We have had a lot of feedback from readers about our first article about the Zillow.com website which was published here during the week of February 12. Nearly all of the writers were extremely negative about Zillow. For example:

"This website is highly inaccurate. The values of the homes are not even close. A house valued at 380K will come back on Zillow as 212K. They have a serious database problem."



"It's a very BASIC GUESS. It cannot 'see' the house (if the house is) damaged by a storm or fire or anything else. It cannot 'see' (if) the property is waterfront or has a spectacular view - or not - and compares (it) to only what's close in proximity - or the property that is gutted on the inside. Hey, it is free so everyone can use it at their own risk. VERY misleading in many areas and only decent for 'cookie cutter' neighborhoods in disclosure states. You can trust it about as much as a pie in the sky."

"Fun with Zillow. Change the year your house was built. For some reason if your house was built in 2000 instead of say, 1956, it is worth a couple of hundred K less. At least mine was. Add a bedroom. 25k less. Two houses. Next door. Same sq footage, yard and house, 200k difference. Suffice to say, needs work. And in the short term will cost some poor sod who has to sell his house and has a bad zestimate a lot of money."

The bottom line seemed to be that readers felt we were much to kind to this new and highly hyped website.

Well you didn't give us a chance to finish.

We tried to give Zillow a lot of leeway as it is a Beta site and is obviously evolving by the day. But there are definitely problems, the most obvious of which is that there is limited data available for much of the country or Zillow utilizes information which seems to bear a poor relationship to actual market value. At a later date, after giving Zillow a little more time to get its act together, we will look at metropolitan areas where Zillow itself rates its performance well and compare its "Zestimates" and home descriptions to listing prices and other information available on real estate websites and newspaper open house ads.

But, even as a Beta site and given that we didn't have a lot of contacts in areas where Zillow gave itself multiple stars for data accuracy and availability, we did find some strange bits of information and data that was certainly not static.

For example, Salt Lake City, Utah showed some very unusual numbers when first checked two weeks ago. A 13 room house in one of the city's better neighborhood was sold by a relative in 1994 for around $320,000. She then bought a luxury condo in a high rise for $260,000. Salt Lake is not Las Vegas. Home values have accelerated lately, but have been flat for many years. When we first checked out the house and the condo last week they had a Zestimate of well over $1.5 million and $1.2 million respectively. Unfortunately we did not print out the information but called someone else who knew the area and who confirmed what we had seen. This person also checked on a nice three bedroom slab ranch she had once owned slightly south of the city and was speechless to see it Zestimated at over $3.7 million.

A week later we rechecked all three of these properties and found that - whoops - there was no Zestimate on any of them and that the assessed values of these homes were a much more realistic at $280,000, $320,000 and $297,000. When we checked again today, all but the slab ranch had completely disappeared from the database.

The most innacurate information on Zillow came from a close suburb of Washington, DC in northern Virginia. There a friend reported that her house and the house directly across the street were both Zestimated at $471,000. My friend's house is a charming four bedroom, 2 bath early 20th Century craftsman style home in pristine condition. The kitchen was remodeled and a family room added some 20+ years ago and both kitchen and family room have all of the bells and whistles. The neighbor's home was gutted and nearly doubled in size at least four years ago. My friend's home is assessed at $692,000 and recent neighborhood sales ranged from $750,000 to $950,000. The house across the street has been valued by realtors at $1.2 million. The Zillow Zestimates bears no relationship to any of the real numbers she provided yet Zillow gives itself four stars for its accuracy in Northern Virginia, states that its data quantity - i.e. the availability of data elements - at "most" and ranks its Zestimates as coming within 10 percent of the actual selling price 70 percent of the time. This was substantially higher than the three stars, "most" and 66% ratings it gives itself in the Boston/New Hampshire/ Connecticut area which we found to be much more accurate.

It is a shame that Zillow came on line before it was ready for prime time. The hype has been incredible - ABC's Good Morning America featured it this week and, as we stated at the beginning, CNN, CNET, Motley Fool, USA Today, Business Week, and dozens of metro newspapers have given it a lot of free air or ink. Many people have visited the site, came away disappointed and may not bother to go back. Better that Zillow had all of their ducks in a row before going public.

There is another website that is threatening to "revolutionize" real estate. While Redfin is geographically limited (to the Seattle area) it offers a new way of eliminating real estate agents from your life. This time the target is the buyer, the method is a sort-of auction, and we will take a look at it this week.