What an eye-tracking lab teaches you about your product.
We started running our own usability sessions with a Tobii eye tracker eighteen months ago. The lessons changed how we design — here's what we got wrong, and what we now do on every project.
Eighteen months ago we set up an in-house usability rig with a Tobii eye tracker, a one-way camera, and a recording bench. The plan was to use it for one client project. The data was so much richer than we expected that it's now part of every product engagement we run.
What eye tracking actually shows you
- 01Where users look first on a new screen, and for how long.
- 02Which elements get scanned and which get ignored entirely.
- 03How the eye moves between content blocks — usually not the order the designer assumed.
- 04Where users get stuck — repeat fixations on the same element are the giveaway.
- 05When a page is overwhelming — saccades become erratic and the user disengages.
What we keep getting wrong
- 01Hero images don't get looked at as long as designers think. Most users glance and move on.
- 02Calls to action that look like 'design' get treated as banners and ignored.
- 03Three-column layouts force the eye to choose — and users often pick the middle column and skip the rest.
- 04Long-form copy is scanned, not read. The first sentence of each paragraph carries 80% of the perceived information.
- 05Forms with visible required-field stars don't reduce errors — they raise abandonment.
What we now design for
| Insight | Design change |
|---|---|
| Hero images get a brief glance | Headline does the real work; image is reinforcement, not the message |
| CTAs that look like design get ignored | CTAs look like buttons — solid, contrasting, button-shaped |
| Three-column UIs lose the side columns | Default to one column above the fold; two on wider screens |
| Paragraphs are scanned, not read | First sentence carries the argument; rest is supporting |
| Required stars raise abandonment | Show "required" only when missing, not as decoration |
How we test now
- 01Six users, in-person, per major version.
- 02Same script across sessions, same tasks, same probe questions afterwards.
- 03Recorded gaze plot reviewed by the design lead within 48 hours.
- 04Quantitative findings (time-on-task, error rate) tracked across versions to prove improvement.
Six sessions catch problems that hundreds of hours of internal review miss. The data closes the loop between 'we think this works' and 'we know this works'.
The lab is among the most valuable studio investments we've ever made. Six sessions catch problems that hundreds of hours of internal review miss. We'd recommend any studio building products to set one up — Tobii's gear is genuinely good, and the data closes the loop between 'we think this works' and 'we know this works'.