How Eye‑Tracking Reveals Hidden User Behaviors in Short‑Video Apps

Why Eye‑Tracking?

The rapid rise of short‑video platforms has dramatically changed how people consume information, with users spending nearly two hours daily on short videos. Traditional surveys and interviews often miss fine‑grained behaviors, so the team introduced eye‑tracking experiments to capture invisible user actions before launching a new UI.

Research Objectives

Two main factors were examined:

Effect of title position and interaction area (like, comment, share) placement on video browsing.

Effect of different recommendation styles on browsing behavior.

Experimental Design

The study used Tobii Pro Glasses 2 in a controlled lab. Participants freely browsed videos for 2–3 minutes, then completed a click‑task. After eye‑tracking, they answered subjective preference questions.

Evaluation Dimensions

Five dimensions were measured: significance, interest, convenience, comfort, and immersion. Both eye‑movement metrics and subjective ratings were collected, while physiological metrics such as pupil diameter were deemed unreliable for micro‑design differences.

Materials and Controls

Eight fully functional SDK test packages replicated the live app experience. The test set included 30 high‑quality entertainment videos, a standardized Huawei P30 device, and consistent screen brightness. Variables such as practice effects, fatigue, and lighting were tightly controlled.

Participant Selection

Old users of the app were targeted, with balanced gender and age ratios. Participants also had experience with competing apps and used the target app at least once per week in the past six months.

Key Findings

What users look at: When the title is placed at the top, users focus on the title and interaction area, quickly deciding whether to continue watching. When the title is at the bottom, users watch the video first, making the video’s opening quality more critical.

Impact on operations: Title‑top layouts increase visibility of the “second scene” feature, while proximity between title and interaction area can cause mis‑taps and lower click‑through rates.

Best scheme: The optimal layout depends on business goals—title‑top for expanding the second‑scene feature, interaction‑area‑optimized for engagement, or recommendation‑focused layouts for longer watch times.

Hidden issues: Eye‑tracking revealed subtle problems such as unnoticed mis‑taps on author avatars and delayed responses that traditional interviews missed.

Conclusion

The eye‑tracking study provided granular insights into user attention distribution, informed UI trade‑offs, and uncovered usability problems invisible to conventional methods, offering valuable guidance for future short‑video product design.