By: Carl Sullivan
Most of us lament the lack of solid information about how audiences really use our Web sites. Do they glance at photos or headlines first? Is left-to-right reading the norm online? Do users scroll “below the fold”? And the $64,000 question: Does anyone look at advertisements?
Beyond the routine data on page views, unique users, and session duration, wouldn’t it be nice to know how readers scan a Web page before clicking off to somewhere else?
Well rejoice, weary Webizens. Later this year, some light will be shed on how people interact with news sites. At a date to be announced, the results of Eyetrack III will be released on Poynter.org. Some of the preliminary findings surprised me, and I suspect, will surprise you too.
Late last year, researchers tracked the eye movements of 50 Internet users as they perused mock news Web sites. The survey sponsors are the Poynter Institute of St. Petersburg, Fla.; the Estlow Center for Journalism and New Media at the University of Denver; and Eyetools Inc. of Sacramento, Calif. Project managers Steve Outing and Laura Ruel are still analyzing the data and writing the final reports, but Ruel offered a sneak preview this week at the America East New Media World conference in Hershey, Pa.
Eyetools provided sophisticated special cameras installed at the bottom of computer monitors to track where a person looks on a Web page, and where they stop to fixate on particular content. Aggregating this data, the researchers have built “heatmaps” which illustrate which parts of the mock pages got the most eyeballs. These heatmaps sort of look like a Doppler radar image from a weather report — but here red and orange don’t indicate heavy precipitation, they indicate hot spots where most users focused their gaze for at least a fraction of the section. The areas of the Web page that got a cold shoulder from viewers are a dark icy blue.
The researchers built different types of mock pages based on the standard types of news sites online today. By comparing the heat maps, the researchers can show various Web designs are used differently.
One preliminary finding: Home pages that ran blurbs underneath headlines tended to cause a drop off in reading — when compared with home pages that ran headlines only. Ruel isn’t ready to draw any conclusions from this particular finding, but speculates that Web users accustomed to performing searches on Google may only look at the headlines in Google searches and rarely read the blurb below — unless they specifically don’t get enough information from the headline. Users may have a similar mindset when visiting news sites. If this theory is true, “good headline writing has exponentially gone up in importance on the Web,” Ruel says.
Home pages that had clean breaks at the fold (the bottom border of the page visible in a Web browser when the page first loads) typically got fewer views below the fold. In other words, it might be best to avoid lines or other graphical elements which may suggest to viewers that the page ends at the fold. Instead, Ruel says page designers might consider having text with breaks that obviously continue below the fold — to encourage scroll downs.
Photos? They typically got far less eyeballs than text. The one caveat is that photos with recognizable human faces in them did tend to get more glances. Not surprisingly, size is also a factor. But still, text gets a lot more action on news sites. “People tend to be very task-oriented on the Web” so maybe they’re less interested in photographs, Ruel suggests.
On shorter, less-packed home pages, the navigation and advertisements got more looks from users. Also, top navigation tends to get looked at more than left-hand nav bars.
Here’s the part you probably won’t want to share with your advertisers. On every type of mock news page presented, users discerned where the ads were and ignored them. Ooops! We all know the sad tale of low click-through rates, but it was still shocking to see the ads virtually ignored on the Eyetrack heatmaps. Not only are readers not clicking; they’re not even looking. The one exception in this study was a simulated all-text classifieds section similar to the “Top Jobs” feature that runs on some newspaper Web sites. Quite a few eyeballs stopped on those ads.
Another low eyeball section? The lower right-hand corner of the Web page. No matter what was there, users tended to avoid looking at this graveyard of a space.
Ruel says the study also shows that multimedia elements don’t necessarily increase reader recall and comprehension of the content. So in some cases, presenting news the old-fashioned way — in plain text — may be best way to use new media.
Watch Poynter.org for the publication of the full research.
(Editor’s note: Eyetrack project manager Steve Outing writes a monthly column for E&P Online.)