To quote a study conducted by Redfin:
When home buyers search for homes for sale online, are they seeing all the homes for sale? Do new listings show up in a timely manner? Are all the homes they are seeing actually still for sale?
To answer these important questions, we analyzed a sample of 6,401 home listings in 33 zip codes from 11 metro areas. The listings analyzed were from three local brokerage sites and two highly trafficked national real estate portals.
Local brokerage sites are considerably more complete, more accurate, and timelier.
The results of this analysis are that the local brokerage sites are considerably more complete, more accurate, and timelier than the national portals.
In each U.S. city or area, all real estate brokerages contribute and share listing data. Only real estate brokers can be members of a local MLS. In contrast, national portals, such as Zillow and Trulia, mostly rely on individual agents and real estate brokerages to re-post their listings on the portal websites, or the portals aggregate data from syndicators of real estate information.
Each of the local real estate brokerage websites analyzed contained 100% of the homes listed for sale in the MLS. Portal sites contained just 79% to 81% of these listings – a fifth of the homes for sale did not appear on portal sites.
Percent of Agent-Listed Homes Available to Consumers
Figure 1 – Listing Completeness
The results varied by market, with one or both of the portal sites showing at least 85% of the listings in five markets: Boston (Trulia at 91%), Denver (Zillow at 86%), Philadelphia (Trulia at 93%, Zillow at 86%), San Diego (Trulia at 94%, Zillow at 88%), and Washington DC (Trulia at 86%).
Trulia was missing the most listings in Seattle, where only 63% of the homes for sale appeared on the site. Zillow was missing the most listings in Phoenix, where only 65% of the homes for sale appeared on the site.
|Market||Redfin||Long & Foster||Windermere||Trulia||Zillow|
|Los Angeles, CA||100%||n/a||n/a||81%||73%|
|San Diego, CA||100%||n/a||n/a||94%||88%|
|San Francisco, CA||100%||n/a||n/a||74%||78%|
Figure 2 – Listing Completeness by Market
(n/a = the respective brokerage does not serve this area or it was not included in the study)
On portal sites, over a third of the results are not really for sale.
Local brokerage sites get home listings from direct feeds provided by the local MLS. In most cases these feeds are updated every 15 to 30 minutes, which allows homes to appear on local brokerage sites as soon as they are listed by an agent. Portal sites obtain their listings from a variety of sources: individual brokerage feeds, third-party listing aggregators, and direct entry by agents. Occasionally portal sites gain access to MLS data, but even when this happens, that access is only partial. The median delay between when a home was listed on the MLS and when it appeared on the portal sites was nine days for Trulia and seven days for Zillow.
Median Days Between List Date and Appearing For Sale on Site
Figure 3 – Listing Delay
Since local brokerages update the data on their sites with new information from the MLS as often as every 15 minutes, when a home buyer searches a local brokerage site, there are little to no results that are outdated and no longer for sale. On portal sites, over a third of the results are not really for sale.
Percent of Homes Shown as For Sale That Are Not For Sale
Figure 4 – False Listing Results
MLS-powered local brokerage sites enjoy three major advantages over portal sites:
- Brokerage sites show 100% of the agent-listed homes for sale, compared to only 80% on portal sites.
- Brokerage sites show new homes for sale seven to nine days earlier than portal sites.
- 36% of home listings that appear as active on portal sites are no longer for sale.
In order to compare the accuracy of real estate listings across various types of search sites, we first searched the MLS directly in each market, and then performed matching searches on Redfin, Windermere (in Portland and Seattle), Long & Foster (in DC and Philadelphia), Trulia, and Zillow. All searches took place August 27 through August 31, 2012. All searches in a given market were performed on all sites on the same day.
For each zip code in the study, we searched each website for all actively listed, for-sale homes of the following types: single-family, townhouse, condo, apartment, TIC, coop, and loft. All “for sale” and “active” types on each site were searched. Pending sales were not searched. Zillow’s “make me move” homes were not searched. For sale by owner (FSBO) listings (available on Zillow and Redfin) were searched, but excluded from the analysis.
All searches were performed without creating or signing into an account on the various consumer sites. For each result on each site, the following data was collected: Status, List Price, List Date, Data Source, URL (link to listing on site).
The following errors or were not counted against any site in the final analysis:
- Unlisted, not-for-sale foreclosures returned in a search of homes for sale (as long as they were labeled as a foreclosure or pre-foreclosure).
- Pending homes that are no longer actively for sale returned in a search of homes for sale (as long as they were labeled as pending).
- Duplicate results in which the same home appeared more than once in the search results on a given site. The most accurate listing was used for the analysis; any additional listings were disregarded.
- Address errors, where a portion of a home’s address was entered incorrectly.
- Location errors, where a home’s location on the map was placed incorrectly.
- Property type errors, where a home was incorrectly categorized (e.g. a condo listed as a multi-family home).
Each of the 6,401 records analyzed is available in a spreadsheet for anyone interested in looking at the data in more detail.
Each of the above steps was executed in the following randomly selected postal codes in the following eleven metro areas. The total number of active listings for each market was between 150 and 200, for a total of 1,718 active for-sale listings (the 6,401 total listings analyzed included many homes that were not actually active for-sale listings).
- Seattle, WA – 98122, 98117, 98011
- Portland, OR – 97030, 98684, 97232, 97205
- San Francisco, CA – 94114, 94022, 94546, 95014
- San Diego, CA – 92114, 91910, 92106
- Phoenix, AZ – 85042, 85015, 85204
- Denver, CO – 80226, 80246, 80212
- Chicago, IL – 60304, 60194
- Los Angeles, CA – 90404, 90028, 92501, 92840
- Washington DC – 20815, 20037
- Boston, MA – 02155, 02445
- Philadelphia, PA – 19066, 19004, 19012