I wrote in this space last week
that some Web sites had doubled or tripled their sales results after learning
what visitors were actually seeing on their pages.
The company that makes studies like this possible is called Eyetools Inc. This
San Francisco-based firm uses sensors atop computer monitors to record what
people actually look at on a Web page. This monitoring process, called
eyetracking, eliminates the headgear that was worn years ago by human
subjects to measure their interest in TV commercials.
One of the most important findings of the study I previously reported on was the
role of images on Web pages. People's eyes are immediately drawn to images
on-screen, so much so that their normal left-to-right reading patterns are
disrupted. Placing an image on the right side of a Web page can reduce visitors'
comprehension of words on the left side.
There's more to eyetracking than just sales, however. Almost every aspect of a
company's Web site can be studied using this technique and I'll show you today
how you can benefit from some of this research for free.
A Blog A Day Keeps The Blues Away
Eyetools used to charge $20,000 to $30,000 for tests it conducted at corporate
sites, but it now emphasizes studies that cost only $1,000 to $2,500 (provided they
can be performed in Eyetools' San Francisco lab). If you're not sure your
company should spend even that much on eyetracking tests, you're in luck. The
company last month quietly began giving away some of its best results for free.
Interesting research findings, which Eyetools is not restricted by customer
agreements from publishing, is now regularly posted on a blog by company CTO Greg
Edwards and others. Your company can probably learn a great deal from the
analysis of other companies' Web sites, no matter how different your industry
may be.
One of the most recent blog reports shows the eye patterns of visitors to
Google.com, a project conducted with search-marketing firms Enquiro and Did-it.
In the image below, the red areas show those parts of the Google page that were viewed by 100% of the
visitors tested. The green areas were viewed by 50% of the visitors, and the
blue areas by less than half. The solid red horizontal line indicates material
that was originally off-screen until visitors scrolled down.
Eyetools' analysis of the above eyetracking patterns is:
• Top listings. The top three "editorial" (unpaid) listings on the
Google page were seen by 100% of the visitors. The fourth listing was seen by
only 85%, and the exposure declined from there, with only 20% of the visitors
ever seeing the tenth listing.
• Top sponsored links. Sponsored links at the top of the page, and
above the first editorial listing, were seen almost as often, with 80% to 100%
of the visitors focusing on them.
• Right-hand sponsored links.
Sponsored links on the right-hand side of the Google page didn't fare as well.
Ads in the first three positions in the right column were viewed by only 50%,
40%, and 30% of visitors, Eyetools says. The fourth ad was seen by only 20%, and
the remaining four ads were noticed by just 10% of visitors.
The message seems to be that placing sponsored links in Google can get you as
much readership as being No. 1 in the editorial listings, but only if your ad is
positioned atop the results, not on the right.
Generalizing To Content Sites
The results of Eyetools' tests can be useful not just to sites that carry
advertising but also to sites that are trying to get visitors to click links on
a home page to jump to inside pages.
The firm analyzed the home page of the Washington Post, which had just been
redesigned on Feb. 15. The eyetracking results are shown in the image below.
• Good readability to start. The top portion of the Post's home
page shows good design, Eyetools' analysis says. The use of white space and
ample line spacing makes people read much of the material and even encourages
them to scroll down, something many Web surfers rarely do.
• A rapid fall-off down below.
Below the "fold" (the bottom of the screen), the story is quite different. A
lack of white space and cramped line spacing sharply reduce the reading of this
area, Eyetools says.
Conclusion
Since Web pages are designed to serve many purposes, a behavioral study of what
people do at one site may not apply to another site. But the way people scan Web
pages is very consistent, even automatic, Eyetools says.
This makes a test of your own company's Web site a unique opportunity to see
your pages through others' eyes, Edwards said in a telephone interview. "We
usually recommend 10" as the minimum number of test subjects for meaningful
results, he says. That relatively small number, fortunately, means that an
investment of only $1,000 or so is required for a simple test of a typical page.
For more information, Eyetools' main site is at Eyetools.com. Its blog, where the examples
shown above are discussed, is at blog.Eyetools.net.
I wrote in this space 
IE 7
Firefox 2.0
