Eyetrack: How you read the news online

What are you looking at?...

And what were you looking at a second ago? What will you look at next, and while we're at it, have you checked out the ad - there, just below the navigation bar?

Are you back yet?

You may not devote a great deal of thought to just how you go about digesting the content of web pages, but it can be of vital interest to the designer of a website (not to mention any potential advertisers for that site). And while some researchers might try to judge a layout's effectiveness through traditional surveys, a more fittingly high tech method of measuring the path of viewers around the screen involves actually tracking the movements of each visitor's eyeballs, in a controlled laboratory setting. Eyetrack introduces visitors to both the means and results of measuring our most fundamental interaction with the internet.

A joint project of the Poynter Institute (a journalism school founded by the chairman of the St. Petersburg Times), The Edward W. Estlow Center for Journalism & New Media at the University of Denver, and Eyetools (an eyetracking research company [which, by the way, will happily test your own homepage for $500]), Eyetrack III set out to gain objective, empirical data about how surfers read news-based websites.

While a conventional survey risks falling victim to the differences between observation and retention -or the simple fact that some test subjects may not want to admit that they spend more time reading comics than international news- Eyetrack simply recorded the viewer's optical reaction to a series of real and mock web pages. Results were then plotted - as movement charts for individual visits, and Heatmaps of cumulative reactions to single pages. (Similar in appearance to weather radar images, Heatmaps display the color red over parts of a page where eyes have lingered, and shift to cooler shades in less trafficked areas.)

Eyetrack's own homepage is basic, with only a single image in the top half of the screen (introducing visitors to the two methods in which experimental results are presented). The site's main sections are laid out in a two-row navigation area above the content -the first row relating to procedure, the second to observation - while a handful of teasers and more specific article links are arranged below. Linked to the aforementioned image, Eyetrack's first article provides a "Best of" overview of the study, and also whets the appetite with a surprise or two. (For example, despite the apparently obvious visual advantage the web has, in the ability to add as many no-extra-cost, full color photographs as desired, it turns out that test subjects generally began their perusal of the selected websites with the text rather than images.)

The second row of navigational options leads the visitor to additional - and more detailed - experimental results. Homepage Design compares variations on five typical homepage layouts, with articles on such factors as the usefulness of blurbs accompanying headlines, and the relative merits of compact (all content visible at once) versus extended (scrolling necessary) homepages. (The latter study incidentally revealing that even paperless online newspapers place some articles 'below the fold.')

Article Page Design (you're on one of those now) is next, followed by Multimedia Results, first compares differences in viewer recall between text-based and multimedia versions of the same story, and then reviews the effectiveness of seven specific multimedia presentations. Finally, Advertising Results explores which sectors of web page real estate equate to Super Bowl half-time, and which to late night infomercials.

Throughout the site, a changing list of supplemental readings is arranged down the left side of the page - the changes reflecting the main content on a given page. The transitory nature of the list, along with the fact that some links are to new pages, while others are simply to anchors on the existing page, can lead to some confusion about how much of the website you've actually seen. Additionally (on my Mac, at least) the small font size was just on the other side of legibility - requiring me to blow up all the text in order to read the index. But there are definitely links worth pursuing here, such as a pair of movie clips showing surfers' eye movements through a website, and a quiz which tests the visitor's ability to predict where a reader might be likely to "fixate" on a mock newspage.

With a sample size of 46 subjects, each spending only one hour online, and with no requirements to view any specific pages available in the study, Eyetrack can't be considered - and isn't claimed by the partners - as being a statistically significant representation of the viewing public.

It is, nevertheless, an interesting experiment, and for those of us whose careers don't revolve around visual design and ad placement, it's the kind of thing we probably never would have seen before the internet. Before the web, a similar study of magazine layout would have likely been printed in a trade journal, and remained unknown outside media circles. Accessible on the web, Eyetrack is appearing in dozens of recommended links collections, and anyone with the inclination now has the opportunity to see some of the thinking that goes into the pages that occupy so much of our time.

Still, there remain a few burning, yet unanswered questions. One can't help but wonder if there was a good deal of angst -during Eyetrack's own design stages - over the layout of a website dedicated to studying effective website layout. And when it was complete, did the Eyetrackers Eyetrack the Eyetrack site? What did the results tell them? Perhaps it would be better if they didn't self-test. (If you gaze too intently, there may be a risk of disappearing into your own navel.) Perhaps they did, and we'll learn the results in Eyetrack IV.

Eyetrack III can be found at http://www.poynterextra.org/eyetrack2004/index.htm.