Aesthetic-usability effect
Definition
People tend to perceive attractive designs as easier to use, and are more tolerant of minor usability problems in products that look good.
The aesthetic-usability effect is the tendency to judge good-looking interfaces as more usable than plain ones, whether or not they actually are. It was first studied by Masaaki Kurosu and Kaori Kashimura of Hitachi's Design Center in a 1995 CHI paper. According to NN/g's summary, they tested 26 variations of an ATM interface with 252 participants, who rated each layout on ease of use and on aesthetic appeal. The paper's abstract reports that apparent usability "is strongly affected by the aesthetic aspects rather than the inherent usability".
Why it matters
First impressions shape how people approach a product. A polished interface earns some goodwill: people try harder, assume the product is competent and forgive small annoyances. NN/g describes it this way: "People tend to believe that things that look better will work better".
The same effect is a risk for research. In usability tests, participants who struggled with a task may still praise the design afterwards, which hides problems you need to fix. Attractive visuals can make a flawed flow look finished.
How to apply it
- Do invest in visual quality: consistent spacing, clear typography and a coherent color system all contribute to how usable a product feels.
- Do treat aesthetics as part of usability, not decoration. Good visual hierarchy also helps people find things.
- Do watch behavior, not just opinions, in testing. Count errors, hesitations and task completion, and weigh them above "it looks great".
- Don't let a high-fidelity prototype stand in for usability testing. Polished mockups can get approval for flows that will fail in use.
- Don't trade function for looks, such as low-contrast text, hidden labels or tiny targets chosen for a cleaner aesthetic.
Common mistakes
- Assuming beauty covers real problems. NN/g notes that visual appeal forgives minor issues only. Serious usability problems still drive people away.
- Relying on satisfaction ratings alone. Self-reported scores are exactly where this effect shows up.
- Reading the original study as proof that beautiful designs perform better. The 1995 work measured perceived usability ratings of layouts. It shows how attractive designs are judged, not that they are faster or less error-prone in use.
In AI products
AI output adds a new layer to this effect. A fluent, well-formatted answer with confident headings and tidy bullet points looks trustworthy, even when the content is wrong. A clean interface around that output adds to the impression. That can push people toward automation bias, accepting results without checking them.
Design so that polish does not stand in for accuracy. Show sources where they exist, make uncertainty visible, and keep the presentation of a tentative answer different from a verified one. The goal is calibrated trust: people should trust an AI feature as much as its actual reliability warrants, not as much as its formatting suggests.