The n=1 trap: designing for yourself, without noticing

False consensus makes our own experience feel universal. Add AI to the process, and those assumptions can spread faster. Here are four questions to ask before you hand over a design.

A white silhouette in blue and magenta light, surrounded by dark figures

I've never set foot on a yacht. Yet since 2023, part of my work has been designing journeys and content for yachting, retail and luxury. I've also carried out UX audits for private aviation. I've never flown on a private jet either.

I'm 21, I live in Nice, and I get around mostly on foot. My day-to-day life and my users' lives can be worlds apart.

Still, whenever I start a project, the same thought sneaks in: "It's obvious. Anyone would do it this way."

That thought is the trap.

Your experience is a sample of one

In 1977, psychologists Lee Ross, David Greene and Pamela House published a study on the false consensus effect: our tendency to overestimate how many people share our choices and views. We take ourselves as the starting point, then assume everyone else is much the same.

In statistics, a sample of one is written n=1. It is a poor basis for a claim about everyone.

You can hear this bias in design reviews. "The button is obvious." "The journey makes sense." "Everyone knows what that icon means."

Obvious to whom? To someone with your eyesight, your language, your device and your years of experience with interfaces?

Knowing your tools inside out can make this harder to spot. Spend all day in Figma and you start seeing an interface as the person building it, not as someone encountering it for the first time. We know the rule. We still need the reminder: the designer is not the user.

"Obvious" is something you learn

I grew up with two languages that run in opposite directions: French from left to right, Arabic from right to left. When a change in reading direction can undo an interface that seemed perfectly clear, "obvious" starts looking less like a fact and more like a habit.

The same goes for contrast you can read without effort, industry terms you barely notice anymore, or a gesture you assume everybody knows. A design choice can make things easier for one person and harder for another.

"Is it clear?" is only half a question. The other half is "For whom?"

AI can scale the same assumptions

In June 2026, I spoke on VivaTech's Sovereignty Arena stage. The talk was called "L'IA Apprend de Toi. C'est le Problème.", or "AI Learns From You. That's the Problem."

AI models learn from material we have already made: interfaces, writing, images and past decisions. Our habits and exclusions are in that material too.

Ask for a persona and you may get assumptions dressed up as research. Generate images and familiar bodies, settings and cultures can come back again and again. Generate copy and you can get something plausible that still misses the people you need to reach.

The risk is speed. An assumption that once shaped one project can now run through everything you produce with a tool. A convincing answer can feel like a challenge to your thinking when it is really just reflecting it back.

Four questions before you hand it over

"Get out of your bubble" is easy advice to give. This is more practical: a short check before sending a piece of work.

1. Who is this obvious to?

Write down the answer. If it is "me", there is still work to do. Someone who did not build the journey needs to be able to understand it.

2. Who might this design leave out?

Think about contrast, text size, language, dexterity, culture, devices and connection speed. Microsoft's Inclusive Design approach calls this "Solve for one, extend to many." Removing a barrier for someone who has been left out can help others too.

3. What have I assumed without checking?

List what you think users know, want, own and understand. An untested assumption is a bet. Presenting it confidently does not turn it into evidence.

4. Who has tried it besides me?

Jakob Nielsen's 2000 article makes the case for testing with five users. The useful starting point is not a lab or an elaborate setup. It is putting the design in front of someone who is not you.

If you use AI, add one more question: what choices did the model make for me? Its answer contains decisions you may not have made consciously. Finding them is your job.

What n=1 is really warning you about

Design is a kind of translation. We turn the world of a client and their users into an interface. If the only perspective we understand is our own, we are not translating. We are repeating ourselves.

So next time a screen feels obvious, pause. Sometimes "obvious" is just a bias that has gone unnoticed.

Georges Baradehi, independent Product Designer and UX consultant in Nice.

Explore my UX audit