The search results delivered by Ask Jeeves, Google, and Yahoo differ substantially from one another, according to a study conducted by search site Dogpile.com in conjunction with researchers from the University of Pittsburgh and Pennsylvania State University.
Using a random sampling of 10,316 keywords taken from query logs, the study found that just over 3% of the returned results--10,712 of 336,232 links--were shared by Ask Jeeves, Google, and Yahoo. Some 12% of the returned results were listed by two of the three search engines. And 85% of the results were unique to one of the three search engines.
The findings demonstrate that the major search engines are not interchangeable and that metasearch engines offer a broader range of top-ranked results, says Dr. Jim Jansen, assistant professor at Penn State's School of Information Sciences and Technology.
That's a message Brian Bowman is eager to sell, as befits the VP of marketing and product management for InfoSpace Search & Directory, the company behind metasearch engine Dogpile. "If you're using just one engine, you're not getting all the results from the Web," he says.
As a metasearch engine, Dogpile aggregates search results from Ask Jeeves, Google, and Yahoo. MSN Search results should be available on Dogpile later this year.
The newly redesigned site lets users compare search results from different search engines on a single page. It also features a new tool called "Missing Pieces," which lets users enter a keyword and displays a graphic to illustrate the extent to which the returned results from Ask Jeeves, Google, and Yahoo overlap.
Beyond simply gathering search results from several sources in one place, Dogpile tries to make the results more relevant by combining sponsored and algorithmic results in a single ranked list. The text label "Sponsored by" distinguishes one kind of result from another. This is a significant departure from Ask Jeeves, Google, and Yahoo, all of which separate paid links from unpaid ones more clearly using color, graphics, and position on the page.
Dogpile also adjusts the order in which search results appear using its own search technology. Depending on how "commercial" Dogpile's engine judges a search to be, users will see either more or fewer sponsored links in the results returned. For example, a search using the keywords "digital camera prices" returned eight sponsored listings among the top 10 results. The keywords "president of china" didn't return any sponsored links among the top 10 results.
While Dogpile's self-sponsored survey--in which a metasearch engine finds there's value in metasearch engines--makes a strong case for using the site, consumer usage patterns have a logic all their own. As Yankee Group analyst Su Li Walker sees it, Dogpile faces a challenge winning consumers away from more familiar search engines.