Author: Joe Goldberg, Principal Research Scientist – Oracle Applications User Experience
Mar. 18, 2009

“Eye tracking” is an important new usability research tool that enables Oracle Applications User Experience team members to determine where a user’s visual attention is focused during online tasks. This information helps drive interface design decisions, so that tasks can be completed more quickly and with less confusion. New analysis tools allow application designers to improve and optimize designs to accommodate visual search tendencies.

Joe Goldberg

Although relatively new to enterprise software companies, eye-tracking methods have been routinely used by marketing, television, and movie consulting firms to address issues such as design impact, confusing motion edits, and clarity of branding. Companies such as eBay, Microsoft, Google, and Facebook routinely use eye tracking to understand how page layout is visually navigated by end users.

What Does Eye Tracking Look Like?

Eye-tracking visualizations provide compelling summaries of a user’s visual attention. This first image shows a Las Vegas skyline with aggregated eye-tracking data across many users. The representation is known as a heatmap, with the red areas indicating screen locations that generate greater visual interest and longer dwell times than other areas. The second image shows an opacity heatmap, which only exposes image areas that have been viewed, where longer viewing results in more translucent image areas. The third image shows a user's visual scanpath. The circles indicate fixations, where the eye dwelled for a short time; larger circles indicate longer fixations. The scanpath tells a story about the user’s task-completion strategy, from first looking at the screen until the task is completed. A three-dimensional heatmap is shown in the fourth image, providing contours instead of a heatmap.

Red indicates where there was more visual interest in this photo showing eye-tracking data of the Las Vegas skyline .	In this gaze opacity heatmap, eye-tracking data from a single user is shown. The most viewed areas are the clearest and brightest.
In the same image, but with a visual scanpath recorded, the circles indicate locations where the eye dwelled briefly. Larger circles show were the eye spent more time.	In this three-dimensional view of an eye-tracking dataset, higher peaks are associated with a greater number of fixations.
	Photos courtesy of Joe Goldberg, Oracle Applications User Experience

These data visualizations serve different purposes for the user interface designer. The scanpath data visualization retains sequential or temporal information as a user scans an interface, allowing the study of user strategies while solving tasks. Various models can be developed to predict where the user will look next. The heatmaps excel at aggregating visual attention location from many users over a period of time. The time window over which the aggregation occurs can also be manipulated, allowing an assessment of temporal changes in visual attention.

How Did Eye Tracking Start at Oracle?

In this modern eye-tracking system, infrared cameras are housed just below the display on the monitor.

Dramatic advances in technology have made eye tracking a routine and non-intrusive measure of software usability. Eye tracking started in the late 1800s as a tool to help understand how people process the visual environment. In those days, the technology required people to wear tight-fitting lenses with attached sticks that made marks on smoked paper in response to eye movements. Mid-20th- century methods placed surface electrodes on the skin to measure small electrical changes as the eye moved horizontally and vertically. Current methods reflect infrared light off the surface of the eye into an infrared camera that is hidden behind a panel on a user’s computer display. Following a 10-second calibration procedure, the user is typically unaware that equipment is measuring his eye location on the screen.

Oracle initiated eye tracking as a usability method in 2001. Oracle began to use new eye-tracking technology in late 2008 to support the evaluation of Fusion designs. The system is portable, and has recently been taken to user group conferences and Oracle OpenWorld, to evaluate various usability issues.

How Does Eye Tracking Inform the Design of End User Application Experiences?

The computer interface can be considered a workspace where tasks are completed and events are monitored. Just like a physical workspace, users are more productive and tasks are completed more quickly if objects are located close together, and if they are easily visible. A software application that results in shorter and more direct scanpaths typically has a better interface than one with highly disorganized scanpaths. Also, unclear visual navigation is marked by dramatically differing scanpaths across users, determined by a nearly equal likelihood of viewing any portion of a page at any given time.

Superimposed Areas of Interest, or AOIs, are indicated by varying colors in this image, and provide a way to organize eye fixations and their transitions.

Certain issues must be considered when designing an eye-tracking study. As an example, the direction of one’s gaze usually corresponds to what is on one’s mind, but when fatigued or informally browsing, the gaze may lead or lag one’s current thoughts. To help avoid this, eye-tracking studies should present focused tasks to users.  In addition, using a mouse may influence eye-tracking studies. The eye may follow the cursor in some cases, so when a mouse is not absolutely required, the cursor is typically not shown.

Data analysis methods that can help to recommend design improvements are still under development. One method of analyzing eye-tracking data requires an indication of regions that are of most interest to the designer or developer. These areas of interest (AOIs), shown below, can be used to tally statistics such as:

Designers can use scanpaths, heatmaps, and AOI statistics to tell the story of what the eyes are doing while a user completes tasks. These stories, in turn, can be aggregated across other users and/or tasks to discover general trends. One trend that has recently been publicized, for example, is that North American users who view a text-heavy page for the first time tend to initially scan the left-hand column, followed by horizontal scans at the top of the page, then a shorter horizontal scan near the middle of the page. This same trend would not be expected for those who are more familiar with the page content.

What’s Next at Oracle
We are developing software that will look for patterns in the scanpaths of different users and/or user interface conditions. The software could show, for example, the most popular search strategies while solving a task.  Understanding the scanning strategies across users can help to find those interface features that affect user productivity while completing common tasks.

While eye-tracking methods are not a substitute for usability testing, they can inform the interface designer about unclear or distracting features on an interface, and can help determine why certain errors are made while completing tasks. Eye tracking is rapidly becoming an additional, compelling tool in Oracle’s usability evaluation arsenal, to make Fusion interfaces as productive as possible.