What's the most popular search engine? There's a new metric available for determining this, the total number of "search hours" users spend at a site, a figure that sees Google leading by far over other search engines.
Search hours can be provided for the first time based on figures from two different sources, Jupiter Media Metrix and Nielsen//NetRatings. Both companies have recently began releasing figures showing the total number of people performing searches at popular search engines and portals, along with the average time spent by those searchers at these sites.
By multiplying the two figures, unique visitors by average time spent per visitor, you get what I call "search hours." This is the total time spent by all visitors searching at each search engine.
At the top of the list in search hours is Google, where users spent nearly 13 million hours searching during March 2002, according to Nielsen//NetRatings. In contrast, second-place Yahoo came in with less than half that figure, 5.4 million search hours, followed by MSN Search with 4.9 million search hours logged. Figures are for US home and work internet users.
Jupiter Media Metrix also puts Google at the top of the list, in figures from February 2002, with 11.7 million search hours. Yahoo again comes second at 6.5 million, then Ask Jeeves is placed third with 4.8 million hours. Again, figures are for US home and internet users.
Search hours should not be taken as an absolute measurement of what's the most popular search engine, but the new figure is probably far more useful than the standard "audience reach" statistics that have long been released. Those figures show what percent of the internet audience has searched with a particular service in a given month.
When looking at audience reach, it is either Yahoo or MSN Search that tops the list, depending on which rating service you cite. However, audience reach doesn't reflect total usage. This is because the audience reach figures count someone only once, even if they go back to the same search engine many times per day.
For example, let's say you go to Google five times per day and search every day, during a given month. In the same month, you also go to Yahoo once and perform a single search there. Despite all your activity at Google -- 150 searches -- you still get counted only once as a "unique" visitor.
In contrast, citing search hours provides a better blend of both popularity and usage. For instance, even though Google gets fewer unique visitors than Yahoo, those visitors spend more cumulative time searching at Google over a given month, so Google ranks better when search hours are considered.
There is one big potential flaw with search hours. It might take longer to find what you are looking for at a search engine with bad results, so looking at cumulative time could reward bad search engines.
I can't disprove this notion, but I think it's unlikely. Even other portals acknowledge that Google's usage per month is extremely high because people go back there often. It's hard to imagine that people would keep going back and spending time at a search engine that wastes their time.
A third metric of assessing popularity is ranking search engines by the amount of traffic they send to web sites. Using this metric, Google and Yahoo recently got top honors, in releases by recently made by two different traffic tracking services.
According to StatMarket, Yahoo accounted for more search referrals than any other search engine, 36 percent, based on worldwide traffic measurements for April 24, 2002. Google was just behind with 32 percent of referrals, followed by MSN with 13 percent.
Measuring service OneStat.com placed Google well ahead Yahoo, saying Google was responsible for 47 percent of search referrals worldwide, followed by Yahoo at 21 percent, then MSN Search at 8 percent, AltaVista at 6 percent and Terra Lycos (presumably Lycos.com) at 5 percent.
However, you begin to lose faith in the accuracy of the OneStat numbers when told that Ixquick is sixth on the list for driving traffic, 2.4 percent, ahead of AOL Search at 1.6 percent. Ixquick is a great meta search engine, but for it to outperform the incredibly popular AOL Search service is not reflective of many people's experiences, especially in the United States. It suggests that OneStat's "global" data is perhaps skewed more toward a particular country.
Nielsen//NetRatings Search Engine Ratings
March 2002 figures for search hours, as well as audience reach and average time spent per visitor can be found here.
Jupiter Media Metrix Search Engine Ratings
March 2002 figures for search audience reach and average time per visitor, and February 2002 data on search hours.
NetRatings Vs. Jupiter Media Metrix
Special bonus sidebar article for Search Engine Watch members, charting the latest figures from both services, so you can see the differences visually.
Jupiter Media Metrix Releases New Search Specific Traffic
SearchEngineWatch.com, Feb. 19, 2002
Further background on how search-specific figures have only recently begun being released and why this is significant.
Jupiter MMXI Europe Ratings
Recently updated with visitor figures for Europe.
Jupiter Media sells European clients to NetRatings
Reuters, May 9, 2002
Savor those Europe stats above, because they look to be the last from Jupiter for Europe. The company's contracts have been sold to NetRatings. I'll see if I can start getting Europe data in the future from NetRatings to post.
NetRatings settles with Jupiter Media Metrix
San Francisco Chronicle, May 8, 2002
One time suitors NetRatings and Jupiter Media Metrix settle a patent infringement lawsuit over tracking technology. Jupiter Media Metrix's cash is dwindling, so we may be looking at only one major rating service left by the middle of the year.
Google Challenges Yahoo as the No. 1 Search Site in the World
StatMarket, April 24, 2002
More on the StatMarket figures. First URL probably won't work in the long term, so use the second URL and scan for the story, if the first fails.
Google is the most popular search engine on the web
OneStat.com, April 15, 2002
Figures here from OneStat.com, with additional ones ranking Dutch search engines.
NOTE: Article links often change. In case of a bad link, use the publication's search facility, which most have, and search for the headline.