09.02.26 By Doug Dimon
Every product team has heard some version of this sentence, usually delivered with a slightly apologetic wince:
“It works great, it’s just… not very pretty.”
Followed immediately by…
“It kind of looks like a Windows 95 program.”
Here’s the uncomfortable truth: that sentence isn’t a complaint about font color. It’s a prediction about adoption.
There’s a well-documented finding in human-computer interaction called the aesthetic-usability effect. In 1995, researchers Masaaki Kurosu and Kaori Kashimura at Hitachi’s Design Center tested 26 versions of an ATM interface on 252 people, having them rate each one on both looks and ease of use. The versions people rated as more attractive, they also rated as easier to use, even when the underlying functionality was identical. Looking better didn’t just make people like the interface more. It made them believe it worked better, before they’d done a single task on it.
Don Norman later built an entire book, Emotional Design, around the same idea: Attractive things make people more forgiving of friction, more patient with small flaws, more willing to stick around long enough to actually learn the thing.
And that first impression forms fast. A 2006 study by Gitte Lindgaard and colleagues at Carleton University found that people form a lasting opinion about a web page’s visual appeal in about 50 milliseconds – a twentieth of a second, faster than a blink. That snap judgment then colored how people rated the site’s usability even after using it for much longer. You don’t get a grace period to win someone over. The eyes vote first, and they vote almost instantly.
This is where a lot of enterprise software quietly misses the mark. Somewhere along the way, “utility” and “ugly” became synonyms, as if a tool employees use dozens of times a day is somehow exempt from the same psychology governing every consumer app on their phone. It isn’t. Nobody’s brain has a separate, more forgiving usability module that switches on at 9am when they clock into work software. The person who feels a small jolt of pleasure opening a well-designed app on their personal phone brings those same expectations with them to the office. And when they open an internal tool that looks dated, cluttered, or hastily assembled, that reaction is immediate. It may be subtle, but it shapes how they feel about the experience before they’ve completed a single task. First impressions influence trust, satisfaction, and ultimately adoption.
There’s a newer version of this visual design problem today: AI-generated interfaces.
It’s never been easier to prompt your way to a screen that looks convincing at first glance – cleanish layout, a reasonable color palette, rounded corners in all the fashionable places. And that’s exactly the trap. An interface that looks good as a screenshot isn’t necessarily one that works well for someone using it all day, every day.
Those are very different standards. The gap between “looks plausible” and “feels right in real use” is where professional design expertise still matters.
What AI-generated design tends to miss isn’t obvious wrongness. It’s the much subtler stuff: a type scale that’s technically consistent but doesn’t actually guide the eye anywhere, spacing that follows a rule instead of a rhythm, a color palette that’s harmonious in isolation but falls apart the moment real content – long names, error states, empty states, someone’s actual messy data – gets poured into it.
AI is very good at producing something that resembles good design. What it often struggles with is the judgment required to create a digital experience that remains intuitive, consistent, and effective after hours of real-world use.
That’s the actual value a professional designer brings that a generation tool doesn’t: not just knowing something looks wrong, but knowing why, and having the judgment to fix the right thing instead of the obvious thing.
AI can be a genuinely powerful way to get past the blank page. It can generate a solid first draft in minutes and accelerate the early stages of design.
The problem comes when that first draft is treated as a finished product – because it “looked good enough” in the demo. That’s how you end up right back at the top of this post, except now it’s an AI-generated interface instead of a legacy one. Users may not be able to articulate exactly what’s wrong. They’ll simply feel something’s off. The experience doesn’t feel intuitive, enjoyable, or trustworthy, even if they can’t pinpoint why.
It is not a vague request to the product team to “make it prettier” at the end of the project. It’s a set of deliberate design decisions made early, applied consistently, and reinforced throughout the product development process:
Build (or adopt) a real design system before you build the product. Consistent spacing, a real type scale, a defined color palette used with intention rather than whatever hex code was closest. This is the difference between a product that feels considered and one that feels like it was assembled from whatever components were lying around – because it usually was.
Get the typography hierarchy right before anyone argues about brand colors. Most “ugly” enterprise software isn’t ugly because of bad colors. It’s ugly because every piece of text is fighting for the same visual weight – no clear distinction between a page title, a section label, and a field description. Fix the hierarchy and the interface instantly reads as more organized, even before anything else changes.
Use whitespace like it costs money, because in a sense it does. Cramming more onto a screen doesn’t make software more powerful. It makes it exhausting. The interfaces people describe as “clean” are almost always just interfaces with room to breathe.
Polish the small moments, not just the big screens. Loading states, hover states, the tiny animation when something saves successfully – these cost very little to get right and are exactly the details that separate “functional” from “this feels like a well-made product.” Nobody notices good micro-interactions consciously. Everybody notices their absence.
Benchmark against the best experience your users have anywhere, not just your closest competitor. The bar in your users’ heads isn’t set by the other enterprise software they’re forced to use. It’s set by their phone, the four apps they check before breakfast, and every gorgeous interface they didn’t have to be trained on. That’s an unfair comparison, and it’s the one happening in their head anyway.
Let AI create in context, not in isolation. AI is most effective when it’s grounded in the right knowledge layer, design system, and business requirements from the start. Use generative tools to accelerate exploration and eliminate the blank-page problem, but keep designers involved throughout the process. AI can produce options quickly. Designers provide the judgment to evaluate, refine, and adapt those ideas to real-world user needs. The most successful teams aren’t replacing design with AI. They’re combining the speed of generation with the expertise required to turn promising concepts into products people genuinely enjoy using.
None of this requires turning your CRM into an art project. It requires treating visual design as a functional requirement with a direct line to adoption, not a coat of paint applied after the “real” work is done.
People don’t just prefer good-looking software, they trust it more, forgive it more, and stick with it longer – sight, quite literally, unseen past the first fraction of a second.
“It looks like a 90s Microsoft program” was never really a comment about aesthetics. It’s a polite way of saying, “I don’t think I’m going to enjoy using this.” And when it comes to adoption, that feeling matters more than many teams realize.
Don’t let dated design cost you adoption. Bridgenext’s UI/UX design services help you turn AI-accelerated first drafts into interfaces people actually trust and enjoy using. Talk to our team →
