Yet Baidu.com Inc. is strong enough in China and NHN Corp. in South Korea to crack the global top five in comScore Inc.'s inaugural report on worldwide search patterns. The report, based on August traffic patterns, was scheduled for release Wednesday.
In the past, comScore reported search numbers for only a handful of countries. The numbers from comScore and rival Nielsen/NetRatings are closely watched by industry analysts, even as the measuring firms use online recruitment techniques dismissed by many traditional pollsters.
According to comScore's qSearch 2.0 service, more than 37 billion searches worldwide went through Google in August. That's about 60 percent of all searches, higher than Google's 50 percent in the United States.
Yahoo Inc. was second worldwide with 8.5 billion, followed by Baidu at 3.3 billion, Microsoft Corp. at 2.2 billion and NHN at 2 billion.
In China, one of the few countries where Google isn't dominant, Baidu shows how one regional player "can break into the top five globally by their complete control of a very, very large market," said Bob Ivins, comScore's executive vice president. Baidu's numbers would likely keep increasing, he said, with China's online population.
Baidu entered the Chinese search market much earlier than Google, while NHN successfully tapped Koreans' preference for human interaction over software in getting search results.
Danny Sullivan, editor in chief of the industry Web site Search Engine Land, said Baidu's strength shows that Google isn't invincible and "can be beaten in markets ... but it's still going to be a challenge."
Sullivan said search numbers from comScore and Nielsen offer "an extremely rough guide" to the state of the industry, and a shift in market share - or lack of it - could help point to whether a new service or strategy is working.
But many financial analysts say the search numbers aren't as useful as cash flows and ad revenue in gauging a stock's performance - in Google's case, sailing past $600 for the first time Monday.
"This is a data set that investors understand has its useful purpose, and predicting quarterly (financial) results is not its useful purpose," said Rob Sanderson, an analyst with American Technology Research.
The worldwide totals were calculated using measurements from 2 million Internet users in some 170 countries. Participants agree to install tracking software and are often enticed by free products and services like screensavers, games and online backup.
ComScore also has a smaller panel recruited using traditional, phone-based methods, and the search figures are adjusted to account for differences in age, income and other factors between the panels. ComScore, like Nielsen, says it needs online techniques to assemble a large enough, albeit self-selected panel.
With Internet users more fragmented than the television audience, Ivins said, "I don't think you can do what we're doing on a scale we're doing with a toolkit from the last century."
Many traditional pollsters are skeptical.
Such an approach assumes that those who initiate participation, such as responding to a free offer, behave similarly to those who do not - an assumption that still needs more research to prove, said Nancy A. Mathiowetz, president of the American Association for Public Opinion Research.
Jordan Rohan, an Internet analyst with RBC Capital Markets, is aware of limitations but considers the numbers "the best I've got."
ComScore estimates that about 750 million people worldwide used Internet search in August, each person averaging about 80 searches.
Europe and Latin America tend to have more searches conducted per person. Ivins said Internet penetration in those markets grew as search technology was already developed, unlike in the United States where human-powered directories were initially strong.
"It just became natural for them" to use search, Ivins said.