The Brancatelli File



August 5, 1999 -- Two scientists recently spent six months rating search engines and came to what now seems an inevitable and foregone conclusion: The Web is expanding so fast that none of them can keep pace.

According to the scientists at the NEC Research Institute, there are now at least 800 million publicly available pages on the Web. Of the eleven search engines they studied, only Northern Light managed to cover as much as 16 percent of those 800 million pages.

The report, summarized at, raises another obvious question: If the search engines fall short when it comes to quantity, how do they stack up when it comes to quality?

Not much better, according to an test I performed on ten of those search engines. I went looking for information on business travel and came back with links to African safaris, newspapers in South Korea and South Africa and even a site that urges business people to use commerce to "glorify god."

My little ad hoc test was simple enough. I wanted information and I wanted it fast, so I entered the term business travel (no quotation marks, boolean terms, or other fancy techniques) into each of the search engines. Then I examined the first fifty results for relevance to business travelers.

What did I consider relevant? Any result that linked me to a page that delivered some form of useful business-travel information. I discounted links to the home pages of travel agents, airline and hotel home pages. (If I wanted an airline, hotel or travel agent, I would have entered one of those terms in the search field.) I also eliminated results that pointed me to other Web sites' links pages. My logic there was simple enough: I came to the search engine for links to business travel, not links to someone else's links to business travel.

As you can see by the chart below, the best performer was Google, which returned 18 results I deemed relevant. That's about one relevant link for every three results. Not great, but all the other search engines fared even less well. What you can also deduce from the chart is confirmation of what most search-engine users have always believed: If relevant links doesn't pop up fast--in the first 10 or 20 results--there probably aren't too many good links buried farther down.

Here's a snapshot of how each search engine performed in my test. They are listed according to the Web-coverage percentage assigned to them by the NEC Research Institute study.

NORTHERN LIGHT is impressive in its ability to find pages containing the words "business travel." But it did a poor job finding pages that actually delivered useful information. More annoying: about a dozen of the first 50 results were links to link pages. Worst of all, those pages all offered links to relevant business travel sites that Northern Light itself didn't find.

SNAP was, to be blunt, nearly useless. Almost all of its first 50 results led to leisure-travel or travel-agent pages. Snap didn't even flag nine Web sites cited in its own "business travel" category. One saving grace: Snap's one-page Business Traveler's Guide, which popped up as search result No. 11. It was chock full of excellent data.

ALTA VISTA started strong. Its "related search" feature, which appeared atop the first page of results, offered excellent alternative search paths. And its first result was on-target. But the first 50 results were cluttered with travel agents, leisure-travel sites and three cites for a bed-and-breakfast inn in Port Townsend, Washington.

HOTBOT hit the mark for quick connections. Its first two results were solid business-travel sites. Six of the first 20 results were also on-target. But you have to wonder why HotBot found's "Twisted Travel" page, but not that Web site's "Business Travel" page.

MICROSOFT struck out on 48 of its first 50 results. And you know that Twisted Travel page from Microsoft found it four times.

INFOSEEK performed admirably throughout. Its first two results, and five of the first 10, were useful. The first search results page also offered an excellent "matching categories" search.

GOOGLE may be the search engine that you've lusted after. Fifteen of the first 20 results were dead on the business-travel target. Google also did an excellent job finding business travel related pages on general sites. Two examples: an article on the deductibility of business-travel expenses from the California Society of Certified Public Accountants and a primer on business travel for women from Oh, by the way, Google found the Business Travel page without referencing that Twisted Travel page.

YAHOO! performed, well, er, like Yahoo! After entering the search term, surfers are not brought to Web site results, but dumped instead into Yahoo's "category matches" page. Among the category results: pet travel, "adult" travel and travel booksellers. After clicking on the "Web Sites" hot link, the first 50 results were almost all leisure-travel sites. And Yahoo! found's Twisted Travel and its "Active Travel" pages without finding its Business Travel page. Surfers comfortable with Yahoo's category-matching system will be fine. If not, the search is confusing.

EXCITE found that elusive Business Travel page without turning up either the Twisted or Active travel pages. Unfortunately, that's all Excite found in its first 50 results. Actually, the first item Excite returns on its search results page is a "Try This First" area with a business-travel hotlink. But that led to a promotion for a book.

LYCOS, like Yahoo!, brings you to a "categories" page and you must click a second time to get to the Web site search results. Once there, you got four solid links in the first 10. However, half of the next 40 results were links to pages located at just two Web sites.


Search Engine

Web Coverage1

Total Pages Found2

First Good Result3

Total Good Results4

Least Relevant Link5


Northern Light





A page to "assist clients to glorify god through their businesses"

Most results led to travel-agent pages or pages with links to business-travel sites that Northern Light should have returned.






A Dallas travel agent's wedding registry

First good result was a link to an internal Snap page. Most results linked to leisure-travel information.

Alta Vista


87, 309



The Korea Post Web site

Sloppy, too many duplicate results and too many travel-agent links.






Yahoo! Travel's Japanese home page

Six of the eight relevant results were returned within the first 20 cites. Too many leisure sites.






A page offering "travel tourism jobs"

Microsoft returned lots of duplicate results and virtually all of the original sites were about leisure travel.






Virtual tour of China's Forbidden City

Returned many foreign pages that referenced business travel. Very few bizarrely off-target results.






A page promoting African safaris

Excellent fast resource, but too many duplicate results after the first 20 links were returned.






Out&About, a gay-travel Web site

Yahoo! didn't seem to differentiate between business and leisure sites. Not useful for quick searches.






Cape Business News, a South African Web site

Off-target links throughout, many leisure-travel results. Cluttered, hard- to-read layout of results.



Not listed



A page about home-based travel agencies

Confusing and unnumbered results. More than half the results were pages from two Web sites.

KEY: 1Percentage of 800 million Web pages covered, according to NEC Research Institute.2As reported by the search engine. 3First result returned by search engine considered relevant for business travel. 4Number of first 50 results returned by search engine considered relevant for business travel. 5Among the first 50 results returned by search engine.

This column originally appeared at

Copyright 1993-2004 by Joe Brancatelli. All rights reserved.