Big data, the vast amount of information generated daily in our connected world, is upending business models in industries from finance to tourism. It is also permeating real estate, a multi-pronged, multi-billion-dollar sector known for its reliance on tradition and intuition.
“Real estate has traditionally been a late adopter in tools and technologies,” said John D’Angelo, who leads New York-based Deloitte’s real estate sector in the U.S. “The adaptation of data and analytics is no exception. But I think what we’re seeing broadly in the industry, really in the last 18 months, is raising awareness of the value of large data sets and putting data and analytics to work alongside people.”
Today, there is a growing cadre of technology startups, as well as behemoth real estate companies, that are beginning to seek insights through the real-time crunch of enormous data sets that power Artificial Intelligence tools. The generated knowledge is starting to underline vital real estate decisions—what property to build, how to price a for-sale residence, and how to reach buyers, among others.
A New Kind of Data
Utilizing big-data-fueled tools, developers better gauge potential projects; agents more efficiently serve clients; and home shoppers and sellers receive a real estate experience free of clutter and redundancy.
Yet, to understand the potential of “big data,” a buzz term splashed across headlines and uttered at industry conferences, it’s useful to differentiate it from the conventional data collected in the real estate industry.
The customary type of market and property information— such as lot size, price, area population growth, nearby schools—resides on multiple-listing services and is being democratized through websites such as realtor.com, Zillow and Redfin.
In real estate, big data is relatively novel, non-traditional data that illustrates granular insights not before gleaned. Examples include the amount of light a home receives throughout the day, the level of noise pollution in a neighborhood, residents’ preferences for gym equipment and the popularity of nearby entertainment spots.
“It’s not just about when [a property] is built; is it made out of wood; how many stories is it; how many bedrooms and bathrooms does it have,” said Zach Aarons, co-founder and partner of MetaProp, a venture capital company in New York City focused on real estate technology. “Now there’s 200 other data points” to potentially parse.
That proliferation of information boasts the ability to augment the answers to some of the most fundamental questions in residential real estate, especially on the high end.
How to Price a Home?
Setting an asking price is a crucial exercise that often hinges on the agent’s local knowledge and expertise. Similar homes are considered; neighborhood amenities are factored in; price strategies are talked over. Both agents and developers rely on conventional data about the market (think current supply and past sales) and the property (think size, quality of used materials, number of bedrooms, among other features) to set the price. Developers also consider the costs of construction. No matter the pricing model, though, the right listing price translates into a successful transaction. An ask too high or too low means a home could struggle to draw in buyers.
For luxury abodes, which usually take longer periods and more efforts to sell, price is paramount but also tough to determine. That is because high-end homes are often custom abodes with unique amenities that are hard to liken to anything else existing on the market. Hence, subjectivity often plays a role in valuations. But agents often struggle with it. And, it turns out, so do some big-data-powered Artificial Intelligence algorithms.
Predicting the value of high-end properties is the “greatest challenge for the Zestimate,” Zillow’s home value estimation tool, said Jeff Tucker, economist with the company. In San Francisco, where many homes reside in the upper bounds of price ranges, the Zestimate has a margin of error of 3.6%, the highest for any major market, Mr. Tucker said.
“We’re actually able to use machine learning to examine features of the home that appear in the photos that will reflect its quality or its sale value,” Mr. Tucker said.
But for the algorithm—just like for real estate agents—the strain stems from the few “training data” points, Mr. Tucker said. There are simply not enough comparables in the luxury segment.
Yet, wielding AI to gauge the value of luxury houses is hardly a futile effort. When comparable data is scarce, technology can better connect the gaps and find price benchmarks than humans can, said Joseph Sirosh, chief technology officer of Compass and formerly of Microsoft.
Mr. Sirosh’s sentiment, however, does not discard agents’ expertise. “AI can actually make the best use of all available data to set the price,” he said. “It’s human judgment versus AI, and then both come together and the end result is actually significantly better.
“I talk about AI to empower AI, that is artificial intelligence to empower agent intelligence,” Mr. Sirosh said.
How to Market and Sell a Property
ARTICLE CONTINUES AFTER ADVERTISEMENT
Customarily, luxury real estate agents would tap their network to sell high-end properties, which call for a refined approach to marketing that often targets only a few thousand potential affluent buyers worldwide. Thus, the reach and strength of agents’ connections in the world of high-net-worth personalities easily break or make real estate deals.
But what Austin, T
exas-headquartered AI-powered brokerage REX, which does not advertise on multiple-listing services, has discovered is that technology might better market luxury properties than agents themselves.
“We learned this by selling some high-priced homes,” said REX Chief Technology Officer Andy Barkett. “We have a huge advantage in that our system does a really good job of advertising to international buyers. We find buyers in other languages because AI doesn’t really care who the buyer is.
REX holds broker licenses in nearly 20 states and Washington, D.C. Because of REX’s heavy reliance on data and technology to attract buyers to its listings, it charges sellers a fee of only 2%, three times less than the industry standard. That is what initially piqued Kingstone Shih’s interest in REX.
Last spring, Mr. Shih, a 52-year old oral surgeon, used REX to sell a vacant, three-bedroom, two-bathroom home in Palo Alto, California, for nearly $4 million. He never met his REX real estate representative, who lived in Woodland Hills, a community in the Santa Monica Mountains some 400 miles south of Palo Alto.
“I don’t think I even talked to her on the phone, everything was over email; the transaction was over the internet,” Mr. Shih said. “That’s not really the traditional way. You have to feel comfortable with that process.”
Through direct advertising and traffic on its own website, REX helped Mr. Shih sell the house in about a month..
The brokerage, Mr. Barkett said, tracks people’s interactions on its website as well as with its ads, adjusting the latter to glean information about home preferences, budgets and commitment to buying or selling, among other things. A sign that someone is interested in purchasing is getting pre-approved for a mortgage, Mr. Barkett said. Meanwhile, a serious seller may seek a deep-cleaning service.
What to Know About a Home?
REX focuses on people-centered intelligence with the aim to steward clients’ home buying and selling experiences. Meanwhile, other companies zoom in on the properties themselves. Israeli-founded Localize.city, which operates in New York, is a case in point.
Combining public and commercial data with its own insights, Localize.city churns detailed knowledge about New York City’s addresses that ranges from how noisy a neighborhood is to how likely a nearby development is to how dangerous the closest intersection is.
“There’s plenty of research that shows that 40% of homebuyers today regret their purchase after two years in the U.S.,” said Steven Kalifowitz, Localize.city president. “And, a lot of that is related to what they bought isn’’t what they thought they were going to get.”
Thus, Mr. Kalifowitz’s company, which employs data scientists and urban developers, produces and gathers a bevy of property statistics not available in traditional listings or city records.
For instance, the four-bedroom penthouse on Greene Street in New York’s SoHo neighborhood, which is asking $14.5 million, soaks in light for 10 hours a day in the summer, four more hours than what is average for Manhattan. Yet, the sunlight—and views—the residence flaunts today may dim in the future as a building permit has been filed for a lot nearby, Localize.city warns.
Localize.city sprouts that type of expedient snippets for any address in New York, luxury or otherwise; on the market or not.
Boosting Real Estate Profits?
AI is not only analyzing the properties of today. It can help build the homes of the future too. A residential development, arisen through AI-facilitated decisions, should ultimately better cater to residents by anticipating and advancing their urban lifestyles.
“Data allows us and our competitors to try to fine-tune our strategies as to where we want to build and what types of products we want to build and what sort of amenities and features we’re going to need to put into those new buildings,” said Jared Sullivan, vice president of research at Chicago-based residential and commercial real estate investment and development firm, CA Ventures.
A 2018 article by McKinsey said that non-traditional variables can account for substantial differences in the performance of residential projects that, by traditional metrics, are identical. From proximity to eateries to even seemingly small distinctions, such as the placement of a pool, may influence a building’s desirability and, thus, financial margins.
Even with units for sale, such perks maintain a monetary advantage for developers. George Ratiu, chief
economist with realtor.com, said that proximity to neighborhood amenities—from restaurants and grocery stores to bicycle lanes and public transportation—commands higher property values.
“Some surveys of home buyers and owners show that consumers are willing to pay a premium for proximity to public transit, even if they themselves are not using public transit,” Mr. Ratiu said.
Reducing Development Costs
By effectively aligning new residences with the priorities of their future inhabitants, who are willing to pay more for what they desire the most, big-data technologies can earn significant profits for developers. But AI can also act upon the other major variable in financial forecasts: cost.
A Miami-based startup, which makes zoning codes easier to parse, Deepblocks exemplifies how AI can reduce expenses for developers early in the life of a project.
Deepblocks’ CEO Olivia Ramos said the company does “back-of-the-envelope calculations,” which reveal the development possibilities for plots of land as well as the initial costs associated with them. Deepblocks has automated the process of preliminary property analysis, saving anywhere from “a couple thousand dollars to $40,000, depending on the size and complexity of the project,” according to Ramos.
Traditionally, a property analysis requires weeks— even months —of collaboration among developers, architects and financial performers, as well as others. Any alteration they make to a plan costs money and time.
“We’re taking those weeks of work and all those conversations into an automated digital process,” Ms. Ramos said. “That’s revolutionary because it means that now you can set 10 times as many options in a 10th of the time.”