Eye-Tracking in Advertising Research

A fixation is a brief period during which the eye holds relatively still over one location and visual information is encoded; a saccade is the rapid jump between fixations, during which vision is largely suppressed. Attention and information uptake happen during fixations, so all marketing-relevant metrics, dwell time, time-to-first-fixation, and refixation counts, are computed over fixations rather than raw samples. The classification step uses spatial and temporal thresholds to separate the two, and getting those thresholds right is essential because mislabeling saccades as fixations would inflate apparent attention. Wedel and Pieters treat this fixation-based parsing as the foundation on which the validity of all downstream attention measures rests.

No. Eye-tracking measures attention, the gateway to processing, but not comprehension, emotion, or persuasion. An element can receive long dwell because it is engaging or because it is confusing, and a glance can register a brand without changing any attitude. The method's value is to establish whether the right elements are noticed and in what order, which is a necessary but not sufficient condition for impact. To connect attention to outcomes, researchers model recall, attitude, or choice as a function of the attention metrics, and in practice combine eye-tracking with self-report, facial coding, or implicit measures so that attention, emotion, and persuasion can be disentangled rather than conflated.

Each respondent's fixations are first assigned coordinates and durations on the stimulus. To build a heatmap, a Gaussian kernel is centered on every fixation and weighted by how long the fixation lasted, then all kernels are summed across fixations and across respondents to form a continuous attention surface over the image. Hot regions indicate where many or long fixations accumulated; cool regions indicate neglect. Gaze plots and scanpaths complement heatmaps by preserving the temporal order of fixations. These aggregations make the data legible to design teams, but they should accompany, not replace, AOI-level statistics, since a heatmap shows where attention fell without testing whether differences between design variants are reliable.