How many search engines do you use regularly? A Nielsen/ NetRatings look at the top three engines back in January—Google, Yahoo! and MSN—found that 58% of Google users that month also ran queries on one of the other two engines. The numbers of Yahoo! and MSN users who also switched to a back-up engine were even higher: 70% or above.
So it’s not like users don’t recognize that different engines produce different results. Still, we’re creatures of habit, and our mouse tends to gravitate to the same Web site when we want to search. Web monitor Hitwise found that those three engines got more than 93% of the U.S. Internet search traffic in July, with Google laying claim to 59% of the visits—a 14% increase over July 2004.
But recent research produced by metasearch engine Dogpile in collaboration with the University of Pittsburgh and Pennsylvania State University suggests that searchers who use one search engine all the time are missing out on a large portion of potential results—results that other engines consider relevant enough to put on their first results page. Even those using two or three engines are not getting all the Web has to offer in answer to their queries.
The study, published in July, looked at results for 12, 570 random user-entered search queries on the four top engines—adding Ask Jeeves to the troika. Researchers examined the first page of organic search results to see how much overlap there was among the four engines. That “first page” qualifier is significant because relatively few searchers look beyond the first results page; in a separate study, Dogpile found that almost 90% it they don’t see what they want, they tend to start another, more focused search. An earlier study by Dogpile found that almost 90% of all click activity happens on the first search results page.
The conclusion of the July overlap study: In only 1.1% of cases did a search produce the same first page of organic search results on all four of the engines. Three engines produced the same total first-page results in 2.6% of the cases; and two engines showed the same results in 11.4% of the cases. In 84.9% of the cases, the first-page results produced were unique to one engine.
In other words, if you’re only searching with one engine, you may be missing a lot about two-thirds of the time. If you’re using Google, the study found, about two-thirds of your first-page results are unique to Google. For Yahoo!, that figure is 71.2%. For MSN, it’s 70.8%. And Ask Jeeves’ first-page results are unique 73.9% of the time, researchers found.
Dogpile has a dog in this fight, so to speak; as a metasearch engine, it pulls combined results from the Web indices created by Google, Yahoo!, Ask Jeeves and, in a deal announced in early August, by the newly re-launched MSN Search.
“Research from comScore/ Media Metrix says that people are suing about 2.8 search engines a month,’ says Brian Bowman, Dogpile’s vice president of marketing and product management. “So they’re naturally flipping among search engines themselves, because they’re not finding what they want in one place.” In a sense, those users are already metasearching.
That same comScore research studied “success” rates for searches on the four leading engines by measuring the number of times a user clicked on a result on the first search page. By that measure, Google had a success rate of 55.6%, Yahoo! 50%, MSN 46.6% and Ask Jeeves 39.7%. By contrast, Dogpile’s ratio of search-to-click conversions was 62.9%.
It’s Dogpile’s contention that it can sway users to look for results from more than one engine at a time and find what they’re looking for more efficiently. If so, it wouldn’t be the first time a large portion of the searching population went meta. Bowman points out that in the early days of the Internet, in 1995 and 1996, metasearch had a 30% share of the search market. “There were so many engines taking such different approaches to presenting information that metasearch was really an important way to get a handle on all the information coming at consumers,” he says. That changed in 2000, when Google became the single voice of search on the Web and even Yahoo! went to them for search results. “The value just wasn’t there for metasearch,” he says.
But over the last 18 months, all four of the top search engines have announced aggressive, expensive initiatives to expand their search capabilities. That renewed industry movement has reinvigorated metasearch as a concept and Dogpile as a search industry player, Bowman says.
Now the job is to get the word out to users about Dogpile and its search advantages. Bowman says the company has allocated $5 million for direct marketing itself online for the rest of the year. He won’t get specific about the measures involved—“we’re very protective of our special sauce”—but he emphasizes that the campaign is going to revolve around direct response rather than brand building.
“Look at the wonderful efforts of MSN around March Madness,” he says. “They spent between $80 million and $150 million and didn’t really get new consumer to come to the site. They were very successful at getting existing consumers to search more no net-net, they made money. But it shows the extraordinary challenge facing single search engines trying to advertise through traditional brand building.”
Part of the effort will involve customer education about the benefits of using metasearch. Dogpile recently revamped a very simple, but very effective interactive search comparison tool on its Web site to include all four of the major single search engines. Visitors can enter a search term and see a Venn diagram—those overlapping circles—mapping the number of key search results they’re missing by using only one, two or even three separate engines.
Consumers can get another visual lesson in search overlap on the Dogpile main Web search results page. By clicking radio buttons, they can break out top search results for each of the four individual search engines and highlight the results unique to each. The page produced even by a very current news search term—say, “John Roberts” three days after his nomination to be Chief Justice—produces a surprising number of unique results: Out of the 34 top results on the four engines, 21 were unique, according to Dogpile.
The value proposition for advertisers is equally compelling, according to Bowman. The July UP/ Penn State study didn’t only examine organic search results for overlap; it also looked at the overlap among sponsored search listings between Google and Yahoo! (Ask Jeeves uses Google’s paid listings, although it has started to sell its own; MSN gets sponsored results from Yahoo!) On any given search query, Yahoo1 and Google offered overlapping paid listings only 4.7% of the time. In about 15% of cases, one of the pair had one or more paid listings while the other had none. And in almost 30% of the queries studied, neither Yahoo! nor Google showed a sponsored listing.
“When the consumer’s query is commercial in nature, you want to have the best possible list of merchants for them to select from,” Bowman says. “But the overlap of advertisers between Google and Yahoo! paid listings is extremely small. We’re the only search company that offers both Yahoo! and Google paid listings.” Those listings are combined with natural search results on the page, although they are designated “Sponsored by” before the URL.
It’s going to be an uphill battle. The public’s understanding of metasearch is pretty low, and Dogpile is fighting the force of sheer Web habit. Still, Bowman says that Dogpile expects it will be able to build its current 1% market share to 5% by pushing its value message to consumers. “Our goal is not to leapfrog Google or yahoo! but to carve out our own niche in a very profitable way,” he says.