Troubles at Zillow demonstrate problems with big data

Tim Whitney

Zillow has suffered a really big hiccup in its business.

It’s stock value has dropped 70 percent since earlier in the year as the company lost more than $500 million. In addition, Zillow is expected to lay off a quarter of its work force as it shuts down the Zilllow Offers segment of its operations.

In my opinion, it sounds like the Zillow iBuyer — short for “instant” buyer — program fizzled because the company started believing its “Zestimates” were valid.  

I’ve never met a real estate broker who thought the Zillow Zestimates were spot on and, therefore, a competitive market analysis wasn’t needed to determine the value of a property. The only people who believed the numbers were accurate were sellers — for obvious reasons. If your algorithm can’t figure out when you’re in the midst of a supercharged market, you need a new algorithm. Even the head of Zillow finally admitted the company failed to predict the pace of home price appreciation.

The way I understand the concept is Zillow would use its big data to generate a future value of a particular property and then try to back into the price they could afford to pay. Add in a quick estimate of renovation costs along with a profit margin, and now you can calculate the sales price you’re targeting. While this tried-and-true method might work on a small scale if you know your market, it appears difficult to manage on a large scale.

There are a few other platforms out there — Redfin, Opendoor and Offerpad, among them — acting as fix and flippers and showing some success in a few targeted markets such as Phoenix and Orlando where they’ve been using big data as well to guide their decisions.

While I believe you might gain an edge over other buyers and sellers by having lots of data, it doesn’t guarantee results. 

Did I mention all real estate is local?