Users perceive aesthetically pleasing designs as more usable — even when they are not. Visual quality creates a halo effect that increases tolerance for friction and minor usability problems.
In 1995, researchers Masaaki Kurosu and Kaori Kashimura ran a study at Hitachi Design Center in Japan. They gave participants two versions of an ATM interface — functionally identical, but one was laid out neatly and the other was arranged awkwardly. Participants rated the attractive version as significantly more usable, even though neither version was actually easier to operate.
This wasn't aesthetic preference or personal taste. The participants genuinely believed the prettier interface worked better. Don Norman later named this the Aesthetic-Usability Effect: when something looks well-designed, users assume it works well. They are more forgiving of its flaws, more patient with its errors, and more likely to attribute problems to their own mistakes rather than to the interface.
“Attractive things work better — not because of any engineering reason but because of the emotional response they evoke.”
— Don Norman
Banking interfaces are a canonical example of the Aesthetic-Usability Effect in practice. Both dashboards below show exactly the same information: balance, recent transactions, spending, and account activity. The data is identical. The interactions are identical. What differs is everything about how they look — and that difference completely changes how users perceive the reliability, security, and ease of use of the product before they touch a single element.
The bad dashboard isn't broken — it displays all the information correctly. But the visual density, the default browser fonts, the table-based layout and the dated colour scheme all signal something about the care that went into the product. Before users do anything, they have already formed a judgment.
| DATE | DESCRIPTION | AMOUNT | BALANCE |
|---|---|---|---|
| Apr 12 | WHOLE FOODS MKT | -$84.20 | $4,218.40 |
| Apr 11 | NETFLIX.COM | -$15.99 | $4,302.60 |
| Apr 10 | SALARY DEPOSIT | +$3,200.00 | $4,318.59 |
| Apr 9 | AMAZON MKTPLACE | -$47.30 | $1,118.59 |
| Apr 8 | STARBUCKS #0842 | -$6.80 | $1,165.89 |
All the information is here. But the visual language signals a product that hasn't been cared for — which signals a service that might not be either.
Same balance. Same transactions. But this product feels trustworthy, careful, and modern — before the user does anything at all.
This is not about decoration. The visual quality of a banking interface communicates something real about the organisation behind it: that they invested in the product, that they pay attention to details, that they care about the experience. Users correctly read visual quality as a signal of product quality — even when they shouldn't. The Aesthetic-Usability Effect means that signal works, whether or not it is perfectly correlated with the underlying engineering.
Error states are where the Aesthetic-Usability Effect has its highest practical impact. When something goes wrong — a payment fails, a file doesn't upload, a page doesn't load — the user is already frustrated. The aesthetic quality of the error experience determines whether that frustration crystallises into distrust or is dissolved by warmth and clarity.
A raw, technical error screen — red header, monospaced stack trace, dense paragraph of system language — confirms every anxiety: this product is unreliable, complicated, and doesn't care that I'm stuck. A well-designed error screen with a human illustration, plain language, and clear actions communicates the opposite. Same error. Completely different memory of having encountered it.
Stack trace on a public error screen. Technical noise the user cannot act on. Every element signals carelessness.
Warm tone, clear actions, human accountability. “It's not your fault” does real psychological work. The error is the same — the experience of it is completely different.
Notice the detail at the bottom of the good screen: a short error reference code that users can copy if they contact support. This is the only technical information that belongs on a public error screen — not because it's useful to the user directly, but because having a reference number signals that the system is tracking the problem. It provides evidence of competence without exposing the machinery behind it.
The Aesthetic-Usability Effect applies to AI-generated content as directly as it does to designed interfaces. A response that arrives as a monospaced wall of text — even if the information it contains is accurate and complete — reads as less authoritative, less trustworthy, and harder to process than the same information formatted with clear hierarchy, readable fonts, and intentional spacing.
This matters because users who perceive an AI response as clearly structured and well-presented are more likely to trust its content, engage with follow-ups, and return to the tool. Identical information formatted differently produces different judgments of the AI's capability — purely through the Aesthetic-Usability Effect.
Same advice — hard to trust, hard to scan. The content is good. The presentation signals a rough prototype, not a reliable tool.
Same advice — easy to trust, easy to act on. Numbered steps, bold labels, a distinct tip card. The AI feels more capable because its output looks more considered.
The good response isn't longer — it's actually slightly shorter. The visual structure does the cognitive work that would otherwise require more words. Numbered steps tell users how many items to expect. Bold labels let users scan before they read. The tip card at the end stands out as the most memorable piece. The same information, more legibly packaged, is experienced as higher quality output — regardless of the model that generated it.
The Aesthetic-Usability Effect is both a generous gift and a strict constraint. The gift: invest in visual quality and users will be more forgiving, more patient, and more likely to succeed when they hit friction. The constraint: the effect works on first impression and during smooth interactions, but it cannot survive genuinely broken functionality. Users who persist through a beautiful product's errors will eventually run out of goodwill if the errors don't get fixed.
The practical implication is that aesthetic quality and functional quality are not in competition — they are compounding. A well-designed interface that also works reliably earns disproportionate trust. A well-designed interface that doesn't work loses that goodwill faster than an ugly one would, because the gap between expectation and experience is larger. Design for both, in that order: make it work, then make it beautiful, knowing that making it beautiful will make the working feel better too.
Kurosu, M., & Kashimura, K. (1995). Apparent usability vs. inherent usability. CHI '95 Conference Companion. · Lindgaard, G., Fernandes, G., Dudek, C., & Brown, J. (2006). Attention web designers: You have 50 milliseconds to make a good first impression. Behaviour & Information Technology, 25(2), 115–126. · Norman, D. A. (2004). Emotional Design. Basic Books.