Fun with Zillow — #5 — Don’t invite David Blaine or David Copperfield to your home
September 20, 2006
Zillow-bashing is a favorite pastime in the real estate blogosphere, second only to its cousin Redfin-bashing.
Tags: For buyers, For sellers, Heat maps, Mapping software, Microsoft Mappoint, Real estate, Real estate data, Real estate mapping, Real-estate-technology, Zillow, Zillow heat mapsThe Zillow bashers have (at least) a triumvirate of beefs with Zillow:
1) Its raw data is suspect
2) Its home-valuation algorithm, even if it were based on good data, could not possibly lead to an accurate Zestimate because it doesn’t, and can’t, consider non-quantifiables like views, amenities, smell, etc.
3) To protect consumers from over-reliance on a Zestimate, Zillow’s home page should have a big, bold disclaimer about 1) and 2).
In an effort to address 1) above, Zillow today announced that home owners would be able to “wikifi” their home’s data if necessary.
So is Zillow’s raw data (mostly from county records, I believe) really that bad? The results are in from a quick, highly unscientific study comparing Zillow’s data on 10 random properties in Menlo Park, CA, with local MLS records.
The county got 6 properties completely right. Of the remaining 4:
- One was off by half a bathroom. Not so bad, unless you’re visiting there and drank too much beer.
- One was missing a full bathroom, a bedroom, and 800 square feet. Better not invite too many guests!
- The third had the correct square footage, but two bedrooms and a half bath had disappeared. Maybe that’s David Copperfield’s home?
- The final one was missing 6 square feet of lot size (not a real problem since the lot in question was 21000 square feet), but the home itself was missing! Maybe David Copperfield and David Blaine lived there?
Zillow scores 90% for its data feed. Zillow had the right information for the 6 properties that the county had correct, and it had the same incorrect information on 3 properties that the county had. The final property (the David & David home) simply wasn’t there!
A challenge to my fellow bloggers out there: If you have access to your local MLS and want to run a similar experiment, email me at kboer@apr.com and I’ll send you a link to an iRows spreadsheet I’ve set up.
Here are my results:
Comments
Got something to say?





Subscribe
