If you perform a Google image search for ‘eye tracking,’ your results will consist primarily of heat maps – heat maps of webpages, heat maps of advertisements, heat maps of grocery store shelves, heat maps, heat maps and more heat maps. They are the most recognizable eye tracking analysis tool. They are the most commonly requested eye tracking deliverable. At this point, it isn’t too much of a stretch to say that the heat map has become the logo for the eye tracking industry as a whole.
However, this post will not be another puff piece about the unmitigated value of this oft-used data rendering. EyeTracking, Inc. will toot its own horn just this once to say that we were the originators of the heat map (or GazeSpot as we call it) back in the 1990s, and then we will proceed to a more objective discussion. What we’d like to talk about today is the manner in which these graphics are misused and misinterpreted. In doing so, we hope to shed some light on what gets lost in the heat map.
Take a look at the example on the right. This GazeSpot shows the aggregated visual behavior of ten users interacting with the Services page of eyetracking.com. Over 7,000 data points are represented here, and yet it doesn’t tell the whole story. Where is the eye drawn first? Is there a pattern in the way users move between elements of the page? How long do they stay here before clicking away? What usability problems are encountered? Did one user’s atypical viewing habits unduly influence the rendering as a whole? No matter what you may have heard, none of these important questions can be answered by the heat map alone.
And what about the pictures in our example? One of the most common misinterpretations of heat maps is the assumption that a particular non-text element was not viewed because it does not have an associated hot spot. Actually, the pictures on the page shown here were all viewed by all users. The reason that they don’t show up as hot spots is that it takes much longer to read a paragraph than it does to view an image. Thus, the impact of each user’s glance toward the picture grows more diluted with each second spent reading the text. As you can see, interpretation is not always as straightforward as it seems.
This is not to say that the heat map has no value. In fact, we use them quite often in all kinds of different studies – websites, packages, advertisements, applied science and more. They are both elegant and intuitive as a means of demonstrating the total amount of attention allocated to specific features of a medium. However, attempts to apply them to deeper research questions are misguided. Any expert in the analysis of eye data will tell you that heat maps serve a precise purpose, one that should not be stretched too far.
In our experience, there is no graphic deliverable that really tells the whole story of visual behavior. That’s why we use a range of different ones – GazeTraces, GazeStats, GazeClips, Bee Swarms, GazeSpots and Dynamic GazeSpots (which are video heat maps with the added dimension of time). All of these deliverables are integrated with statistical analysis of the data, as well as traditional marketing research and usability measures to fully describe the interaction with our test materials. That’s the approach that we recommend for any comprehensive eye tracking study – use all of the tools at your disposal. While there are many fascinating results to be found in a heat map, if you aren’t careful how you use it, you might just get more lost.