TechCrunch Asks the $24,000 Question

A strange thing happened in online real estate last Friday. After three years of fierce competition and $100+ million in venture capital, TechCrunch’s Erick Schonfeld finally asked real estate websites the $24,000 question: who has the most homes for sale?

“The most important success factor for these sites,” Erick writes, “is how comprehensive they are.” By comprehensive Erick means within a market: nobody wants to see every listing in the U.S., just every listing wherever he or she wants to live.

So Erick published a study from Roost, a lead-generation site with a feed from the MLS — the database that, in nearly every market, all brokers use to share listings. (Redfin has MLS listings like Roost, but also for-sale-by-owner homes and bank-owned foreclosures not yet in the MLS.)Photo credit Roberdan

The study showed that on the strength of its MLS access, Roost has far more broker-listed properties than sites like Trulia, Yahoo and Google — which have to get their listings through one-by-one agreements with brokers.

Redfin wasn’t included in the study, and can’t even get a copy to assess its accuracy. But it was sad for us to see everything but the kitchen sink thrown at the study, from people who never denied its basic conclusion: that a site with MLS access has many more homes for sale in a market than one that doesn’t.

And while the debate in TechCrunch’s comments section raged over the quantity of listings, no one acknowledged the huge differences in quality: what brokers share with media sites is only a teaser photo with a few basic details, rather than the complete MLS listing shown by Redfin and Roost, with dozens of fields and photos, updated as transactions occur. The difference in quality is plain for anyone to see: compare Redfin’s Washington DC listing to what appears on media sites.

But since everyone focused on Swiftboating the quantity claim, let’s take a look at the arguments:

  • The study was arbitrary. That’s true. The parameters were arbitrary: 3-beds, 2-bath homes between $400,000 – $450,000, in Miami, Dallas & San Diego. But to argue a study is skewed, don’t you have to say how it was skewed? Which parameters would have shown that Trulia, Yahoo or Google has more listings?
  • Actually, “we have roughly 70% coverage in most major metros.” Trulia argued that its coverage was better than reported, at 70%. Setting aside that missing 1 in 4 listings is itself a calamity, we wondered 70% of what? Of MLS listings? The claim seems as arbitrary as the study Trulia was trying to rebut.
  • Consumers care most about “filtering options,” not seeing all the homes for sale. Redfin has a dozen advanced filters but 94% of our searches are on the basics: price, beds, baths. One of the most common filters is no filter at all.
  • Only 80% of the properties in the U.S. are sold via the MLS.  Academic studies have found that even in Madison, Wisconsin, reputedly the nation’s capital of for-sale-by-owner homes, 86% of the properties were sold by a broker via the MLS. In other areas with an MLS, that percentage is much higher. Where did 80% come from?
  • The MLS doesn’t include foreclosed homes, which comprise 31% of home sales in San Diego. But in fact foreclosures are likely already in the MLS, because banks hire brokers to put them there. Redfin obtains listings from the banks and directly from the MLS, so we know how the numbers break out. Here’s what we have as of this past weekend:

    The suggestion that 31% of listings are not in the MLS because of foreclosures is off by an order of magnitude.

  • The study was “silly.” It isn’t silly to worry about whether you’re seeing all the homes for sale. And trying to make a joke out of the whole issue by saying “I was thinking about commissioning a study saying that I’m a size 2 supermodel” implies that Roost is trying to prove something unimportant and obviously untrue, when in fact it is trying to establish, however artlessly, something important and almost undeniably true.

But here’s what’s great about TechCrunch. Erick didn’t just print the he-said she-said arguments between websites. He queried all the websites himself, and found that Redfin had slightly more listings (6,300) than Roost (6,036), and Roost had far more listings than Trulia (4,395).

Zillow, which had the lowest number of MLS listings in the Roost study, claimed the most listings of all, 7,661, but Erick couldn’t confirm this because Zillow’s site searches on the county of San Diego, not the city. Redfin gets listings from Zillow as well as the MLS, so any listings on Zillow that aren’t on Redfin should appear shortly.

Hats off to Erick for having done the research.

He shouldn’t have had to. The pledge of every new real estate site has been to bring transparency to real estate. But so far, most sites have been anything but transparent about what they have and what they don’t. Now Joseph Ferrara at Sellsius blog is asking us all to stop hiding the ball and say where our listings come from and what we’re missing. Since April 2008, Redfin has done just that. Trulia, Yahoo, Google, please step forward.

Photo credit: Roberdan on Flickr.

Discussion

  • http://www.roost.com Drew Izzo

    Nicely put Glenn! And one point of clarification. http://www.roost.com does have for sale by owner listings.
    b/w any area that we cover, we cover as you do – completely.
    To Joe’s point:
    Because each local MLS does not have a geographic border, we occasionally have listings in an area that we do not cover completely. You’ll notice that we’re honest here with this message: “ listings are under construction. They’re all accurate, but not complete like other areas.”

  • http://http//blog.Franklyrealty.com Frank Borges LL0SA= Broker FranklyRealty.com

    Hey Glenn,

    I’m sure you will agree that all of this is WILDLY confusing for consumers.

    Recently somebody told me that they “prefer to search on Zillow.” I asked, “Why, because you like to see 65% of the homes for sale?” Without the MLS, they have “a bunch” of homes in one area.

    Sure they can talk about Millions, but tell me how many are in Arlington Virginia!

    I’m working on FSBOs for my MLS site FranklyMLS.com (currently it launches a link to Craigslist to find the FSBOs) and of course foreclosure data.

    But the foreclosure sites also have a MASSIVE numbers problem too. My site will pretty much disclose that this data sucks, but they want it, so we will give it to them.

    Homes listed on these sites are 6 months old and sometimes sold twice already. Why keep them in? Because then you can say you have 1.2M vs a competitor’s 1.1M homes.

    I spoke to an Investor/Realtor (on a panel) that says that these foreclosure sites suck, but he makes a ton of money off of them. The data is so bad, that he bought a zip code and advertises his services. People come to him… and he sells them… drumroll, regular REO homes listed on the MLS.

    He has never seen one person, or investor use that site to buy anything.

    For the most part if you have the MLS and Craigslist results, you have 95-98% of the homes for sale in one area. Maybe add ForSaleBy0wner.com for the last couple points (but don’t they all list on Craigslist too?)

    If only Craigslist would allow us to feed their data.

    Frank
    Broker FranklyRealty.com, FranklyMLS.com

    p.s. How about a study on who has the MOST photos? We are about to cross the 30,000 additional photos added to our WIKI MLS.

  • Pingback: Why Are We Wasting Our Time? | BloodhoundBlog: National real estate marketing and technology blog | Realtors and real estate, mortgages, lending, investments

  • Pingback: Consumers Deserve Real Estate Info They Can Trust | A Bird's Eye View

  • Andrew Mattie

    I’m not sure that “skewed” is an appropriate term for the study. It’s just not statistically significant or representative of anything outside of the defined variables in any way. I’m not knocking it — I’m only claiming that readers of the study shouldn’t attempt to surmise the quantity or completeness of the listings within any single database except for within the narrowly tailored parameters listed.

  • http://www.cyberhomes.com Brett

    Brett Shaw from Cyberhomes:

    What really bothers me is that as soon as one site advertised that they had over 3 million listings, so did everyone else. All of a sudden, I saw up to three listings on the same home on zillow. Not only that, but there are different prices as well. Which one is it? All this does is confuse potential buyers and make them question all of the other data on the site. Do not confuse quantity with completeness.

  • http://www.oakland-ca-homes-for-sale.com/blog/ Oakland Cowboy-Oakland, CA Homes for Sale

    These are interesting statistics. I agree with the rest of you that you can’t put 100% faith in any of the studies. I guess at best they are good for showing trends.

  • Pingback: Quora