research methods
Eye tracking
What is eye tracking, and what can it tell you?
Eye tracking records where people look on a screen, and for how long, producing gaze paths and heatmaps of attention. It answers a question self-report can't: not what people say they noticed, but where their eyes actually went - revealing what got seen, what got skipped, and the order it all happened in.
Also known as: eye tracking, eye-tracking, gaze tracking, attention mapping
Prefer to watch?
Watch the recap 0:48
The demo
A landing page. Before anything else: click the one spot you think a visitor's eyes land on first. Then reveal where attention actually tends to go.
What this demo shows (text version)
A sample landing page has a logo, a hero with a smiling face, a big headline ("Move money in seconds"), a line of body text, and a call-to-action button. You first click where you think a visitor looks first; the demo then reveals an illustrative gaze heatmap. Attention concentrates on the face and the headline, with the body text and especially the call-to-action getting far less - even though the CTA is the thing the business most wants seen.
That mismatch is what eye tracking exposes: attention goes to faces, large type, images and movement, and skips ad-like or sidelined elements, regardless of what a team intends. It's great for "did they even see it?" but it only shows what was looked at, not whether it was understood - so pair it with think-aloud, and don't mistake a fixation for comprehension. (The heatmap here illustrates well-known tendencies, not measured data.)
Eye tracking shows where attention actually lands - and it's reliably not where teams assume. Eyes go to faces, big type, images and movement; they skip ad-shaped boxes and things in "ignored" zones. It's powerful for "did they even see it?" questions, but it only tells you what was looked at, not why or whether it was understood - so pair it with think-aloud, and remember attention isn't the same as comprehension.
You guessed where people look first - and the heatmap probably surprised you. Attention floods the face and the headline and barely touches the call-to-action you cared about. That gap between where you put the important thing and where eyes actually go is exactly what eye tracking exposes - and why "but it's right there" is so often wrong.
Attention follows predictable pulls: human faces (and the direction they're looking), large high-contrast type, images, motion, and the natural scan patterns (the F and the Z). It actively avoids things that look like ads (banner blindness) and anything parked in a low-traffic corner. A heatmap makes "is this being seen?" answerable in a way no survey can.
Its big limitation: looking is not the same as understanding. Eye tracking tells you an element was fixated on, not whether it was read, comprehended, believed or acted on - someone can stare at a label and still be confused. So it answers "did they see it / in what order" far better than "did it work", and is strongest paired with think-aloud for the why.
It's also become more accessible: dedicated hardware gives precise data, but webcam-based and predictive "attention" tools now approximate heatmaps cheaply, and even a quick first-click or five-second test gets at the same "what gets noticed first" question without a lab. Use the method that fits the stakes - and treat all of them as evidence about attention, not proof of success.