Yahoo! or Gotcha!?

Yahoo! or Gotcha!?

I thought they'd cleaned up this mess years ago. Like maybe back in 1998.

Back when listings first appeared on the Internet—hurrah!—and removed real estate agents as the gate-keepers of listing information—hurrah!—trouble in virtual Paradise immediately broke out in the form of what might be called "padded listings". Listing Web sites made claims like "3,000,000 listings!" Fantastic—data! give me more data!—except that many of those listings were stale. Still showing as active, they'd really sold months ago.

Let me interject a tiresome moral here, something to the effect that the wonderfulness of the Internet is directly related to the wonderfulness of the organizations that operate on the Internet, which wonderfulness is directly related to the professionalism of those organizations, which professionalism is directly related to their willingness to either do or reject the expedient, which willingness is directly related to oh-so-fallible human nature faced with the choice between drop-dead reliability and a quick buck—so that the vaunted scientific wonderfulness of the Internet turns out to rest squarely on the spindly underdeveloped shoulders of human nature.

This changes everything!

Back to online listings. Years went by, and I operated under the impression that the problem of stale Internet listings had been mopped up. No longer did clients contact me excitedly about listings that had sold six months or a year ago.

Until a few months ago.

The first client is looking at $2M properties, a market that, at least in Silicon Valley, has far more buyers than sellers. While other markets, locally and nationally, wallow in unwanted inventory, the "move-up" market here has been crippled, no, not by panicked sellers and timid buyers, but by a lack of inventory. Foreclosures? Not a problem. Lay-offs? Not a problem. Tight credit? Not enough of a problem, at least not like further down the price range. Seasoned homeowners, not afraid of their own shadow, ready to pull the trigger on a bigger house in a better neighborhood? Yes! One-hundred-seventy-five people through an open house? Yes yes yes!

Just to complicate things, this $2M move-up client of mine currently lives in another area, one that's a national poster child for the housing slump, with plunging prices and catastrophic foreclosure activity. Her friends slap her on the back, congratulating her on the perspicacity to buy now, in a "buyer's market", not knowing that she's actually buying in an extreme seller's market. And just to complicate things even more, she's relocating and really needs to buy a home, so it's frustrating for her to hear me say, "Sorry, there's nothing more I can show you. That's all there is."

So on February 13, 2008 this client seizes the initiative, goes to Yahoo! Real Estate and searches for homes herself, using the zip code of the area in which she wants to live. Eureka! She finds ten—count 'em, ten!—homes that look perfect. Suddenly the market is flooded with promising homes. Homes I hadn't told her about. Excited, she emails them to me.

I panic. How did I miss all these homes? I check the Multiple Listing Service every day.

But you know what? Some of these homes look mighty familiar. In fact, weren't they on the market a while back? Yes, I'm sure they were, and I'm equally sure they sold.

So I go to the MLS, and, sure enough, every single one of these "94402 homes for sale" is no longer for sale.

Remember, she did this search February 13, 2008. Here are the sale dates of those ten listings supposedly active on that date:

February 15, 2007 (only two days short of its first anniversary!)
March 19, 2007
April 19, 2007
April 24, 2007
May 7, 2007
July 14, 2007
September 28, 2007
October 7, 2007
November 28, 2007
December 6, 2007

Puzzled, the next day, February 14, I go to Yahoo! Real Estate and search, not by zip code, but by city. I search, not the city where my client is looking, but another, smaller city—I know, not scientific, but I don't have time to work with a huge data set. Yahoo! shows ninety-six homes, both single-family and condos, for sale in that city. The MLS shows just ninety-two. Not an exact match, but close enough for Internet work.

However, I notice that Yahoo! says their "snapshot" of the market was updated February 11, 2008, three days before. Lots of things happen in an active real estate market in three days, and this particular city is still an active market.

So to reconcile the Yahoo! and MLS searches, I make these adjustments to the MLS single-family inventory:

four sold between February 12 and 14
one came on the market
and two were taken off the market unsold

So the correct number of active single-family listings as of the update date, February 11, was seventy-five.

Next I make the same adjustments to that city's condo inventory:

one sold between February 12 and 14
none came on the market
and none were taken off the market unsold

Which means that the correct number of active condo listings as of February 11 was twenty-three.

Add twenty-three to seventy-five and that's ninety-eight MLS listings as of February 11, compared to ninety-six Yahoo! listings. That's an acceptable variance, given the margin for error inherent in the adjustments.

So those ten "active listings" my client had found on Yahoo! that'd sold months before must all be flukes. The apparent problem was that she'd searched by zip code rather than city. Was it just the particular zip code she'd used, or would all zip code searches have yielded incorrect results? I don't know, and I didn't feel like verifying the accuracy of every Yahoo! zip code search. As far as I was concerned, the case was closed.

But the case didn't want to be closed. Because the next day another client asked me about three listings he was eyeing. This client is also a buyer, but he's looking in a different area and lower price range than the first client, and he doesn't feel any pressure to buy. As he says, the pundits predict that real estate will be very soft for a very long time. To back up this theory, he mentions three listings on Zillow that have been on the market for a very long time. I look them up on the MLS and find that:

the first house sold in twelve days back in December 2007
the second house also sold quickly, back in August 2007
the third house is still on the market

Remember, this was February 15, 2008. This client's Yahoo! Real Estate search turned out to have a less-than-sparkling 33 percent accuracy rate. Now, I ask you: what if two of every three homes I pulled up in front of, with excited clients in the back seat, turned out to be sold? Months ago? How much credibility do you think I'd have with those clients? Former clients.

A month or so goes by and again I forget about Yahoo! Then a client who's looking for a very specific kind of property, one that'll be hard to find, asks me excitedly about a home she's found on the Internet that's exactly what she's looking for. Again I panic because, again, I hadn't seen it on the MLS. I look up the property and discover that it was taken off the market, unsold, in April 2007. Yet it still shows as an active listing, one year later, on one listing Web site. I ask which site it is. "Three guesses, and the first two don't count", as we used to say when we were kids.

What's the problem? Is it the listing Web sites themselves, who in the interests of big numbers and low costs aren't diligently removing listings as they sell? Or is it the organizations that provide listing information to these sites, organizations that, as far as I know, are MLSs around the country? If the fault lies with the MLSs, then why is the information they provide their agent subscribers (for several hundred dollars a year) invariably correct, at least in the case of my local MLS, yet the information they provide to Web sites not?

Here's another question: what does this say about the efficacy of being an "Internet-powered, self-directed real estate consumer"?

There's another problem, one that goes beyond merely sending self-directed buyers off on wild goose chases, that I wonder if anyone is even aware of. How many bubble bloggers, market timers and other amateur market watchers are getting their information about unsold inventory from these Web sites? How many of them are avidly tracking inventory numbers, a vital market indicator, that may be grossly inflated? How many market timers are trying to time the market with bad numbers?

And finally, what does this say about the Internet's role in real estate? Data, the Internet's only specialty, has never been helpful information by itself. Bad data, I feel safe to say, is even less helpful.