At this point in time, the importance of data in decision making is undoubtable. With the advent of big data and machine learning consulting, entire industries like finance have been transformed to a great extent. In this article, we explore the current applications of predictive analytics in real estate and impact of this technology in the near future.
Although real estate professionals often manage huge investments, for decades their decisions have mostly been based on a very limited set of financial factors. If it’s an already-built facility, they would skim through the property rent rolls and expenses. Even in large-scale property development projects, investors often consider land price and capital structure as top factors influencing their decision. Other than that, it mostly comes down to intuition and industry experience.
However, the number of factors that influence property demand has been steadily growing along with urban development. While every investor would certainly link the price to the property’s proximity to a subway station, these relationships are rarely linear. Instead, property prices are formulated by a combination of variables, which can also be interconnected with each other in unconventional ways.
In fact, nontraditional variables can have more impact on the end price than traditional ones. Today, winning investment teams consider not only the number of coffee shops in a nearby area but also their scores in Google reviews. Such granular statistics help investors assess property value as accurately as never before.
It’s also worth noting that the use of conventional datasets to assess property value can be unreliable. For example, besides the fact that U.S. Census data doesn’t provide the crucial insights that investors need, it can also be inaccurate due to its reliance on mailed responses.
Many real estate investors make their decisions based on historic data. Nowadays, however, critical data has to be processed in near real time, to follow the reactive data analytics strategy. With the exponentially growing number of data points impacting a property’s potential returns, the human mind is no longer capable of adequately accounting for all of them in a very limited timeframe set by the fast-paced real estate environment.
In the investment realm, decision-making speed often correlates with success. But by the time an investment team gathers and analyzes all relevant data to make a decision with the highest success probability, the opportunities are likely to be seized by others. This calls for predictive analytics consulting and advanced AI-based forecasting tools.
The predictive capabilities of machine learning have radically changed investors’ perspective on forecasting. With companies recognizing the unrealized potential of big data analytics to forecast market scenarios coupled with the advent of machine learning, the real estate tech sector has been seeing heavy investments since 2015. According to CB Insights, 2019 marked a record $8.9 billion in venture capital invested in real estate startups.
The Quantarium valuation platform’s models combine neural networks, genetic modeling, and machine learning to accurately assess properties in the residential real estate market. The Quantarium algorithm has access to data of more than 150 million households with more than 900 data points for each, making the company’s evaluation model one of the most accurate on the market.
While the company offers many applications for its model, portfolio services have become especially useful for both lenders and real estate agents. For instance, mortgage companies use Quantarium’s valuation model to predict when property owners are going to sell, default, or refinance, enabling them to identify future risks and opportunities.
In a nutshell, the real estate business often comes down to accurately predicting property value. In this sense, Israeli-based Skyline AI is one of the most prominent industry players.
The startup focuses on commercial real estate and has access to one of the biggest transaction databases in this sector. Most importantly, Skyline AI recognizes the importance of non-traditional data and puts many resources in research. The company claims that a certain combination of alternative datasets can reveal far more about an area’s economic growth than traditional historic data. For example, the company has discovered that the number of Airbnb listings correlates with the rent price fluctuations. Similarly, car ownership rates and credit card data can serve as reliable metrics of investment feasibility in an area.
Skyline’s state-of-the-art AI-powered model continues to prove its effectiveness. In March 2020, the company collaborated with AION Partners, a New York-based principal and operating partner for real estate investments, announcing another multimillion acquisition deal of two residential building complexes in Philadelphia.
Unsurprisingly, information retrieved from social media can tell a lot about our preferences, thus helping real estate professionals build accurate consumer profiles and predict investment feasibility in a specific location. Spatial.ai has developed the ‘GeoSocial’ dataset, which sources and analyzes data from social media conversations across 72 segments, allowing real estate organizations to better predict demand in selected locations.
For example, Griffin Morris, co-founder of Spatial.ai, reveals that people who talk about spending time with their families are more likely to have above-average income. This insight comes from combining machine learning and natural language processing, allowing access to quantified behavioral information down to the block.
In 2020, Spatial.ai partnered with Cherre, an award-winning real estate data platform, to augment social media data with that coming from thousands of public, private and internal sources, allowing investors to approach property value assessment more holistically.
By coupling machine learning and IoT data analytics, property management can better understand energy consumption patterns and optimize resource efficiency. Sensors can be placed in heating systems, elevators, and workspaces to collect data about occupancy in each space.
For instance, Deloitte uses PointGrab’s computer vision and ML-powered platform to optimize the utilization of its workspaces in its London headquarters. By utilizing PointGrab’s Virtual Traffic Line feature, the system can predict what exact space will get dirty in the next hour and notify cleaners in advance.
BuildinIQ’s Predictive Energy Optimization service improves the energy efficiency of both commercial and public buildings. By installing IoT sensors in heating, ventilation and air conditioning (HVAC) systems, the BuildinIQ platform can automatically adjust temperatures and pressure in air handlers. Continuously analyzing local weather forecasts, building occupancy, and energy prices, the BuildingIQ system fine-tunes the resource spending strategy for the next 12 hours. Moreover, the platform can detect anomalies in HVAC operations, allowing facility teams to proactively address potential risks.
However, it’s critical to note that IoT networks are susceptible to cybersecurity attacks. For example, in 2013, hackers managed to access millions of customer profiles by breaching Target’s smart HVAC systems. Even with anti-hacking measures in place, companies should take an active role in maintaining the highest levels of security on their premises.
Identifying potential home buyers and sellers correctly is among the biggest pain points of real estate agents. For decades, the most effective solution was in sending marketing materials by mail. However, nowadays the absolute majority of home buyers start their journeys online. With predictive analytics models, real estate agents can finally eliminate guesswork and better understand which consumers are likely to buy or sell their properties soon.
For example, SmartZip sources data from more than 25 datasets to generate the list of properties that are most likely to be sold in a selected area in the near future. In particular, SmartZip’s predictive analytics model analyzes a multitude of various MLS data points like property characteristics (number of bedrooms and bathrooms, square footage, etc.) and historic information (when the property was listed and withdrawn).
Unsurprisingly, much of SmartZip’s predictive power comes from social media data aggregation. Google search history and certain behavioral cues including liked posts on social media are also analyzed by SmartZip. Moreover, the model takes into consideration various consumer data like recent purchases. If a person repeatedly buys supplies for home renovation, together with other data points this could mean that the owner intends to list their property at a higher price. With such copious amounts of data to analyze, SmartZip helps brokers, lenders, and real estate agents to significantly narrow their target audiences and optimize marketing spend.
The reluctance to trust new technologies shouldn’t come as a surprise in such a conservative industry as real estate. Many companies still resort to lengthy feasibility studies that can cost as much as building a predictive analytics platform from scratch. Marc Rutzen, CEO at Enodo, an AI-assisted underwriting platform, claims that clients’ skepticism remains one of the most significant hurdles. However, as with any other disrupting technology, it’s just a matter of time when predictive analytics becomes an industry standard.
Other than that, tapping into advanced analytics is not a walk in the park. While real estate professionals have been dealing with data in one form or another since ever, in a predictive analytics context data aggregation, cleansing and processing are completely novel processes. Getting into them requires a strategic effort on behalf of a real estate company to become data-centric. In this case, using predictive analytics is just another logical step.
Some experts argue that the mass adoption of predictive analytics and AI in general is often hindered by messy and siloed data. Yes, some conventional datasets may not be standardized and large companies often suffer from their lack of interoperability. However, there is currently no shortage of data cleansing service providers. Moreover, as we discussed earlier, solid data governance practices should ensure that new data is ready to be injected into predictive analytics models.
The industry’s landscape is changing. With the power of cloud computing, access to lavish amounts of data, and rapid advancements in machine learning, these emerging predictive analytics applications will continue to change how brokers, agents, lenders, and investors approach the real estate business.
As new exciting relationships between seemingly unrelated data points are getting discovered and verified, conventional gut-based investment and marketing approaches will be rightfully challenged. However, despite the undeniably huge potential of predictive analytics in real estate, the industry players should still treat the outcomes of such models as recommendations rather than a single source of truth.
The full potential of these technologies is yet to be unleashed. Depending on the case, the most advanced analytics systems in real estate can reliably predict outcomes for a five-year time span maximum. With the advances in related technologies and even bigger datasets, we can expect a greater prediction accuracy within longer time spans.
Given that the mass IoT adoption is underway, predictive analytics capabilities will expand further, introducing even more data points and consequently enabling even more granular and reliable predictions.