Using AI to Determine the Best Use of Real Estate


Real and personal property is a basic delineation in English common law that corresponds roughly to the differences between immovable and movable objects.  Interests in land and fixtures, such as permanent buildings, are classified as real property interests. The real estate operations industry consists of companies engaged in developing, renting, leasing, and managing residential and commercial property interests. The industry includes real estate brokerage and agent services, real estate appraisal services, and consulting services. The real estate operations industry excludes real estate investment trusts (REITs). Residential real estate is one of the largest sectors of the U.S. economy and supports millions of professionals who provide services related to home purchases, rentals, financing, and home maintenance and improvement.

In 2015, sales of 5.3 million existing and 490,000 new homes in the United States had an aggregate transaction value of $1.6 trillion, generating $65 billion in commissions paid to real estate brokers. During that same period, there were approximately 46 million rental housing units in the United States, with a national vacancy rate of 7.3%. In the commercial real estate market, aggregate transaction values in 2015 were estimated at $313 billion.

In the past, property valuation has been one of the most important parts of a real estate broker’s job. Recently, the task of comparing current property valuations has become automated by tools such as Redfin Estimate and the Zillow Zestimate. With hundreds of variables to analyze to correctly determine the value of any parcel of real property, predicting real estate values is a perfect use case for artificial intelligence (AI). A Seattle-based startup called CityBldr has created a software-as-a-service (SaaS) platform using AI to help determine the best use of all properties, to help developers find the most underutilized properties, and to help property owners understand the value of their properties as potential development sites.

Current property valuation methods compare sale prices of proximal similar properties. For example, a home in one neighborhood sells for $500,000, so a similar home nearby would use this comparable sale to substantiate its current value. But this may not be the most profitable use of the land. For example, if the property is zoned for other types of construction, it might be more valuable to tear down the house and turn it into multi-family housing, or a multi-story hotel, office space, or a number of other uses depending on its location and the need for various types of properties in that area.  CityBldr uses AI to predict what a developer might pay the property owner for the site’s development value.

The tool draws on 16 different public sources of data including zoning codes, tax history, transit, and parcel data, and has generated three proprietary data sources.  In the end, it analyzes more than 180 variables on 118 million U.S. properties to determine how plots of land can be improved to maximize their value.

According to CityBldr CEO Bryan Copley, “In addition to identifying every building use permitted on every piece of property, our predictive analytics help developers and investors make decisions based on forward looking indicators versus lagging indicators alone. In our case, we decided to use AI because we realized computers could empirically, scientifically understand the best use for all land better than humans ever could. Machine learning helps us make sense of disparate data sets and defines clear relationships between actions and outcomes in real estate development.”

Comments are closed.