This is part one of a two part series about the use of data science in commercial real estate. In order to find out more about how property firms are innovating using data science we interviewed six leaders in the space that work for a wide variety of companies. Read the second part of the series on how to make data science fit into a real estate organization here.
The idea of data science has been around for a long time, rooted in the field of statistics. The difference is that the last couple of decades of advancements in computing power, data warehousing, and data architecture, combined with the past decade’s massive leaps in cloud computing and open-source tools, have enabled practitioners to turn theory into practice. From this, almost every industry has started to find ways to improve decisions with computationally driven analytics, finding valuable patterns in increasingly complex data sets.
Macroeconomic and demographic data has been limited in its ability to predict the performance of commercial real estate assets. But data science has started to usher in a transformational era for the industry. Commercial real estate firms can now access more market data and leverage data science skill sets and powerful technologies to identify relationships between asset performance and market factors. This can then be used to optimize investment and portfolio performance. Data is becoming an integral part of many commercial real estate organizations, each with its own specific use case that helps them challenge the “traditional” way of doing business.
Assessing opportunities at scale – Joanna Marsh at Investa
As the General Manager of Innovation and Advanced Analytics at Investa, one of Australia’s largest institutional real estate investors, Joanna Marsh spends a lot of time thinking about how to boost her organization’s ability to make better decisions. One of Marsh’s first priorities involved creating a system that would flag assets due for a transition. By looking at the property data points ranging from liens to development approvals, Marsh hoped to find signals that could generate an early advantage in the acquisition process. When she started at Investa in 2018, her efforts were aided by advanced technologies that were finding their way into commercial real estate: machine learning and artificial intelligence.
Marsh knew that those two advancements would open a much bigger world of potential use cases. But that presented a series of huge issues that she had to overcome. First was the technical challenge. “The problem with most off-the-shelf AI and machine-learning products is that they don’t work in real estate,” said Marsh. “We’ve been seduced by the promises that AI companies make, and the industry is now jaded since those promises never manifested. So the industry gets more comfortable with going back to Excel, and the way they have always operated.”
But knowing that technical paths would eventually materialize, Marsh had to confront a second huge challenge: “We came up with 100 potential use cases, and that was just the beginning. To get real, we had to prioritize potential use cases based on strategy and cost,” said Marsh. “After winnowing down the ideas, we translated the technology and financial expectations into a weighted multi-criteria analysis model. And the use cases that flowed to the top were the ones where enough input data already existed to make the time-to-return attractive.”
After narrowing down the concepts, there was still one more huge wall: the internal realities of the enterprise. “For projects to gain traction, we needed to hook into one of the income levers of the business to help pin use cases against fees and revenues,” said Marsh. “That’s part of why we like leasing use cases, since those generate high volume where we can drive real revenue generation.”
Marsh may have created one of the industry’s more sophisticated processes for assessing and advancing data science concepts in the commercial real estate field, but in the end, her most challenging obstacles still involve the human element. As Marsh explains, “A lot of people in the industry are really good spreadsheet jockeys, so we always need to ask, what would they rather be doing? And once we really figure that out, we advance.”
A lot has changed since Marsh’s early attempts. Still, she says that the process has led to a much stronger understanding of her fundamental objective at Investa: support internal teams to win on asymmetric returns.
A personal journey to scale in data science – Mark Franceski at Avison Young
Mark Franceski didn’t come to data science from a coding or statistics background. Before joining Avison Young last year, Mark Franceski spent his prior 10 years as the Vice President of Research for The Bozzuto Group, a large East Coast multifamily management, development, and construction firm. As Bozzuto’s portfolio doubled during his time there, Franceski realized that growth required a much stronger level of analysis than before. But even harder to confront was how that shift would also require a much stronger level of analytical skill development in himself.
One of Franceski’s realizations was that as the world of data science was emerging, the nature of his work would change. According to Franceski, “That required big shifts to upgrade the types of data inputs used, upgrade how analysis would be presented and shared, and a lot in between.”
Serving as both the executive and the doer back at Bozzuto, Franceski realized that he had to upgrade his own skill base. “That required learning how to deal with dimensional modeling, Power BI, inexpensive low-code options platforms, and deploying platforms like StratoDem Analytics to squeeze out deeper insights faster than prior tools,” according to Franceski.
The net result in that chapter of Franceski’s work was pivoting from generating reports to creating an agile engine to handle a wider range of use cases. “Back at Bozzuto, we used to deliver slide decks. Now the organization uses dashboards, from the property level all the way up the chain to the chairman,” said Franceski.
Now as an Executive Director at Avison Young, serving as the global AVANT leader for the residential and capital markets sectors, Franceski has a much larger scope and platform to drive data science advancements. AVANT is Avison Young’s predictive analytics system to generate stronger market insights for the firm’s clients.
As Franceski describes the AVANT system, “There’s an advantage created in integrating data to pick up trends, enable better mapping and visualizations, and make higher-level analysis understandable for clients. It’s created a first-mover advantage over our larger competitors, and scored some early wins so clients are betting that we’ll continue our progress towards real advancements in predictive analytics.”
Franceski has been betting on data science, and Avison Young has been betting heavily on AVANT. According to Franceski, AVANT started with a small team from across the industry. Now there are more than 150 employees around the world working on AVANT. This puts Franceski at the forefront of a major data science initiative, globally across the multifamily and capital markets categories, in part because of a continual effort to build his personal data science knowledge as fast as the data science field advances.
Strategic bets on where to spend – Andrew Weakland at W.P. Carey
Andrew Weakland is Senior Vice President and Director of Systems Development at W.P. Carey, one of the largest net-lease REITs with 1,400 properties across North America and Europe. Despite the technological advancements he has brought to his company, there’s a twist to how Weakland leads the way: he’s not investing in building a large in-house structure for data science.
“I would never try to hire a top data scientist internally,” said Weakland. “Our core focus is real estate investing, and that talent is also rare. But for a top-tier data scientist, you’re also competing against the FAANGs (Facebook, Amazon, Apple, Netflix, Google), and I simply won’t compete in dollars for top engineering talent,” said Weakland. “By the time you build out that team, the cash committed for that team would be better deployed into growing the real estate team. Or even deployed directly into real estate.”
Yet still, W.P. Carey expects Weakland to generate insights that can help the entire team beat the market. That requires Weakland to figure out how to harness best-of-breed applications in the market, and that job has become easier as the quality of applications has accelerated. “The amount of venture capital that’s flowed into PropTech is helping the market mature faster,” says Weakland. “And the best of the vendors are making those tools more digestible for the commercial real estate market.”
In the end, just having a data science team still doesn’t help a property firm make better decisions. For that to happen the data needs to get into the hands of decision makers. “For us, we’re looking for solutions to move from the constant requests for reports to putting the data in front of the team when they need it, instead of in the hands of data scientists,” said Weakland.
Data science is a powerful tool for commercial property companies but for Weakland, those tools only help if they are correctly deployed. “This isn’t about theoretically bringing in tools to replace a team,” he said. “It’s about making sure the tools resonate with people who’ve spent their entire careers in real estate. These tools go nowhere if someone asks, ‘why does this matter to me,’ then disregards the tools.”
As Weakland points out, “These tools help our work get better when we find points that match the intuition of the professionals on our team, and tell stories that match their innate feel of the market. It doesn’t help if we tell people they are wrong or aren’t looking at the right factors.”
“In the end, we make highly complex binary decisions. Should I buy or sell? And we have to consider lots of variables and factors,” said Weakland. “And we need to optimize the human intelligence of highly niche talent by arming them with better tools. Machines aren’t going to be making those binary decisions for the humans on our team. But machines are getting better at helping us find anomalies, which our teams can then exploit.”
Weakland’s approach to data science deployment may differ from some, but as he points out, “We aren’t tech professionals. We just happen to be industry professionals who are one step ahead on the tech side.”
The impact of data science on real estate – Zander Geronimos at MetaProp
The need for more data science technology in the property industry has produced an ecosystem of innovation around it. As the head of strategic partnerships and business development at venture capital firm MetaProp, one of the prominent early-stage venture capital funds in the real estate arena, Zander Geronimos has an unusual vantage point for observing the path of real estate data science. Despite the current downturn in the venture market, he sees a lot of new entrants into the space. “We’ve been getting more excited about ventures that find a small number of top firms as clients, build for those needs, and overserve those customers,” Geronimos said. “Then it’s the incumbent and mid-stage companies like Altus Group, MRI, VTS, and Realpage who are racing to tie pieces together.”
But real estate data is only a piece of a larger portion. More often than not Geronimos sees companies creating what he calls a ‘symphony of software’ where investment teams across asset types start to cooperate from a data perspective. “It is only a recent phenomenon that will continue to grow as folks become more comfortable sharing information through their platforms,” he said. Data science has come a long way since the term was first invented but there is plenty more room for innovation when it comes to how it is used in real estate. A few innovative players are already finding success using big data to inform their real estate decisions. Once they begin outperforming their peers it will be a race to see how fast property companies large and small can find ways to harness data science.