As you know, I originally planned to test eye tracking during the poster exhibition Critical Imagination. The idea was simple: visitors would look at the posters, the eye tracker would record what caught their attention, and afterwards I would compare the data to the design intentions behind the posters. Which elements were seen first? Did people read the text or only scan the visuals? Did the composition guide the viewer, or did important information disappear into the background?
In theory, it would have been a very useful setup. Poster exhibitions are all about visual attention. A poster has only a few seconds to invite someone in, communicate an idea, and make them stay long enough to understand more. Eye tracking seemed like the perfect method because it can show gaze paths, fixations, heat maps, and areas of interest. Instead of only asking people what they remembered, I wanted to see how they actually looked.
But on the day of the exhibition, the eye tracker did not work.
At first, this felt like a failed test. The whole method depended on the device, and without it there was no automatic data, no heat map, and no clean visual proof of attention. However, after the initial frustration, the situation turned out to be more interesting than expected. Even if the eye tracker had worked perfectly, the testing would still have had a major problem: not enough people from the intended target groups attended the exhibition. That means the collected data would probably not have been representative enough to support strong conclusions.
In that sense, the technical failure exposed a bigger methodological issue. A working tool is not the same as a good study. Eye tracking can produce impressive-looking data, but that data only becomes meaningful when the research setup is strong: the right participants, a clear task, a controlled process, and enough observations to compare patterns. Without that, a heat map can look scientific while still saying very little.
This made me rethink what “testing” a poster can actually mean in an exhibition context. A gallery is not a lab. Visitors arrive with different motivations. Some come because they know the designers, some because they are interested in the topic, and some may simply pass through. They talk to each other, move unpredictably, stand at different distances, and may spend more time in front of one poster because a friend is standing there. These behaviours are not errors; they are part of the exhibition experience.
So instead of seeing the failed eye tracking test only as a problem, I started thinking about alternative ways to understand attention manually.
One option would be timing and tracking observation. This method is often used in museum and exhibition research. A researcher discreetly follows or observes a visitor and records where they go, where they stop, and how long they stay. For the poster exhibition, I could have created a simple floor plan of the space and marked each visitor’s route. Every time someone stopped in front of a poster, I could record the approximate dwell time. This would not show exact eye movement, but it would show which posters attracted people enough to physically stop.
Another option would be a manual “attention map.” After viewing the posters, visitors could be asked to mark the area of a poster they noticed first, the part they found most memorable, and the part they found confusing. This could be done with transparent overlays, sticky dots, or printed miniatures of the posters. It would not be as objective as eye tracking, because it depends on memory and interpretation, but it would create a participatory version of a heat map. In some ways, that might even fit the theme Critical Imagination better, because visitors would actively reflect on their own perception.
A third option would be short think-aloud sessions. Instead of silently looking at a poster, participants would be asked to describe what they notice while viewing it: “I see the image first,” “I am trying to understand the headline,” or “I do not know where to look next.” This method is not perfect, because speaking changes the viewing experience. Still, it can reveal confusion, expectations, and design interpretation in a way that eye tracking alone cannot. Eye tracking might show that someone looked at a title for five seconds, but a think-aloud comment can explain whether they were interested, confused, or trying to decode the message.
Video observation could also be useful, as long as it is done ethically and with permission. A camera placed at a distance could record visitor movement, posture, and stopping behaviour. For example, it could show whether people leaned closer to read small text, whether they skipped certain posters, or whether groups discussed specific works. This would not track the eyes, but it would track bodily attention. In a physical exhibition, that kind of attention matters.
The most realistic future approach might be a mixed method. Eye tracking could still be used, but only in a smaller, more controlled session outside the exhibition. Participants from the intended target groups could be invited separately, shown selected posters, and asked to complete a structured viewing task. Then, the exhibition itself could be studied through observation, dwell time, informal interviews, and visitor mapping. Together, these methods would give a fuller picture: eye tracking for precise visual behaviour, observation for real exhibition behaviour, and interviews for interpretation.
The failed test also reminded me that research methods should not become more important than the research question. I wanted to know how people engage with posters. Eye tracking is one way to approach that, but it is not the only way. In fact, the absence of the technology made the human side of the exhibition more visible. Who attends? Who does not attend? Who feels addressed by the posters? Who walks past? Who stops, reads, discusses, or ignores?
Those questions may be harder to measure than gaze points, but they are just as important.
In the end, the eye tracker not working did not completely ruin the project. It changed the direction of the reflection. It showed that testing visual communication is not only about capturing where people look. It is also about understanding the conditions under which looking happens. A poster does not exist in isolation. It exists in a room, in a social situation, in front of real people with different backgrounds, expectations, and levels of attention.
Next time, I would still like to use eye tracking. But I would not rely on it as the only method. I would prepare a backup observation sheet, define target groups more clearly, recruit participants in advance, and combine technical data with manual and qualitative methods. That way, even if the device fails again, the research does not fail with it.
Maybe that is the most useful insight from Critical Imagination: sometimes a broken tool forces a better question.
Sources
Tobii. “Understanding Tobii Pro Lab’s Eye Tracking Metrics.”
Tobii. “User Testing with Eye Tracking Goes Mobile.”
Nielsen Norman Group. “Thinking Aloud: The #1 Usability Tool.”
Nielsen Norman Group. “Usability Testing 101.”
Yalowitz, S. S., & Bronnenkant, K. “Timing and Tracking: Unlocking Visitor Behavior.” Visitor Studies, 2009.
University of Cambridge Museums. “Visitor Observation: Impact Evaluation Resources.”
Yuan, S. “An Eye-Tracking Study of Attention Capture Efficiency in Commercial Poster Design.” Applied Sciences, 2025.
MeasuringU. “Essential Eye-Tracking Visualizations and Metrics.”
