Eyetracking Web Usability – Book Review

Eyetracking - Book Review

Jakob Nielsen, Ph.D., one of the world’s leading experts in Web usability and Kara Pernice, the managing director of the Nielsen Norman Group, conducted unprecedented research in Web usability using eyetracking and published it a new book simply called “Eyetracking Web Usability.”  I was fortunate enough to have attended a usability conference 2009 in which both Nielsen and Pernice were presenters.  Until then, I was unaware that any technology of this sort existed.  This book speaks to the developer, the designer and also, thankfully, to the corporate executive.  Much of the research has been assembled from 2006 and this book was recently published in its first edition in November of 2009.

Example of an eyetracking computer

So what is usability? Usability.gov defines it as “Usability measures the quality of a user’s experience when interacting with a product or system—whether a Web site, a software application, mobile technology, or any user-operated device.” This definition, although broad, pretty much describes this type of research.  Until recently (within the last ten years) the notion of taking usability studies one-step further and incorporating eyetracking research wasn’t possible.  Previous studies involving eyetracking relied on heavy medieval looking contraptions that were both distracting to the subjects and difficult to calibrate.  Now eyetracking studies (related to computer use) can be done by sitting at a desk with the subject using a computer application or viewing a webpage while infrared cameras surrounding the computer can pick up eye-movements and compensate for head displacement.  This, in a word, is remarkable.

Figure 1 - Gaze Plot

Figure 2 - Heat Map

But how does it all work? At the beginning of the book, Jakob Nielsen and Kara Pernice lead us through a brief introduction about eyetracking metrics.  These metrics rely on two types of eye movements: fixations and saccades.  Fixations relate to what the user is looking at (in focus) and saccades are the transition movements from fixation to fixation. Looking at the data in form of Gaze Plots and Heat Maps one can quickly see the plethora of data available from a few simple sessions on eyetracking. In the figure (figure 1) you can see a series of fixations and saccades on a Gaze Plot.  The size of the dots relate to the time the user was fixated (focused) on the spot.  The numbers on these dots (fixations) correspond to the order of the “looks.”  The lines between the dots are the saccades.  These lines actually note the eye movement and thus are out of focus for the subject.  Out of focus therefore implies that the user is unaware of the content of these areas unless you count peripheral vision. The aggregated data from these gaze plots help to create a heat map (figure 2) that shows where the test subjects, as a group, fixated the most.

So here is the obvious question.  In a class that focuses on online communities, why am I reviewing a book about Web usability and eyetracking? User experience is important regardless of what the content is.  If your user is coming to your community and you want to encourage participation and engagement you need to make it easy.  More importantly, you want the user to not only be successful but have their experience score high in satisfaction.  The point being is that the site may have the content/community the user is seeking but if the user is unsuccessful, unconfident or unsatisfied, they may just seek out the competition.

So here is what I learned.  Perhaps a few of these points seem painfully obvious, based on my years of experience in development and production these mistakes are innocently made on many Web-based projects.  What is carefully outlined in this book, both and the beginning and the end, is that the subjects (users) were given a specific task (either topic specific or general browsing) and the data was collected by eyetracking machines and vocally.  The researcher never spoke. They only recorded the session.  A research study wouldn’t be complete without a clear understanding of what the metrics are for the study.  Quantitative measurements were done based on task time, success, errors and user satisfaction.

Probably the biggest and most interesting finding is the phenomenon of Banner Blindness.  Banner Blindness is a term to describe the users complete disregard for advertisements on the Web. Only until this type of research was available did this evidence present itself.   The book alludes to the notion that this is a conditional effect from previous of Web browsing.  The usability angle is to make sure that content is not presented in a way that could be subconsciously presumed to be a banner advertisement.  The census example is an excellent illustration where many test subjects when asked to find the population of the U.S. completely overlooked the obvious.  Why?  Well there are many reasons the book suggests.  One is the location of the information.  The “right rail” is often used for targeted advertisements.   Also, miss-labeling of certain features of website and information congestion can also lead to problems.  I found it interesting that interactive components like dropdown menus and search fields are high visible attractors for users and consequently prevent users from seeing important information.  The book calls these magnetic elements.

Figure 3 - population clock no fixations - useit.com

Navigation. Simply put, this are the roads and bridges that you build so your users can easily travel through your website.  The placement of these buttons on your website are important and how they are labeled are even more important.  The study found that people are twice as likely to look for navigational elements on the left side of the page or to the top at subnavigation buttons than looking at the global navigation elements.  The authors attribute this to a lifejacket scenario in which the users subconsciously know that there is always a safe way back. Therefore this assurance that the global navigation exists at the top eliminates any reason for a fixation on those navigational elements.  The book also makes a point of noting that one should pay specific attention to menu/navigational wording and arrangements since this can help users feel included or worse it can actually repel or insult visitors.

How important are the elements on a page when people first open a page? The book looks at seven items that were based on rapid fire tests: Login link or field (26%), Logo(22%), search(16%), language selectors(14%), banner self-promotion(7%), contact link(2%), privacy policy(1%).  The study attributes the large number of fixations on the login fields/links and the search component to the magnetic attraction to open fields.  Logos are also described as a sort of anchor or reassurance that the user has landed on the right page.  The importance of that logo’s consistency in placement and size throughout the site is described as an insurance policy for the user.  Although rarely does it get a second look as the user navigates the site.  I found it also very interesting that users have been conditioned on element placement.  For example when looking for shopping cart information:  users looked in the upper quadrants for shopping cart indicators (42% left top, 58% right top).  In my opinion, this is purely due to a standard that has been adopted and made a pseudo rule.  This, like the search location in the top of a page, has now become instinct to users.

The research on images presented in the book is very interesting and is a topic that is detailed in 25% of the book.  The data presented pertains to specific tasks the user is instructed to do. Images can be treated in many different ways.  One way is as a visual obstacle course where users look at the information around the images but not at the images themselves.  Images should pertain to the content (obviously) but one interesting note is that people ignore stock images 85% percent of the time.  Perhaps this is just another example of subconscious quality control that the user is invoking.  Interesting images always win.  Images of attractive people who also look real and authentic (non-models) also get many looks.  Also worth noting was the gender differences when looking at images.  The book reports that in studies where there were images of men with clothing on sites, men looked not only at the clothing but also at the face of the male model. Women only looked at the clothes.

Almost all of our Web browsing visits are sprinkled with advertisements.  Whether it is searching through Google or looking through our favorite shopping site.  Once you are aware of these statistics with respect to advertising you will begin to notice why methods like Google Ads are so successful.   All of the following data is presented based on percent fixation corresponding to the type of ad.  After all, the biggest part of successful ad campaign is getting looked at.  88% of internal promotion that matches the site style gets first place.  This is more like the deals or specials of the site.  Text ads get 52% of the fixations.  These plain text ads, the book suggests, has a visual relationship to Google in that Google is a trusted name and their information is almost always presented in a plain text format. 52% of ads that have clear text separate from graphics (i.e. not having the text running over the image). 51% of ads on search engine results pages get looked at as well.  Only 35% of pure graphical ads even get a glance.  Finally our favorite, the animated graphic ad only gets 29% of fixations.  The location of ads is important as well.  Placing them on the right hand side in a website gets the most fixations compared to the top of the site.  In a SERP (search engine response page) more fixations are gleaned from the top then from the right side.

This book is a great read with lots of illustrations, a touch of humor and enough research data to back up the findings.  It really is designed for a broad audience –both industry professionals and hobbyists.  Although much of the data seems to be common sense, the reality is that as a Web developer and Web producer I still had many “Ohhhh” moments.  None of this information can be taking out of context.  This is what makes this book so interesting.  The test subject’s tasks and their vocal testimonial is the key to this book’s success.  This book provides ammunition in backing up design decisions to corporate suits and also gives producers insight when talking to designers, developers and the top tier of a corporation.

An eyetracking study is not for everyone nor is it needed. The book clearly states this in the closing chapter.  In general, it provides us with some behavioral data and it is available for use when good old fashion Web usability studies just are not providing answers.  Thank you Jakob and Kara.