Hitwise Intelligence - Heather Hopkins - US
« Content Aggregation is King? | Main | Twitter Gaining Momentum But Still Niche »
April 25, 2008
Internet Searches Match Decline in Existing Home Sales - Revised
Legend has it that Bill Tancer (my boss) bought and sold his last house using Hitwise search data. After reading yesterday's Wall Street Journal article about the continued decline in home sales and the stabilizing of housing prices, I checked Hitwise data to see what our data says about demand for homes for sale.
The following chart compares the share of US Internet searches for "homes for sale" (the highest volume non-branded search term relating to real estate) compared with the existing home sales (based on an seasonally adjusted rate published by the National Association of Realtors. (Note the search data is based on a rolling four week average of the share of US Internet searches for the final week of each month).
I ran a correlation analysis and found a very strong (.92) correlation between existing home sales and the monthly share of US Internet searches. This is about as close to a perfect correlation as I've seen and it got me pretty excited. (Yes, I am a geek.) The correlation is unlikely to be as tight over a longer stretch and I would like to extend the analysis back 24 or 36 months. (If you know of a source for the rate of home sales going back further than 12 months, let me know.)
Even if this is over only one year, it is interesting to see how closely online demand (as measured by searches for "homes for sale") and actual sales correlate in the past year.
This tells us that internet searches for "homes for sale" match the decrease in offline demand for homes for sale . We've done some work in the past looking at how internet usage data can be used as an economic indicator. This is a one small piece of that pie.
So, how do things look for April? I looked at the weekly share of searches for "homes for sale" for March and so far in April. Things are flat. The downward slide in demand may have ended but we are unlikely to see a recovery just yet.
NOTE: I made an error when I originally posted this entry. I copied and pasted the wrong column and so confused rate of sale with price. Thank you to Matt at Inman for gently pointing out my error. My bad... I did not find the same strong correlation between housing prices and internet searches. The correlation was .68 between searches for "homes for sale" and the median housing price and .69 for average housing prices.
Posted by Heather Hopkins at 08:57 AM
Posted to Search
Trackback Pings
TrackBack URL for this entry:
http://weblogs.hitwise.com/movabletype/mt-tb.cgi/789.
Comments
This yearly downward trend happens every year, even when demand is high and prices are rising.
In our observations of consumer real estate search trends, the downward trend over the course of the year can be more closely attributed to seasonal differences in home searching behavior. Here at Homes & Land (HomesAndLand.com) we see the same trend occur year after year: slow downward trend in visits towards the end of the year, a monster jump in traffic in January, a gradual upward trend into the middle of the summer, and a more dramatic, gradual decline into December. Wash, rinse, repeat.
Google search trends for the terms "homes for sale" and "houses for sale" over the past years _very_ closely match our overal internal traffic numbers:
"homes for sale", "houses for sale"
-Eric
Eric Adair
Business Development
http://www.homesandland.com
Posted by: Eric Adair at April 24, 2008 01:18 PM
Hi Heather,
We can help with the median price data you are looking for.
Drop me an email we'll help you on this or pr@trulia.com
Interestingly, we're seeing significant traffic growth over at Trulia.com. Over 3x yr on yr. Of course there are many more factors at play for us as a growing business and product. Without doubt, consumers are more in need of the independent data, advice and opinions that Trulia.com provides than ever before.
Speak soon
Pete
Posted by: Pete Flint from Trulia.com at April 24, 2008 01:59 PM
One question: is it "houses for sale" or "homes for sale" and does it matter?
BTW- pretty cool see what usage data might be telling us.
Posted by: Maria T. Pruitt at April 24, 2008 04:27 PM
Maria, Thanks for pointing out my error - it is "homes for sale". I ran the analysis with both terms and the results are pretty much the same, however there's a much larger volume of searches for "homes for sale".
I have corrected my post. Thanks again.
Heather
Posted by: Heather Hopkins at April 25, 2008 06:12 AM
Eric, excellent point - there is a typical seasonal decline that happens after a seasonal lift in January.
I didn't include the data further back but it's likely worth sharing. When we look back one, two and three years, we see a steady decline in the share of US Internet searches for "homes for sale". There is a seasonal lift each year in January and in the summer but on average - things are declining.
I compared the share of US Internet searches in the most recent four week period to a year ago, two years ago and three years ago. The share of Internet searches was down 42% in the past year, 50% over the past two years and 63% over the past three years.
As mentioned in my post, I don't think that the data will correlate as closely when we take the analysis back further but I am eager to get my hands on the data to see what we find.
Hope that helps and thanks again for the thoughtful comment.
Heather
Posted by: Heather Hopkins at April 25, 2008 06:34 AM
Interestingly enough, our clients have seen a fairly sharp increase in traffic so far in 2008 as opposed to 2007.
Is internet search traffic as a whole up or down over the period you studied?
Paul Gage
Senior Product Manager
Z57 Internet Solutions
www.z57.com
Posted by: Paul Gage at April 25, 2008 01:36 PM
Hey Heather I blogged about this over at Inman and a reader wants to know what's up with the Y-axis label that runs 4 million to 6.5 million.
http://www.inman.com/blog/2008/04/24/losing-interest
Looks to me from NAR's data that maybe what you were looking at was the seasonally adjusted annual rate of sales rather than median home price?
http://www.realtor.org/Research.nsf/files/EHSreport.pdf/$FILE/EHSreport.pdf
Posted by: Matt Carter at April 25, 2008 03:00 PM
Thanks for the update, Heather. Because most people start their search for a home online, it makes sense that there would be a correlation between the rate of sales and monthly share of "homes for sale" searches.
What I thought was interesting about the original proposition you put forward -- that there was a correlation between home prices and searches -- was that it might have meant that people were losing interest in even shopping for a home because of their fears about falling prices.
We hear a lot about fence sitters who are waiting for prices to bottom out. It would be interesting to hear from you if "homes for sale" searches pick up before prices bottom out. Maybe there will be some sort of inverse relationship between home prices and searches as prices descend into territory where people think it's time to go bargain hunting.
Posted by: Matt Carter at April 30, 2008 08:00 PM
