How to run eye tracking tests on mobile

Anyone who has used a smartphone knows the flip side of having millions of apps at hand: an experience ruined by a poor interface. That is why product and user experience (UX) teams usually go through several rounds of testing, interviews and analysis — with many iterations — before releasing something genuinely usable. Eye tracking reveals what people look at, in what order and for how long while they use an app or a site on their phone. But measuring attention on a phone brings its own challenges. Here is why, and how they are solved.

Why a phone stand is necessary

A screen-based eye tracker assumes the sensor sits below the screen and that the participant is at a certain distance. With that information it computes the angle the eyes form when looking at different parts of the screen. Using that same hardware on a phone — rather than a computer — means solving two problems:

• Hands. On a phone we use our hands to swipe and interact. If the sensor sits below the device, hands block it and data is lost. That is why the phone goes on a stand that keeps the sensor in the right position, unobstructed.
• Dimensions and angle. A phone is smaller and is held at a different angle than a laptop. The system factors that angle into its gaze calculation, so controlling the sensor position, angle and height relative to the screen is essential for accurate measurement.

The stand strikes the right balance between controlled experimentation and ecological validity: you can test mobile content in the environment it was designed for, without compromising data quality.

How to analyze behavior on mobile

Infinite scroll produces an endless stimulus that is hard to aggregate: each person scrolls at a different speed, stops at different points and takes different paths to complete a task. There are individual differences in what captures attention and what holds it. Three tools make rigorous analysis possible:

Dynamic areas of interest (AOIs)

They track elements that move with the scroll, following each participant’s path. That makes it possible to isolate what gets looked at and what gets ignored, even when the content never sits still on screen.

Gaze mapping

It maps gaze — which moves as the user scrolls — onto a reference image (a capture of the whole document, for instance) using computer vision. That aggregates data from many participants into a single heatmap, even across long stimuli like a website or a feed.

Multimodal analysis

To understand not only qué they look at but what they feel, eye tracking is combined with facial coding (a camera that reads expressions) and GSR (a sensor on the hand not used for swiping). The result is a fuller reading: attention and emotion — valence and intensity — in front of the content, at the same time.