Recent posts

June 3, 2008

Trouble in the Southland

Update as of June 6:  Sandicor has fixed its data feed, and as of a few minutes ago all San Diego MLS listings are up-to-date on Redfin.com. Should listing data issues in any market come up in the future, we’ll hold to the commitment outlined below. Thanks for your patience!

Some of you may have noticed that San Diego listings on Redfin.com haven’t been updated over the weekend. Sandicor, the San Diego area MLS, has been having some issues that mean it can’t export listing data to MLS-powered websites like ours.

We’ve talked to the folks over at Sandicor and they’re working hard to resolve the issues. They run a tight ship, so we’re confident the data will be moving again soon.

road-closed.png

San Diego listing data highway: temporarily closed.

It bothers us whenever the data on our site isn’t as fresh as possible. We’re a member-broker of the MLSs in each of our markets so that we can get the most accurate and up-to-date listings on Redfin.com. We update our site with new data from most of the MLS databases every 15 minutes.

But sometimes problems come up. You should know what you’re getting on Redfin.com, so here’s what you can expect:

We’ll update listings on Redfin.com every 24 hours, or sooner.
This applies to our MLS listings, as well as foreclosure and for-sale-by-owner (FSBO) listings. An update means the latest data from our listing data providers is on Redfin.com. For sources of data that change less often, like tax records and school information, you can expect the data to be updated less frequently.

When we can’t, we’ll tell you why.
Whenever our listing data is more than 24 hours old, we’ll post a note in the forums for each market that’s affected, explaining the issue and giving you some idea of when it will be resolved.

And then we’ll fix the problem ASAP.
If the problem is on our side, you can rest assured that we’ll attack it like rabid dogs until it’s fixed. On weekends, we may be slower to discover problems with our data, since our developers personally monitor the data feeds and they occasionally, but only occasionally, take part in non-development related activities - i.e. “personal life.” If it’s our partners who are having trouble, we’ll give them whatever help they need.

We simply want to build the best real estate website in the world, and as part of that we’re making this public commitment on our listing data. Comments are, as always, welcome.

Photocredit: sillygwailo on Flickr.


May 16, 2008

What Happened to Redfin’s Data Download Feature?

A few months back, we released a new version of the site that let our customers download listing stats and past sales data to a spreadsheet. Then last Tuesday night, we began to require registration on Redfin’s site before allowing customers to download results from the map. People typically use this data when they’re doing a Comparative Market Analysis or CMA.

We made the change because a few of the MLSs to which we belong have requested that don’t allow indiscriminate access to listing data. As one MLS has already pointed out, this will help protect home-sellers from being snail-mail spammed by movers. It’s a way to balance buyers’ hunger for information with sellers’ privacy.

And no, we’re not complaining about the MLS. As we observed this week in the New York Times, MLSs are getting more and more Internet-friendly all the time. We know how hard it is for them to set up rules that all the brokers can abide by when sharing their data, and we like belonging to the MLS. Being able to show all the homes for sale is much better for us than trying to build our own partial database, which is what the non-brokers have to do.

Here’s an example of the data you can download once you’re logged in:

Download homes for sale

Several Redfin users have noticed the change. One asked us the question:

I can’t seem to see/find the download button when pulling data for homes that have sold, to build comparisons.

What am I missing?

Thanks,

ZK

ZK, you’re not missing anything, but now you need to register and log in to see the data. If you register, we won’t spam you, sell or rent your email address, or otherwise do nefarious things. And of course you get to use all kinds of cool site features like email updates on your search, RSS feeds, and the ability to tell the difference between homes for which you’ve viewed details (lighter colors) and those for which you haven’t (darker).

Icons- listings, seen and not seen

We think registration for download is a good thing: serious buyers still get the data that helps them make a sound offer, as well as other benefits. And sellers don’t get spammed.


April 2, 2008

The Measure of a Block

What makes a neighborhood? The type of person who lives there? The architecture? The crime? How the flowers look in the spring?

Capitol Hill, Washington DC

Capitol Hill, Washington DC. Photo credit: Flickr.

Everyblock thinks all of those things make a neighborhood, as well as restaurant inspections, permits, news, and, now, real estate listings. Using our RSS feed of listings, they just added summaries of real estate listings to their data for every block in San Francisco. For example, here’s what’s happening today in Supervisorial District 6, in North Beach.

All this data and a growing number of neighborhood bloggers mean we no longer have to rely on print weeklies for hyper-local news. Here in Seattle, the bloggers cover everything from burning tires to park closures, to old ads for local beers. In DC, they blog about sidewalk posters, a new hemp store, the blooming of the cherry blossoms and police cameras.

Capitol Hill, Seattle

Capitol Hill, Seattle. Photo credit: Flickr.

If you’ve got an idea of what defines your neighborhood, leave a comment.


March 11, 2008

Typical is Boring, but When Selling a Home, Effective

A recent Redfin Forums contributor raised the question: Does a higher buyers’ agent commission = faster sale? From Baltimore to Seattle to Southern California, real estate agents and journalists have endorsed the idea of offering unusually large commissions so buyer’s agents will recommend a listing.

Because we’re curious masochists, we decided to answer that question. But first we had to establish what constitutes a “higher” commission. Higher than what? While traditionally the seller has offered the buyer’s agent 3% of the final home price and reserved 3% for his own agent, the very existence of a standard commission has been hotly disputed. Again, some notes from the fray: Realtors from Canada to Florida insist that a standard commission doesn’t exist, while professors from Berkeley and the Hudson Institute insist that it does.

Of course, Redfin’s business model demonstrates that it’s possible to charge different commissions. But Redfin has always been careful when listing a home to encourage our clients to offer the buyer’s agent 3%, even while accepting a lower fee for ourselves. In this at least, we’re like everybody else. We analyzed commissions paid to buyers’ agents for all broker-listed residential homes sold in King County in 2007, and found that 79% paid commissions of exactly 3% to buyers’ agents. Whether 79% adoption constitutes standard behavior or not, it is certainly common behavior.

The average commission, factoring in the lower and higher commissions as well, was 2.88%. The chart below (source: NWMLS) shows the average commission grouped by list price of the house. Most agents will tell you that the commission as a percent of the list price is lower for more expensive homes. We found this to be true: for list prices above $2 million, the average commission started getting closer to 2.5% than 3%.

Real Estate Commissions by List Price of Home

Because of the difference in commission rate for more expensive homes, we broke our data set into two parts: homes listed above $2 million and homes listed below $2 million. We focus here on the homes listed below $2 million because it’s a more interesting set– it’s much larger and the effects are a lot greater as well. This data set included 22,673 home sales.

Table: Home Sales, King County Residential Homes, 2007

  Commission lower than 3.0% Commisison equal to 3.0% Commission greater than 3.0%
Sale-to-List Price 99.9% 99.3% 98.5%
Days on Market 89 68 129

Rick West in our real estate group and Chris Wilkins in engineering ran the numbers, and the results were not as expected. That’s exactly why we created The Real Estate Scientist: to ferret out the data behind the myths and assumptions.

Homes offering a commission higher than 3% actually had a lower sale-to-list price than homes offering a 3% commission, by about 0.82%, or the equivalent of $4,095 on a $500,000 home. We had hypothesized that a higher commission would correlate to a better result for the seller. In contrast, homes offering a commission below 3% did get a better result for the seller: they had a higher sale-to-list price, by 0.54% or the equivalent of $2,719 on a $500,000 home.

The real difference between these groups was in days on market: homes that offered a 3% commission took 68 days to sell, while homes that offered a lower commission took about 30% longer, and homes that offered a higher commission took almost twice as long.

According to our data, setting a higher commission to get better results doesn’t work. It’s best to be typical, with respect to commissions.

How should we interpret this data? Is it the doing of the seller’s agent, who offers a lower commission on a property he knows will sell, and a higher commission on a property he thinks will be difficult to move? Or was the high commission itself a signal of desperation that encouraged negotiating? We tend to think that unusually high commissions are a symptom rather than the cause of a distressed listing. Whatever the case may be, the Seattle data suggest that offering buyer’s agents an unusually high commission isn’t worth it. If you have a different take on our results, just leave a comment.


February 20, 2008

For All You Data Junkies Out There

We’ve been getting local about our markets to see which towns should be hot spots for Redfin. We’re sharing the analysis for all you data junkies out there.

Here’s what we already know about Redfin buyers: they’re very likely to work in high-tech fields, they usually have a college degree, and they typically buy a house for about $500,000. Like these people:

Redfin Customers

We looked for similar demographic characteristics in targeted areas within each of our markets, and combined that with recent activity on Redfin’s site. The data we considered was:

  • Visits to Redfin.com, from Google Analytics. Note that Google tracks where people are when they search, not where they’re searching on the Redfin map.
  • Redfin closed deals. Useful because many Redfin buyers hear about us through word of mouth. We got this from our own database, which surprisingly is only about 50% complete.
  • Demographics of the area. We considered median household income, homeownership rate, and percent of population with bachelors’ degrees or higher, as reported by the US Census. If we couldn’t find city-level data, we used county data.

We used visits to rank order the towns, and then added the other data. The top 20 towns searching on Redfin in the Boston area look like this:

City Redfin.com Visits Redfin Transactions Homeownership Rate Median HH Income % With Bachelors’
Boston 16.6% 5.3% 32.2% $39,629 35.6%
Somerville 12.2% 5.3% 30.6% $46,315 40.6%
Allston 10.7% 0.0% 32.2% $39,629 35.6%
Brookline Village 5.9% 0.0% 69.7% $67,066 42.9%
Newton 2.1% 0.0% 69.5% $86,052 68.0%
Cambridge 2.0% 15.8% 32.3% $47,979 65.1%
Brighton 1.9% 0.0% 32.2% $39,629 35.6%
Milton 1.9% 0.0% 69.7% $67,066 42.9%
Jamaica Plain 1.7% 5.3% 75.6% $60,359 27.8%
Watertown 1.6% 5.3% 47.0% $59,764 47.2%
Newtonville 1.4% 0.0% 69.5% $86,052 68.0%
Natick 1.3% 0.0% 61.7% $62,854 43.6%
Medford 1.2% 0.0% 58.6% $52,476 31.7%
Charlestown 1.1% 0.0% 32.2% $39,629 35.6%
Arlington 1.1% 0.0% 61.7% $62,854 43.6%
Waltham 0.9% 5.3% 46.0% $54,010 38.4%
Woburn 0.9% 5.3% 61.7% $62,854 43.6%
Framingham 0.9% 0.0% 61.7% $62,854 43.6%
Malden 0.9% 10.5% 43.3% $45,654 26.2%
Concord 0.9% 5.3% 61.7% $62,854 43.6%

What’s interesting about the Boston data? Well, a lot of the neighborhoods where people are searching for homes- Somerville, Allston, Cambridge- have very low homeownership rates. But if you know Boston, you know that’s because these are where a lot of college students live. They’re mostly in apartments but maybe they’re looking for a place to buy after graduation. Or maybe they’re just blowing off class and surfing the Net.

Let’s go down the coast to Washington DC. This is a big area that covers several counties in Virginia and Maryland, as well as the cities of Washington DC and Baltimore.

City State Redfin.com Visits Redfin Transactions Homeownership Rate Median HH Income % With Bachelors’
Washington DC 20.9% 20.5% 40.8% $46,211 39.1%
Arlington VA 14.5% 9.1% 43.3% $66,626 60.2%
Baltimore MD 5.1% 4.5% 50.3% $29,792 19.1%
Falls Church VA 3.4% 0.0% 60.6% $82,906 63.7%
Takoma Park MD 2.9% 0.0% 68.7% $76,957 54.6%
Reston VA 2.7% 2.3% 70.9% $83,890 54.8%
Alexandria VA 2.7% 6.8% 40.0% $60,715 54.3%
Dunn Loring VA 2.2% 0.0% 70.9% $83,890 54.8%
Chevy Chase MD 2.0% 2.3% 68.7% $76,957 54.6%
Merrifield VA 1.9% 0.0% 70.9% $83,890 54.8%
Rockville MD 1.8% 2.3% 68.7% $76,957 54.6%
Herndon VA 1.7% 0.0% 70.9% $83,890 54.8%
Oakton VA 1.7% 0.0% 70.9% $83,890 54.8%
West Mclean VA 1.6% 0.0% 70.9% $83,890 54.8%
Bethesda MD 1.5% 0.0% 68.7% $76,957 54.6%
Sterling VA 1.4% 0.0% NA $NA NA
Fairfax VA 1.3% 0.0% 70.9% $83,890 54.8%
Annandale VA 1.3% 2.3% 70.9% $83,890 54.8%
Garrett Park MD 1.3% 0.0% 68.7% $76,957 54.6%
Vienna VA 1.2% 2.3% 70.9% $83,890 54.8%

Washington DC and Arlington ranked near the top, which wasn’t surprising to us. What was surprising was Baltimore: not often regarded as a tech center, and with a much lower median household income than Washington DC, the citizens of Baltimore showed themselves to be quite tech-savvy by searching Redfin for homes to buy (full disclosure: I’m from Baltimore).

Flying across the country to our home market of Seattle, we found some odd patterns. Note that in this data set we didn’t include the three demographic variables from census data, but we did include more of the Google Analytics statistics about the behavior of visitors.

City Redfin.com Visits Redfin Transactions Bounce Rate
Seattle 46.1% 43.2% 2.8%
Bellevue 15.9% 24.6% 1.9%
Kirkland 9.3% 5.3% 1.8%
Medina 4.3% 0.1% 2.2%
Mercer Island 2.9% 1.0% 2.4%
Lynnwood 2.7% 3.2% 2.0%
Seahurst 2.6% 0.0% 2.8%
Redmond 2.0% 2.5% 2.4%
Renton 2.0% 3.3% 1.8%
Bothell 1.5% 1.1% 2.1%
Tacoma 1.2% 1.1% 3.4%
Federal Way 1.1% 1.2% 2.3%
Mountlake Terrace 0.9% 0.1% 2.4%
Kent 0.8% 1.6% 2.4%
Mukilteo 0.8% 0.0% 1.9%
Kenmore 0.7% 0.2% 2.0%
Bremerton 0.6% 0.0% 5.3%
Puyallup 0.5% 0.6% 3.8%
Spokane 0.4% 0.0% 30.5%
Auburn 0.3% 0.6% 1.6%

The best brain candy in the Seattle area data-set, in my opinion, is Medina. This city is right on the coast of Lake Washington and has some pricey real estate. It’s a small area, and so it’s not surprising that Redfin has closed few deals there. What is surprising is that Medina is the fourth largest searcher in Seattle. Could it be that Medina’s most famous resident, Bill Gates, is addicted to searching real estate on Redfin?

Also interesting is the data for Spokane, where Redfin has a lot of searchers but a high bounce rate. Bounce rate is when a searcher finds a site and immediately leaves it. The searchers of Spokane probably realized that Redfin doesn’t have service in Spokane, and left. This makes me think that Redfin should add coverage in Spokane.

Data on our California markets isn’t included here. In San Francisco, the exercise was less illuminating because San Francisco and San Jose by far dominated our search results and closed deals. We can get more granular data, down to the neighborhood level, but we didn’t for this exercise. Ditto for Southern California.


October 11, 2007

DoJ: Competition and Innovation are Good, Even in Real Estate

Yesterday, the Department of Justice published a website that focuses on the importance to consumers of competition in real estate. There’s a good summary on the InmanNews blog. They found that that commissions have risen along with home prices, which is no surprise to anyone who’s been in the market. But the data are interesting, and the DoJ also adds specific information on regulations by state.

The DoJ estimates that the median commission paid by consumers so far in 2007 is $11,302. This includes both buyers’ and sellers’ commissions. (The buyers’ agent is usually paid half of the sellers’ agent’s commission, so while buyers’ agents may charge no direct fee, they are still receiving up to 3% of the total costs paid by a buyer at closing.)

As a point of comparison, we looked at commissions paid by Redfin customers who bought in 2007. We focused on buyers’ commissions only, where we have the most data. Redfin’s average commission was $15,071, and because we refund 2/3 of our commission back to the buyer, $10,194 of that went back to consumers. By market, Redfin’s refund to buyers breaks down as follows:

Redfin Buyer Refund, 2007 Year to Date

This tells us a couple of things:

  1. Redfin buyers are buying expensive homes. Assuming a total commission of 6%, which is typical in our markets, our buyers paid a little over $500,000 for their homes. That’s much higher than the median home price of $218,184, but Redfin is also in some of the most expensive housing markets in the country.
  2. Of course, where house prices are high, consumers pay much more than the median for commissions. Again assuming a 6% commission, total commissions (both buyers’ and sellers’) would have been over $30,000 on average for these homes, much higher than the $11,302 that concerned the DoJ.

Redfin buyers are successfully searching our MLS listings to find properties on their own, then working with our local agents to negotiate and close the deal. They’re buying expensive houses and saving a lot of money. And the DoJ would seem to think that’s a good thing.


October 2, 2007

Making Real Estate Search Faster and More Local

Some things are universal, such as when you click on a link you want an instant reaction. Redfin’s response time hasn’t been as fast as we’d like, so we’ve released a new version that improves performance significantly. It’s on the site today. You’ll see much faster search, especially for past sales search. John Battelle’s blog on what web users value was influential in our decision to prioritize performance.

We ran many different tests against our old user interface and the new one. The chart below shows the performance gains across three browsers. In the case of Internet Explorer 6, we display listings on our map almost four times faster with the new version of our site. The performance gains are smaller but still significant for IE7 (download it) and Firefox (download it), both of which have better overall performance than IE6.

perf-chart.jpg

And we’ve added a cool visual effect: during the (now shorter) search, you’ll see a progress bar and houses will progressively pop up on the map. For the techies among you, we’ll be talking in more depth about the specific things we did to improve performance, including a good bit of data and commentary on our database switch, in an upcoming post on the Redfin Developers’ Blog.

Talking about speed of information, we’ve also sped up the MLS integration (MLS systems are used by brokers to track all the open listings in an area). In many cases we load new listings faster than the real estate sites used by the brokers, so our customers are the first to see newly listed properties. In Southern California, we’ve had issues with MLS data getting onto our site too slowly and we think we’ve fixed those latency issues in San Diego, Orange County, and most parts of Los Angeles. Our goal is to have all new listings in Southern California on Redfin within minutes, which is the type of performance we have in other markets.

In this release, we’ve also added lot outlines (see below) in Boston, Baltimore, and DC and updated the outlines in Seattle and California. Finally, we’ve added new map icons and links to driving directions on the property detail pages.

lot-outlines-and-map-icon.jpg

Other things aren’t as universal. When I moved away from San Francisco, I wondered why it was so difficult to get a burrito the way I wanted it. In San Francisco, you tell them what to put in your burrito; everywhere else, you start with a standard burrito and add and remove ingredients. Invariably something goes wrong.

Food is local, and so is real estate. While we’re only in a few markets now, we try to go deep in those markets. We write about local real estate in Redfin’s Sweet Digs blogs, provide an online customer forum in each market, and of course hire experienced, local agents. As part of that effort, we’re rolling out local sites for San Francisco Bay Area real estate, Washington DC and Baltimore real estate, Seattle real estate, Boston real estate, San Diego real estate, Los Angeles real estate, and Orange County real estate.

People come to Redfin for direct access to MLS listings, so we’ve left the search bar front and center on these pages. But we’ve added information that’s useful for someone searching for a new home, such as schools, census data, and local news and blogs. For example, interested in new construction in the Westchester area of LA? You can find out about it in Sweet Digs on the Redfin LA page.

We like these pages, but we also realize they’re just a start. If you’ve got any suggestions on good local links to include or other feedback on the pages, please send them. And let us know whether you like the lot outlines and new map icons, and if there’s anything else you’d like to see on the map.