For the first time, comprehensive estimates of search activity by Internet users is being made publicly available, and it sheds new light on which search engines are really the most "popular."
We've always known that search engines are popular in general. Both Jupiter Media Metrix and Nielsen//NetRatings publish lists each month that show which web sites are most used. These lists are consistently riddled with search players such as Yahoo, MSN, Google and other familiar search names.
The problem has been that the published figures were not search specific. Traffic to Yahoo would include those who were checking email or using the auctions area. Traffic to MSN would include anyone visiting an MSN web site, rather than just those using MSN Search. As a result, it was impossible for the public to know just how popular these portals were in relation to more search specific sites, such as Google.
That's now changed. Jupiter Media Metrix has now publicly released a top 10 list based on search traffic, which you can find via the Search Engine Watch page listed below. The new figures provide a much better picture of who is most popular in terms of search.
For one thing, the figures put the portals in much better perspective. For example, in the past year, both Yahoo and MSN would typically be shown as having around a 60 percent share of the Internet audience. This means about 60 percent of the people using the Internet would go to these sites in any given month.
When you look at those users who actually search at these sites, the shares are cut in half, to about 30 percent of the Internet audience. That still puts MSN and Yahoo at the top of all search sites, but it also makes the stature Google commands in the search world even larger.
Google had a nearly 25 percent share of the Internet audience that searched in January 2002, putting it just behind two of the web's major portals -- and the Google figure keeps rising each month. More impressive is that Google has achieved this without advertising or having the advantage of a browser that routes people to its site. Word of mouth and good press continues to be what brings Google traffic.
Both Jupiter Media Metrix and NetRatings have released lists of "top search engines" before, but the new figures from Jupiter Media Metrix are substantially different and better than those other lists. Those lists were only for sites that told Jupiter Media Metrix that they should be classified as "search engines." Portals -- which have substantial search traffic -- typically weren't included, so the lists were misleading.
The new Jupiter statistics look at search-specific URLs. In other words, anyone searching at Yahoo will get back a results page served from search.yahoo.com. Similarly, MSN Search will send results out of search.msn.com. This allows search-specific traffic to be broken out from other types of activity.
The numbers are much, much better than what we've had before, but they still remain open to criticism. For example, the Jupiter figures show that MSN Search was the top search site in January 2002, with a 36 percent share of the Internet audience. Yahoo is a close second, at 33 percent, but Yahoo claims that it would be above MSN Search, if certain types of traffic weren't included.
Specifically, Yahoo is upset that people who enter search terms into the Internet Explorer address bar get results from MSN Search that are counted into MSN Search's total. The contention is that some people may enter incorrect domain names or be confused about how to use their browser, resulting in searches that aren't really searches.
I didn't hear back from Yahoo in time for this article to include a quote from them on this issue, but there is a link to an AFP article below, where Yahoo speaks about this. As for MSN Search, it admits that these "autosearches," as it calls them, do get counted. However, it doesn't agree that all of these are errors.
"Jupiter Media Metrix does count searches from MSN Home page, as well as from the address bar. Yahoo has said that address bar searches [autosearch” shouldn't be counted because a lot of them are errors. Well, most of them aren't errors, they are bona fide searches by looking at the query logs. One could debate whether they are intentional or not, I suppose, but that's impossible to prove without asking everyone. As for the DNS Error Service, a feature of our Autosearch implementation, while that is included, guess what--we were comfortably ahead of Yahoo before we debuted the service," said MSN Search general manager Bill Bliss.
In addition, Bliss says that searches conducted on MSN Search via the Internet Explorer search button are real searches that MSN Search doesn't get credit for.
"Their tracking software only tracks pages whose URL appears in the address bar, and since the IE Search button pages do not, it doesn't get counted. Therefore even if one believes we were overcounted in one category, we are undercounted in another," Bill said.
The debate between MSN Search and Yahoo over who's the most popular has broken out before, last October. That battle was confusing, because MSN cited the private figures it had from Jupiter Media Metrix about being the top search engine, which are now being released publicly. Meanwhile, Yahoo fought back with Nielsen//NetRatings figures showing that it was the top portal.
Not only were the figures from different firms, but they also compared different types of traffic. Of course, it would be nice to see another set of search-specific figures from NetRatings. This may come, in the near future.
"There is a lot of analysis required to determine the exact and total search-related URLs to each site. We are working to incorporate a new reporting system that will automate the aggregation of the various search URLs. We anticipate that this project will be completed in the next few months," said NetRatings spokesperson Christine Chan.
A third player in the Internet ratings game, StatMarket, has always provided figures about search-specific traffic. It generates these by measuring visitors to web sites that use its tracking tools. Unfortunately, StatMarket doesn't conduct regular "top 10" or "top 50" public releases in the way that Jupiter and NetRatings do. That makes it hard to depend on the StatMarket as a source for popularity figures.
The latest figures from Jupiter Media Metrix are great to have, but they leave you wanting more, as they focus only on "unique visitors" rather than usage. Usage could show a different picture.
For example, a person might visit Yahoo only once in a given month and would be counted once towards Yahoo's share. The same person might then visit Google everyday in the same month and conduct several searches per day. Despite this, they would still count only once in the "unique visitor" figures.
Even if usage figures begin to be provided, they still have to be carefully evaluated. For instance, Jupiter Media Metrix last year issued a special report suggesting that the less time that is spent with a search engine, the more satisfied a user must be, given that they must be finding what they wanted quickly.
Under this theory, Overture -- with an average usage time of 56 minutes in April 2001 -- was deemed to be more satisfying that Google, which had an average usage time of 269 minutes. Of course, if people like Google's results, they'll do more searches there and accumulate more usage time overall.
Jupiter Media Metrix Search Engine Ratings
A rundown on the new search-specific stats can be found here.
Jupiter Media Metrix Changes its Tune on Pop-Ups
@NewYork, Dec. 26, 2001
More background on how Jupiter Media Metrix is subdividing the figures it reports and moving pop-ups and unders into their own category.
NetRatings, Jupiter Call Off Merger
@NewYork, Feb. 19, 2002
The proposed merger between ratings services Jupiter Media Metrix and Nielsen//NetRatings is off, due to US Federal Trade Commission objections.
Experts Mull Jupiter's Options
@NewYork, Feb. 24, 2002
Follows on the article above, about how Jupiter, facing a cash crisis, is now looking at ways to survive.
NetRatings, Jupiter to Resume Patent Lawsuit
@NewYork, Jan. 18, 2002
NetRatings and Jupiter Media Metrix are involved in a lawsuit over tracking patents.
MSN Search Surpasses Competition to Become No. 1 Search Service In the United States
Microsoft Press Release, Jan. 10, 2001
Figures and stats on how MSN says it ranks in the search space, based on Jupiter Media Metrix figures.
Microsoft, Yahoo debate search engines' popularity
AFP, Jan. 12, 2002
This article cites Yahoo's objections over the figures naming MSN Search as the top search engine.
Yahoo, MSN Spar Over Traffic Figures
SiliconValley.internet.com, Oct. 12, 2001
We're the biggest, says Yahoo. No, we're the biggest, counters MSN. A look at the dueling audience figures. Yahoo claims its 210 million unique visitors worldwide in September -- as measured by Nielsen//NetRatings -- makes it the largest global web property. MSN says it had 270 million unique visitors according to Nielsen//NetRatings-rival Jupiter Media Metrix.
Yahoo, MSN Battle Over Traffic Figures
NewsFactor Network, Oct. 12, 2001
Another story on the audience dispute. Note the analyst comment that the time a user spends on the site is more important than the site's audience figures. That's wrong. The statistics, any of them, mean nothing on their own. So what if a portal has someone on their site for an average of an hour per month? If their competitor only keeps visitors for 30 minutes per month but earns more per visitor, then the time means nothing. Or if a web site has millions more visitors than its rivals but fails to earn much from them, it's not going to stay in business -- as anyone watching the dotcom fallout over the past year understands well.
Search Engine Marketing Finally Getting Respect
The Search Engine Report, Aug. 2, 2001
See the first link listed at the end of the story for more about the Jupiter Media Metrix time analysis report that was conducted.