Nobody Thinks Like You

TL;DR: False consensus makes you think users think like you. They do not. Your preference feels like evidence because your brain keeps using itself as the sample size.

Maybe you chose a layout because it felt clean. Picked a label because it seemed obvious. Cut a feature because you could not imagine anyone needing it. You called that instinct. You might even have called it experience. A lot of the time, it was just you designing for yourself in a room full of people who are nothing like you.

That is a normal blind spot. More experience does not fix it on its own. Researchers have been writing about it for decades. You keep seeing it in design work because the mistake feels reasonable from the inside.

I still catch myself doing this when a screen feels obvious too fast.

False consensus in action

In 1977, Lee Ross, David Greene, and Pamela House at Stanford ran a series of experiments on what they called the false consensus effect. They gave people ordinary, unremarkable scenarios (a speeding ticket, a class assignment, a TV commercial) and asked them to predict what most other people would do. Then they asked what the participants themselves would do. The pattern was clear every time:

Social observers tend to perceive a ‘false consensus’ with respect to the relative commonness of their own responses.

— Ross, Greene & House

In plain terms, you assume most people would do what you would do. And when they do not, you tend to treat them as the exception.

That second part matters as much as the first. Ross and his colleagues didn’t just find that people overestimate agreement. They also found that participants judged people who chose differently as having more extreme, more revealing personality traits. The person who pays the speeding ticket assumes most people would pay it. And they assume that someone who contests it in court must be the kind of person who loves conflict, or is paranoid about authority, or has something to prove. The same logic runs in reverse. Anyone who contests it thinks you are soft for not fighting back.

Both groups are doing the same thing. They start with their own choice, treat it as normal, and then explain away anyone who does something else.

In design reviews, it happens like this. Someone builds a navigation pattern that makes sense to them and then describes users who cannot find their way around as “not our target user” or “low tech literacy.” Someone writes UI copy that feels obvious and assumes people who misread it were not paying attention. I have seen teams do this in the room while the evidence was right in front of them. One person misses the menu label in a test and the room starts explaining the person instead of the label. That bias pushes you toward the wrong call and gives you a ready-made excuse for why the people who struggle do not count.

The brain does this when it has to guess what other people are like. Your own preferences, habits, and reactions are the nearest data you have. You have lived them, so they feel solid. So when your brain tries to estimate what is normal, it reaches for the nearest example it can find, you, and starts there.

In product work, it happens all the time. You use your own product with your own knowledge, patience, and tolerance for ambiguity. You know where everything is because you put it there. You understand the terms because you wrote them. The flow feels smooth because it fits how you think. That is why the mistake feels like confidence instead of bias.

What this costs

In 2011, Color launched with $41 million and a strange idea about photo sharing. You took photos on your phone and the app mixed them with photos from other people nearby. Same party. Same street. Same room. The idea was that people would want that. A live stream built from whoever happened to be around.

A lot of people opened it and did not get it. Not in some grand product sense. In the basic sense. What is this for. Why is this showing me other people’s photos. Why would I keep using it. The company changed the interface within a week.

Later, the founders said they had launched “a network you don’t know how to get good at” and that they had “threw a mountain at people.” That sounds about right. They asked people to understand a whole new behavior all at once. It probably felt simpler from inside the company because the people building it already knew what the thing was supposed to be.

That is what happens. The people closest to the product can still see the point. Everyone else is just trying to work out what this thing is.

I have seen smaller versions of this in product work over and over. A label feels obvious because the team already knows the feature. A flow feels easy because the people reviewing it already know what comes next. Someone gets stuck in a test and the room starts explaining the user instead of the screen. That is usually when the trouble starts.

The hardest thing about false consensus is how normal it feels. Knowing about the bias does not protect you from it. Bosveld, Koomen, and van der Pligt found the same pattern in a different setting years later. Your brain keeps defaulting to similarity, and awareness helps a little, but it does not solve it.

The user you imagine

The shift is simple but not easy. Before you make a decision, name out loud whose preference you are acting on. Whose, exactly. Is this layout choice based on something you observed in a user? Or is it based on what you would want if you were using this product?

If you hesitate, that hesitation is the answer. It means you are pulling from your own head.

One test is worth running. After any design choice, ask: would someone who has never seen this product land here and see what I see? Not after a tour. Not on a second try. Right now. If you cannot say yes with evidence, you are leaning on your own experience again.

Pay attention to how you talk about users who struggle too. If your first reaction to a failed user test is to describe the participant as confused, distracted, or not representative, you are watching the bias happen in real time. The person who does not behave the way you expected is the data. A product that only works for people who think like you is a mirror with buttons on it.

This does not mean your instincts are worthless. Pattern recognition built from real observation matters. But pattern recognition built from your own preferences, dressed up over years as expertise, is still an assumption. Those are the dangerous ones, because they do not feel like assumptions at all.

The most dangerous thing in design is not getting it wrong. It is getting it wrong while feeling dead certain you got it right.