Some of the best design decisions I’ve ever witnessed happened on whiteboards, with paper cutouts, or as rough sketches during workshops. Never in Figma. Often ugly.
In one project, we cut out paper components - based on patterns from the existing design system - and rearranged them live during a workshop. Rectangles and arrows and handwriting, nothing more. And because everyone in the room already knew the components, because nobody could hide behind aesthetics, something shifted. People stopped debating how things looked and started debating whether they solved the problem.
That’s what taste looks like before pixels exist. The rough sketch that nails the problem is worth more than the polished mockup that misses the point.
If you can’t explain your taste, you don’t have taste. You have preferences.
Gatekeeping dressed up as wisdom
Here’s what frustrates me about how the design industry talks about taste.
We fetishize it without defining it. We point at Dieter Rams and say “that’s taste.” We browse Apple’s product pages and say “that’s taste.” But when someone asks “okay, how do I develop it?” the answer is usually some variation of “look at more stuff” or “you’ll know it when you see it.”
That’s not instruction. That’s gatekeeping dressed up as wisdom.
When the first car came out, nobody cared about its color - the competition was a horse. But once cars became commodity, quality and details became everything. Software is in exactly this place right now. Simply shipping a product that works is no longer enough. Everyone can do that, especially with AI. What makes a product stand out is coherence, intentionality, the feeling that someone actually thought about it. And yet the industry that needs taste most treats it like a birthright instead of a trainable skill.
The data confirms this isn’t just a design-community navel-gaze. The World Economic Forum’s Core Skills in 2030 analysis puts “Design and user experience” in the “Emerging skills” quadrant - important, but not yet core. What IS core? Analytical thinking, creative thinking, resilience. LinkedIn’s Work Change Report goes further: 70% of current professional skills will change or become obsolete by 2030. Same direction from a completely different dataset. Execution skills deflate, judgment skills inflate. And for a designer, taste is where judgment skills converge.
So let’s define the damn thing.
Taste is the ability to discern quality, appropriateness, and coherence - and to articulate why.
Elizabeth Goodspeed and NN/g (It’s Nice That, NN/g) both define it as navigating possibilities with intention and strategic vision. I’d push further: the layer that matters most is explaining those decisions under pressure, to someone who doesn’t share your visual vocabulary. In my experience, most designers with genuine taste fail not at seeing what’s right, but at defending it when a VP asks “why not the other version?”
I break it down into three components, and all of them are learnable.
Pattern recognition - your internal database. You’ve seen enough to tell what’s exceptional from what’s mediocre, not in absolute terms, but in context. A brutalist website works for an underground music venue. For a healthcare app serving elderly patients? Catastrophic. Same visual treatment, opposite quality assessment. You build this through deliberate exposure - not scrolling Dribbble (that’s entertainment, not education), but asking “why does this work?” every time you encounter something that stops you.
Contextual judgment - your filter. User needs, business goals, technical limits, brand identity - all the constraints that determine whether a design decision is brilliant or disastrous. Minimalism works for Apple because their brand IS simplicity. Try that same minimalism for an enterprise tool where financial analysts need 47 data points visible simultaneously and you’ll get fired. You build this by obsessively reverse-engineering the decision architecture behind every piece of design you encounter. Not “what did they make?” but “why did they make this and not something else?”
(And before the UX research camp objects - Filip Mishevski argued on UX Collective that taste and empathy are being falsely pitted against each other. He’s right. Taste as I’m defining it already contains empathy - contextual judgment means understanding the user. Taste without empathy is designing for Dribbble. Empathy without taste is a research deck that nobody implements.)
Articulation - your multiplier. “I like it” is not articulation. “This works because it reduces cognitive load by limiting choices to three, uses color to signal hierarchy, and creates a clear visual path from headline to CTA” - that is. Don Norman called it “making thinking visible.” I’d call it something more blunt: if you can’t defend your design decision in a room where you’re the only designer, your taste doesn’t count.
When I recruit designers, I learn most about taste not from portfolios but from conversations - and often they’re not even about design. I’m watching how someone thinks. How they arrive at solutions. How ideas form and develop in real time. Whether they check assumptions or barrel forward. You can teach someone a tool. You can’t teach someone motivation, curiosity, or the habit of questioning their own decisions. The way someone thinks through a problem live, in front of you - that tells me more about their taste than any Figma file ever will.
AI can generate and match patterns. It cannot explain why one solution fits and another doesn’t. And that explanation is what separates you from a prompt.
The window closes in 2030
Two years ago, taste was a nice-to-have. You could get away with solid execution and mediocre taste because execution itself was the bottleneck. Making things took time. That time had market value.
Now? AI generates first-draft designs that match expert work almost half the time. “Good enough” arrives in seconds. Production speed stopped mattering - what matters is selection. Knowing which of the hundred options AI produces is actually worth pursuing, and being able to explain why. When a stakeholder asks for more “pop,” taste isn’t what makes you add gradients - it’s what makes you ask what “pop” means in this context. Attention? Energy? Conversion? Sometimes “pop” means bigger type. Sometimes it means removing everything else. You’re not solving the request, you’re interrogating it.
Tomer Cohen, CPO of LinkedIn, said something in an interview with Lenny Rachitsky that should simultaneously encourage and terrify you: design is “currently the hardest craft for AI to learn.” Currently. Design is harder to automate than code because it requires judgment about hierarchy, spacing, emotional resonance - in other words, taste. You have a window here. But Cohen added: “only until 2030.” You have time. The question is whether you’ll use it or spend it optimizing Figma workflows that won’t exist in four years.
Most “taste” discussions stay in the aesthetic lane. But look at what happens when taste operates as business infrastructure. Linear - $1.25 billion valuation, Series C in 2025, used by OpenAI, Perplexity, Cursor - just published a design refresh that shows this concretely. Here’s a product praised globally. And instead of adding features, the team did the opposite: they simplified. Smaller icons. Dimmer sidebar. Fewer separators. They took things away. Their CEO Karri Saarinen wrote explicitly that design debt has to be paid every 2-3 years, even when the product is successful - because incremental additions erode coherence. Most companies respond to success by adding more. Linear responds by cleaning up. Not knowing what to add - knowing what to remove. Their latest manifesto puts it directly: “People can spend more time on intent, judgment, and taste.” Net revenue retention from 133% to 171% in two years. Taste pays.
(You might think: “Sure, but Linear is a design-led company.” Fair. But Freshworks - India-based SaaS, 900+ competitors - used taste as competitive weapon too. Their “Uncomplicated Movement” drove 20-30% quarter-over-quarter growth. Taste scales beyond boutique studios.)
Ira Glass nailed something related: “Your taste is why your work disappoints you.” The gap between what you know is good and what you can produce. Every junior designer feels it.
In the pre-AI world, closing that gap required years of execution practice. Now AI can execute your vision faster than you can. If you have taste - if you know what good looks like and can articulate what’s wrong - AI becomes the bridge between your judgment and the output. Taste without execution skill might actually be the most valuable combination right now.
But here’s the uncomfortable part. AI doesn’t level the playing field between designers who have taste and designers who don’t. It widens it. Figma’s 2025 AI Report: senior designers trust their judgment when working with AI 43% more than juniors. Top designers use AI to expand what they can do. Average designers use AI to do the same work slightly faster. One group multiplies taste across new domains. The other automates mediocrity.
So let me ask you directly: when was the last time you made a design decision that wasn’t in the brief? When did you last push back on a requirement because you saw something the PM didn’t? When did you last articulate why your solution was better, not just show that it looked better? If those questions make you squirm, that’s your taste telling you something.
And taste transfers across domains - that’s how you know it’s real. I once reviewed a portfolio from a very junior designer, recommended by a colleague. The UX portion was, honestly, rough. Student projects, maybe one real client. Normally disqualifying. But the second half of the portfolio was furniture design. This person had designed chairs for a manufacturer. And I swear - I have never seen better-designed, more thoughtfully constructed seating in a portfolio. The ergonomics, the material choices, the way form followed function without sacrificing beauty. That’s what made me send that email with an invitation to talk.
Because taste built in one domain - understanding constraints, making intentional choices, knowing when something works - carries over. The best designers I know all have at least one serious interest that has nothing to do with interfaces. One does 3D printing. Another is a practicing psychologist who does therapy after her design hours (and it makes her interface work better, not worse). The worst thing you can do for your taste is to consume only UX content. If your entire reference library is Dribbble and Mobbin, your taste ceiling is whatever the design community already considers good. That’s not taste. That’s consensus.
I had a 12-week bootcamp-style program outlined here. I deleted it. Taste doesn’t develop through homework assignments - it develops through habits. Three habits, specifically. Each maps to one component of taste.
Build your swipe file with intent. Not a Pinterest board of pretty things. A curated collection where every item has two sentences explaining why it works. Not “I like it.” Not “it’s clean.” Something specific: “This onboarding flow reduces drop-off because it asks for one piece of information per screen, building commitment through micro-completions.” Twenty items in your domain. Twenty in adjacent fields (architecture, film, writing). Ten from completely outside design. If you can’t explain why it’s good, it’s not in the file. Add five items weekly. This builds pattern recognition.
Hunt for beautiful failures. Find designs that look gorgeous but serve their purpose poorly. Award-winning apps nobody uses. Elegant dashboards that hide critical data. Ask: where did taste fail? (Almost always: someone optimized for peer approval over user success.) This is harder than curating good work because it requires you to separate aesthetics from effectiveness. That separation IS contextual judgment.
Explain your choices before showing your work. In every design review, lead with reasoning: “I chose this approach because constraint X, user need Y, and business goal Z pointed here. I considered alternatives A and B, rejected them because…” Post two design critiques per month publicly - one praise, one analysis. The act of explaining taste forces precision. You can’t hide behind “I just know” when your PM is asking “but why?” The designer who can articulate judgment in a room full of stakeholders doesn’t just have taste - they have authority. This builds articulation.
The question you’re avoiding
Most designers reading this will nod along. “Yes, taste is important. Yes, I should build my swipe file.”
And most won’t do it.
Not because they’re lazy. But because building taste means admitting that your current taste might be mediocre. It means exposing your judgment to criticism, moving from the safety of execution (where the brief tells you what to do) to the vulnerability of curation (where YOU decide what’s good, and you might be wrong).
The designers who develop genuine taste will become the most valuable people in the room. Not because they can make things - AI handles that. But because they can see what’s right, explain why, and direct others toward it. That’s editorial judgment and creative direction rolled into one, and it’s the kind of work that AI makes more valuable, not less.
The designers who avoid this work will keep producing competent, unremarkable output - by hand or by AI. And in a world flooded with competent, unremarkable output, that’s the same as being invisible.
Mike Schindler, writing on UX Collective, argues that “taste” is the wrong word - too foofy, too vague. What we really mean, he says, is practiced judgment, shaped by showing up again and again, building tolerance for ambiguity, seeking critique. He’s right about the mechanism. But I’ll keep the word. Because “practiced judgment” doesn’t make anyone uncomfortable, and taste - with its implied subjectivity, its hint of elitism, its demand that you actually have an opinion - does. That discomfort is exactly what makes it useful.
Start with the swipe file. Start hunting the beautiful failures. Start explaining your choices out loud before showing the work. Call it whatever you want - taste, practiced judgment, trained instinct. The mechanism is the same.
And you can start today. Do you?