Research
The question started from a practical problem: when a new website or landing page is ready, teams want to know whether visitors understand it, where attention goes, and whether the interface supports the intended action.
Common answers tend to come from best practices, opinions, surveys, user observation, cursor tracking, or eye tracking. The specific claim tested here was sharper: can cursor tracking work as a low-cost substitute for eye tracking?
Central finding
Cursor tracking and eye tracking are not interchangeable. Cursor position has a weak positional relationship with gaze, but it does not capture enough of the same information to be treated as a proxy for attention.
What was tested
- Whether cursor position and eye position correlate during website use.
- Whether cursor and gaze movement patterns behave similarly.
- Whether adding more cursor-tracking users can compensate for lower measurement detail.
Correlation is not enough
The cursor and eye position showed a statistically significant correlation of r = 0.22. That sounds promising at first, but the figure matters: the relationship is mostly positional, not movement-based. If cursor movement were a strong substitute for gaze movement, speed and movement patterns would align more clearly. They did not.
Distance and coverage
If cursor position were close enough to gaze to support design decisions, the average distance between the two should be small. In this study it was not: the average distance was 9.39 cm.
The same pattern appeared when time was removed from the comparison. Cursor and gaze could theoretically visit the same places at different moments, but the heatmap overlap was only 2.3 percent.
The scale argument
A common argument for cursor tracking is scale: if it is cheaper, perhaps more users can make up for the loss in measurement quality. The entropy analysis tested that idea directly.
Both methods gain information as more users are added, but they plateau at different levels. Eye tracking reaches about 6 bits, while cursor tracking plateaus around 3.5 bits. In practical terms, cursor tracking does not catch up simply by adding more participants.
What each method can support
Cursor tracking
Useful for interaction traces, click intent, scroll behavior, and interface friction. It can show how people use a page, but not reliably where they look.
Eye tracking
Useful for visual attention, gaze allocation, and whether critical information is noticed. It is harder to collect, but measures a different behavioral signal.
Conclusion
Cursor tracking is not the poor man's eye tracker. It can still be useful, but interpreting cursor data as loose gaze data can point teams toward false optimization decisions.
Publication details
- Author: Adjmal Sarwary
- Publication date: 2017-10-16
- Format: research paper / method comparison
