What's your house really worth? (cont.)

By Jeffrey M. O'Brien, Fortune senior editor

Improving accuracy

The first step on that road is improving the accuracy of the Zestimates. Overall, Zillow has Zestimated the value of 57 percent of U.S. housing stock, but only 65 percent of that could be considered "accurate" - by its definition, within 10 percent of the actual selling price. And even that accuracy isn't equally distributed.

For example, 85 percent of homes in Los Angeles have Zestimates, and two-thirds have been accurate. But only 53 percent of homes in metropolitan New York have Zestimates, and only half of those are accurate. In Louisiana, where one in 50 homes is listed on Zillow, the site is just about worthless. (In at least one case, questions about Zillow's accuracy have taken on a political charge: A community activist group, the National Community Reinvestment Coalition, accused the site of discriminatory valuations in minority neighborhoods. "A lot of people have thrown rocks at us," Barton says. "Have we been sued yet? No, but I'm sure we will be.")

Squirreled away five floors below Barton's office, the company's VP of data and advanced analytics, Stan Humphries, is trying to improve those percentages. Each night his formulas churn through two terabytes of data, comparing every home with similar nearby properties and factoring in newly reported sales.

"When you want to see comparable homes - when you ask, 'What homes are like this house?'' - realtors do that with intuition," Humphries says. "We do it with machine learning. You can infer information. If one house always sells for more than everything around it, we might be able to infer that it has a view. This is a really hard problem to solve, but at some point in the next year we are going to produce valuations that people are going to be completely stunned by."

Barry remains dubious. Zillow has Zestimates for 99 percent of all Phoenix homes and claims that 72 percent are accurate to within 10 percent. But Barry tells of a family who recently came to him believing their home was worth a lot more than it was. Zillow told them it would sell for $505,000; Barry and another agent each independently put the figure at $440,000. He doesn't discourage clients from looking at Zillow, but he definitely suggests they take the numbers with a grain of salt. "It's a tool," he says dismissively. "These folks said they looked at Zillow, but after seeing the Zestimate, they knew they wanted someone with knowledge of the area to see their house."

Finding the fairy tale

Back to my million-dollar pricetag. By now, it's early January, almost a month since I first visited Barton and Frink in Seattle. I haven't heard a peep about my Make Me Move price, and suddenly, just as I'm wrapping up this article, an e-mail hits my in-box. It reads, "Hello, we are neighbors of yours, looking to buy a home in Potrero Hill. In an effort to not waste your time, we are wondering if your million-dollar price quote is at all flexible. We are able to buy in a range close to there, but not quite there. Thank you very much in advance for your time."

Whoa. Now what? My wife and I discuss whether we have any wiggle room and decide that the first order of business is to call the woman and see if she's serious. The following morning, that's what I do. She is. As it turns out, the potential buyer and her husband are perfectly suited to play the role of the house-seeking couple in the fairy tale. They love the neighborhood and have been frustrated by their inability to buy. With the help of an agent, they've made three offers in three years, only to be outbid each time. Oh, yeah, and she's pregnant. They want a place to raise their family and are sick of going through the usual routine. "I've seriously considered putting a note up at Farley's," she says, referring to the local coffee shop, "or writing a form letter and putting it in mailboxes."

She's been a fan of Zillow since she found the site several months ago and immediately identified Make Me Move as a great way to draw a bead on houses before they go up for sale. Scouring the site, she found six pricetags in my neighborhood and inquired about mine and one other. As soon as she talked to the other guy, he decided it would actually take more to make him move after all, and he chickened out. I remain convinced that my number is fair. But she can't get there.

So I propose an idea that'll take $60,000 right off the top: Let's cut out the middlemen. She's torn. She feels loyalty to her agent but is not sure how much he's really worth. "We're doing 95 percent of the looking and all the legwork. He really just comes in when we're ready to make an offer," she says, adding that she'd be nervous about not having a pro to see the process through to the end.

Ultimately, the buyer and I decide to chill, think about how we could make the transaction happen, and pledge to strike up the conversation anew in the spring. Which is fine by me. With a buyer on hold, I'm able to begin scouting a dream property for my family. Just five years ago such a search would have started in a newspaper and ended in the office of an agent, where we would have probably competed with dozens of other bidders. Today I might be able to hop on Zillow and buy a house that's not even on the market - without dealing with open houses, bidding wars or buyer's remorse. Or even an agent.

In an opaque market, where the closest approximation of the value of a home lies in the head of a knowledgeable broker - say, Phoenix circa early 2007 - an agent can easily be worth the standard 6 percent commission. But in a perfect market, where the true value of a home is crystal clear to all parties and the fairy-tale note on a napkin comes in the form of an anonymized e-mail, 6 percent begins to look an awful lot like a luxury. Kind of like a canary-yellow Porsche.

Doris Burke contributed to this article.


Real estate: Who's buying now Top of page