Steve Outing, co-project manager for Eyetrack III, answers a few of our followup questions about the recent research findings released by The Poynter Institute, the Estlow Center for Journalism & New Media, and Eyetools.
Here’s the transcript of our e-mail interview with Outing:
What’s different about your approach to research this time?
We took a much more methodical approach this time, designing and creating 5 mock news websites for our test subjects to spend time with. For these 5, we created a variation of each, to test specific things. Half of our group was one variation of each of the 5 sites, the other half saw the other 5, so there’s statistical significance to some of the findings (while other things were less stringently tested, so we offer “observations"). We also devised a multimedia/text comprehension study, combining eyetracking with written questions afterward to see how much people recalled in viewing a multimedia version of a story vs. a text version.
In the previous (1999-2000) research, it was more free-form; test subjects viewed various news websites while being tracked and researchers pored over the data to look for trends.
Big difference: No head gear. Just a person sitting in front of a computer, which makes for a boring photograph compared to someone with an odd contraption on his head with wires everywhere. This is a big jump in making this testing more realistic to the actual website reading experience.
Another difference: We’re now in the broadband era, where most people have experience with fast connections — if not at home, then at work. In the 1999-2000 study, we still had mostly narrowband usage and that affected how and when people viewed photographs, for example. And multimedia content was less prevalent then, so this time we tested the multimedia news experience.
You say that this report is only preliminary and should be used in conjunction with usability, log analysis and focus groups, how do you see editors and web designers using this research to improve what they do?
I see this as complementing what they’ve been doing. You can learn things from eyetracking that you can’t any other way. My colleagues at Eyetools (our technology partner in this project) tell of a client of their’s that had an important element of its homepage pretty much in the middle of the page. Testing revealed that despite its central location, most people ignored it completely—never looked at it. With a page redesign and subsequent testing, people started fixing their eyes on it.
The power in this, I think, is in combining eyetracking with other methods of site performance measurement.
It appears that online readers are starting to develop some interesting habits for scanning. What are they doing that’s new and how might that be applicable to not just news sites but e-commerce, advertising - or blogs?
I’m not sure how much is new; I suspect that there’s a very slow change in how people read on the web over time, as people get more and more experience as online users. For instance, we still didn’t see people looking first to photographs on news sites; text was still mostly the first thing checked out. The PC is still mostly text based and historically we’re used to mostly text uses — e-mail, word processing, spreadsheets, etc. So perhaps that continues to influence how we view websites. You’ll recall early Eyetrack research indicating that in print newspapers, photos get the first looks.
I thought the results about font and headline size were interesting. Some home pages with contrasting type size (between headline and blurb) were set up to encourage scanning, we learned. When font size was small, and headlines too, people seemed to focus in more—and ended up fixating on and reading more words. It was as though we changed their behavior by altering font and headline size.
Blogs: I’d say our results about how the first 2-3 words of headlines are all that many readers see are important. Maybe bloggers will want to be more careful about headline writing, putting action words at the beginning to capture attention. And perhaps the idea of not having item headlines be too big, as that might discourage reading text and encourage scanning headlines.
Advertising: Placement is obviously important, and our observations support what’s probably obvious to most publishers: ads placed in upper and left will get seen the most; ads above the flag/masthead are often ignored, but placed between masthead and top headline at least get seen; ads inset with body text are seen the most; etc.
E-commerce: Text ads are powerful in holding interest for longer periods (in our test, 6-7 seconds on average for a block of job ads, compared to an average of 0.5-1.5 seconds for banner/display ads). Font size/contrast stuff can apply here, as can the idea of making the first 2-3 words really count.
How influential do you hope this report to be, will it be expanded and how?
I hope it will start a conversation in the industry. I’ve heard from many people that they still look to the first online Eyetrack study (1999-2000) for guidance even today, so I hope this will be seen as useful not only now but for the next couple years.
We tried to go wide on this study, hoping that it sparks additional, more focused research. I think there are some things that we touched on with this study that could be looked at deeper with later, more focused eyetracking. And I hope that the industry tells us what it wants to know more about. For instance, we didn’t do anything on hidden content/dhtml techniques (flyout navigation and mouseover-revealed page content), and I’d love to see that get researched. We just lightly touched on navigation placement. (We tried to cover a lot of ground, and couldn’t fit in everything.)
How this will play out I don’t know. If the industry finds value in this and wants more depth, I’m sure we’ll figure out a way to do more. Poynter has no immediate plans beyond this, but all along we’ve considered this as a way to get the conversation started about how eyetracking might further our aims as an industry and let us better understand how people interact with news websites.
Personally, I hope other organizations start doing more eyetracking research, not just Poynter and its partners. Getting rid of the headgear for test subjects is a big deal in my mind, as it makes the testing experience more applicable to real life. So it’d be great if the industry takes advantage of this jump in technology.
For more about the study, see:
• Official Eyetrack III web site
• Track links to Eyetrack on Blogdex, Technorati
• PDF of Eyetrack III research report (11 MB)
• What News Websites Look Like Through Readers’ Eyes by Steve Outing and Laura Ruel
• When It Comes to Homepages, It Is Polite to Stare by Jay Small
• When, How to Tell Stories with Text, Multimedia by Jeff Glick
• From Banner Blindness to Text Ads, Placement and Size Matter by Kinsey Wilson
• Eyetrack III is a gold mine for web designers by John Zeratsky