data science across the organization, answering questions such as which loans are most likely to refinance or sell, what order tasks should be performed to optimize the due diligence collection process, and which new clients and markets will most likely match previous investment success. As Rutzen points out, “Thes are now answerable questions.”
Creating the optimal environment to put data science to work – Adam Brueckner at Silverstone Partners
Adam Brueckner, Managing Principal at Silverstone Partners, understands what it’s like at both established estate investment funds with significant resources to deploy data science and at small startup real estate investment funds that don’t have that kind of luxury.
At more established funds, there are a couple of sources of headwinds, explains Brueckner. “First, the real estate business is much more than just punching a button on a Bloomberg terminal to buy. Real estate assets are complex and always have nuances and idiosyncrasies. That typically requires a lot more people involved within these complex organizations while a smaller fund can be very nimble.”
“Within the real estate industry as a whole, there is generally less familiarity with data science as it is simply at a very early stage of exploration. When presented with a machine learning black-box model and someone says, ‘trust us,’ this is understandably a very difficult bridge to cross, even for seasoned, established professionals who possess several decades of experience. Ultimately this type of analytics is intended to augment, not replace the existing processes, but the unfamiliarity is a large hurdle,” said Brueckner. “Until machine learning becomes explainable, it makes sense to hold a degree of skepticism.”
“Then there’s the headwind that delivering strong data science outcomes takes time,” said Brueckner. “Even 18 months would be pretty good, but that can invite a downward spiral for data science funding when they can’t generate good results fast enough.”
“And one of the big headwinds is simply cultural. At more established funds, there’s a base of smart professionals who really understand real estate, but built their expertise before data analytics became part of the picture. And they bring on new hires who understand analytics, but not real estate,” said Brueckner. “The result can be like oil and water.”
“The bottom line is that it’s hard to turn a real estate company into an analytics-driven company overnight,” said Brueckner. And that realization led him to his logical conclusion: launching Silverstone Partners, a boutique real estate investment fund driven by advanced analytics at the core, which is a theme we see for some of the most promising newly formed commercial real estate investment funds.
At Silverstone Partners, that means conducting traditional analysis and advanced analysis side-by-side as well as building a team where everyone has a high level of comfort working with advanced analytics. “We still work with the brokerage community, still drive through neighborhoods, still talk with tenants, still examine buildings, still try to assess how it feels in the parking lot,” said Brueckner. “We still do all the things that other real estate investors do. But we’re willing to make the bet on data science advancements. Uncovering emerging trends and pricing discrepancies is doable. We’ve seen enough to know that this will be a differentiator.”
Brueckner points out, “In the hedge fund industry, a lot of inefficiency of the market was driven out by quantitative hedge funds. But in real estate, 95 percent of the industry still invests with a traditional mindset. And this gap is going to create a big advantage for firms willing to leverage both the traditional process and a more advanced data-driven evaluation process…at least until most of the field adapts to this shift. But for now, it’s an exciting time to explore the art of the possible.”
Fitting into the internal flow – Gabe Greenberg at Tishman Speyer
“What was most interesting to us is the idea that in other industries, there’s a huge number of data sources,” said Gabe Greenberg, the Senior Director of Data Analytics at Tishman Speyer. “But in real estate, there’s not really that much data, and everyone has access to most of the same sources.” And that realization led Greenberg’s team to head down the path of connecting extensive data across every possible tax parcel.
Then reality set in. “We looked deeply across markets, built dashboards and tools that analyzed millions of properties across markets, and the acquisitions team thought it was interesting. But they didn’t use it,” said Greenberg. “Those weren’t the tools that they wanted to address the questions they needed to answer.”
But unlike at some other firms where data science efforts fizzle if the early results are inconclusive, Greenberg continued to discuss with stakeholders and sharpen the internal product fit. “We figured out what work is repetitive. There’s better adoption when we take the boring pieces and get those done more effectively,” said Greenberg. “We aim for the questions that are repeatedly asked, then build the tools that make it easier to pull out the actionable insights.”
Now, one of Greenberg’s focuses includes enabling the rapid scanning and scoring of the firm’s relevant markets, as well as tracking changes in underlying submarkets. “By quantifying pieces of the underwriting process and systematically identifying submarket characteristics for stronger investment results, the qualitative human overlay can kick in faster, acquisitions teams can present better deals, and our investment committee can make more informed decisions,” said Greenberg.
“Rather than building out overly complicated infrastructure, we focus on the specific analytics that acquisitions teams need to move forward or pass on a deal,” said Greenberg. “That matters because there are firms that work at immense scale over a huge number of markets, and there are firms that are extremely knowledgeable in a few home markets. We’re carving out our own space by competing with a unique set of insights that others don’t have.”
When it comes to business outcomes, just talking about using data is not enough. The companies that are excelling in the way they use data science are the ones that dedicate time and resources to it. Oftentimes that can even mean restructuring an organization to give technology champions a larger seat at the table. For true incremental improvement in the way a company uses data, the entire organization needs to be thinking about innovation. While that is not always easy, it is necessary for property companies looking to make smarter, data driven decisions. Change is never easy, but it is necessary to keep up with the ever evolving property market and the accelerating pace of technology. Property firms that are able to take advantage of all of the ways that data science can help make better investment decisions will be the ones who find the best deals and deliver the best returns in the future.