The Zillow Fiasco
Zillow is an online real estate media company that generates revenue by selling advertising on its website. In 2011, Zillow and Yahoo Real Estate launched a partnership which created the largest real estate advertising network on the web.
But in an effort to expand and improve profits, the firm decided to bet its future on an algorithm-based home-flipping business. They named it: Zillow Offers. The plan was to use the algorithm to identify homes priced at or under the market, buy them, make minor renovations, and then sell quickly.
It was a decision the company would soon sorely regret. As part of the plan, Zillow would settle for a minor profit on the sale, and then upsell their clients through transaction fees, title insurance, etc. But the algorithm which was supposed to predict housing prices and trends fell far short of the goal. Simply stated: the algorithm didn’t seem to grasp and factor in all the variables in the very complex housing market.
By the summer of 2021, the firm’s executives realized something was desperately wrong. Zillow realized that they were paying too much for homes they intended to flip.
This month, Zillow admitted failure and announced it would permanently close Zillow Offers. This represented one of the sharpest recent American corporate retreats. The company cut about a quarter of its workforce (about 2,000 jobs) and wrote down losses of half a billion dollars.
As a result, the company’s market cap which reached a highpoint of $48.35 billion in February of this year, is now about $16 billion (a loss of about two-thirds of its value).
Computer-driven analysis has become more and more common in equity markets (stocks and bonds). Based on the success of the computer models in the stock brokerage industry, Zillow thought it would make sense to simply devise a computer model which would do for real estate sales what these models did for equity markets.
But Zillow soon discovered that equity markets and real estate markets are two entirely different species. After all, unlike stocks and bonds, homebuying is still partly informed by emotional attachments, personal tastes, and other non-data driven factors. Additionally, the real estate market, unlike the equity markets, is also driven by location. For example, an identical home can sell for wildly different prices depending on which city it is in – and in which neighborhood within the city. Also, residential real estate prices are affected by aesthetic, social, and other factors. As for equity markets, these variables do not apply.
Finally, if a three-bedroom home is advertised for a given price, the algorithm did not take into account the layout of those bedrooms – is it designed nicely, etc.
There were other issues Zillow missed in their very expensive failure. The housing market had become volatile; the pandemic was impacting all markets in a profound way. The supply chain issue and the labor shortage made renovations particularly difficult.
Simply stated: The algorithm was not “aware” of these serious obstacles – it had not been programmed correctly. Too many important factors were not taken into consideration, nor was the human factor.
In a microcosm, lets look at one Zillow purchase to get an idea of how Zillow Offers missed the mark (there are many other examples, of course). The firm bought a home in Winter Springs, Florida from Karin and George Dorsett. When the Dorsett’s were ready to sell, a local brokerage team said they believed the home could sell for $440,000. Zillow offered them just over $450,000 and a brokerage fee of just one percent. (Remember, Zillow is counting on the upsell for most of its profit.) The algorithm prompted Zillow to offer these terms, but the profit margin was too small, as was the commission. Hence, Zillow los money on the transaction.
Ms. Dorsett said, “We are so glad we sold to Zillow.”
Zillow might not be.