Eyetracking Web Usability: review

Time to pick sides: Jakob Nielsen has written an eyetracking book. I can scarcely think of a more divisive pairing: mention either within earshot of a UX aficionado and you’re in for impassioned advocacy or scornful ridicule. Me? I’ll confess both subject and author have left me unconvinced in the past, but I approached Nielsen and co-author Kara Pernice‘s new book with curiosity and as objective an outlook as I could muster.

Eyetracking Web Usability is the outcome of the largest eyetracking study ever undertaken: 1.5 million fixations from 300 participants. Nielsen and Pernice are clearly keen to stress the magnitude and legitimacy of their research. Their test script, posted in full, is well considered and comprehensive, covering a range of tasks representative of real web use.

After a brief recap of eye physiology and saccades, the book begins in earnest with a detailed breakdown of research methods. Findings then stretch across chapters discussing specific web elements in turn: navigation, forms, images and so on. At their best, these chapters reveal flashes of usefulness. A chart of eye fixations versus layout density shows minimal correlation, demonstrating that busy pages simply dilute attention from the most important information. The book also touches on the important role of information scent and microcopy, declaring insightfully that “a link is a promise”.

In typical Nielsen style the text is heavily punctuated by summary boxes. Sadly, it quickly becomes apparent that these make the point just as effectively as the full text. Eyetracking Web Usability is all fat, no meat. Wasted space includes a page on why a 7-point Likert scale is better than a 5-point one, and five pages on male users’ propensity to fixate on dog genitals. The writing, meanwhile, veers from redundant to simply cringeworthy: “Give that Wii a rest, and go prioritise your Web page layout design. You can do it!”

A chapter on adverts (whose raison d‘être is of course to attract the eye) starts promisingly. An ad has a 36% chance of being seen by a user, a figure surprisingly unaffected by user task. However, it soon descends into known generalities: banner blindness and users’ dislike of irrelevant advertising. The chapter encapsulates Eyetracking Web Usability’s main shortcoming. Eyetracking demands specificity: carefully planned tasks on an individual site. Nielsen and Pernice’s 300-person test can only dilute potentially salient points into generalisations that even a novice designer will already know. The conclusions cover ground so well trodden as to be barren.

Despite the authors’ focus on rigour and transparency, serious concerns surround the research methods themselves. Heatmaps from the tests are dated from late 2005. With lab time accounting for five months, the study was therefore complete by summer 2006. Why then was this book not published until the brink of 2010? It is hard to avoid the impression that the results sat untouched for years and were subsequently rushed out in a lull of client work. Eyetracking Web Usability also misses a huge opportunity by focusing solely on informational websites. Web apps are discounted since eyetracking can’t handle dynamic elements, including Ajax and even dropdowns. The results are thus only valid for an increasingly small part of the UX designer’s 2010 workload.

Most worryingly of all, it seems that the tests were conducted in Internet Explorer 6. Browser choice does not appear to have been offered to users, and where browser chrome is shown (it is stripped in the vast majority of the heatmaps), it is unmistakeably IE6. If this is indeed the case, it nullifies many findings since the primary browser innovation of the 2000s – the tab – is unavailable. In IE6 a link is an entirely binary choice: go there, or stay here. Modern browsers allow an important new behaviour: Open In New Tab, creating tentative and plural navigation steps. It’s likely Nielsen’s participants relied far more on the Back button and their short-term memory than today’s users. Their search engine use is also likely to be different, since IE6 lacks an inbuilt search box in the UI.

Eyetracking Web Usability thus lacks the rigour required to be taken seriously as an empirical work; however, its adherence to factual reportage make it a chore to read. Even the most ardent enthusiast will skip over paragraphs that merely disclose participant actions in minute detail. It’s sixth form science at best; utterly literal, over-eager for the praise of the adjudicators. The effect is exacerbated by the disappointingly scant acknowledgment of others’ work. Few external insights or breakthroughs are admitted, although NN/g reports are of course suggested as ways for the reader to supplement his knowledge.

The book’s conclusion will come as no surprise to the reader. “Eyetracking fills in the details… Most companies should not bother conducting their own eyetracking studies.” It is hard to disagree. The book does nothing for the eyetracking industry except cement its status as an expensive diversion; the excessive cover price of £44 only reinforces this. If this is the accumulated wisdom of the largest eyetracking survey in history, we can safely consider the technology inconsequential.

Remember those design principles you learned ten years ago? Eyetracking shows they’re right. Carry on.

Cennydd Bowles