Card sorting settles naming fights faster than meetings

Once I worked on a project where two companies were becoming one, and each of them had its own name for almost everything. Same service, two names. Same section, two different places in the menu. Every meeting about navigation went in circles, because every label carried a team's history and a little bit of its pride.

Nobody in those meetings was wrong, exactly. Both names made perfect sense to the people who had been using them for years. That was the real problem. The room was full of people who already knew where everything lived, deciding how things should look for people who didn't.

I remember one meeting in particular where we spent almost an hour on a single menu item. By the end, the decision depended less on what was clear and more on who had the most patience. That's when it became obvious we were asking the wrong people.

Stop asking the room

So we stopped arguing and ran card sorting with real users. If you haven't done it, the idea is simple. You write each piece of content on a card, give people the whole stack, and ask them to group the cards in whatever way makes sense to them. Then you ask them to name each group in their own words.

We kept it clean on purpose. No slides, no explanation of the merger, no hints about which company used which word. Just the content and the people who would actually have to find it.

The patterns showed up faster than I expected. Some categories that each team considered essential didn't survive at all, because users simply didn't think that way. A couple of the final section names came almost word for word from what participants wrote on top of their piles. Some of the content we'd been fighting over ended up somewhere neither team had proposed.

The argument didn't need to be won anymore. It had been answered, by the only people whose opinion actually mattered.

Users don't care whose word it was. They care whether they can find it.

What I liked most was what it did to the team. Once the results were up on the wall, nobody had to lose. People could change their minds without feeling like they had given in to the other side. The discussion moved from "my name or yours" to "what did users do here, and why," which is a much healthier conversation to have with people you're about to work with every day.

That's something I didn't fully appreciate before. A naming fight is rarely about names. It's about identity, and about who gets to decide. Research gives everyone a way out that doesn't cost them anything.

It works on a deadline too

Card sorting has a reputation for being slow and academic, something you do when there's a research team and a comfortable timeline. It doesn't have to be.

On another project, the whole design phase had two weeks, start to finish. There was no time for a proper study with recruited participants, scheduling, incentives and a report at the end. And still, the product depended on getting the structure right, because the core of it was an order flow that customers would use every day.

So I ran one focused session with the sales and operations leads. They were the people who explain the service to customers all day long and hear every confused question. In a way, they had already done years of informal research without calling it that. They knew which words customers used, which options confused them, and where people usually got stuck.

The order flow came almost directly out of that session. Start with the item you're sending, then show only the options that matter for that item. No giant form at the beginning, no long menu full of things that don't apply. It sounds obvious written down like this, but it only became obvious once we watched the people closest to customers sort the pieces themselves.

It isn't the same as testing with customers, and I wouldn't pretend it is. Internal people have their own blind spots and their own vocabulary. But when the choice is between a quick sort with the people closest to customers and no research at all, the quick sort wins every time. You can always validate later. You can't get back the two weeks you spent guessing.

Or skip the cards entirely

Sometimes you don't even need the cards. In day-to-day work, a lot of the answers are already being said out loud, you just have to listen for them.

On a product for car buyers, the stages of the buying pipeline use the exact words the buyers already used while working. Fresh, in the works, ready, bought. We didn't invent any of them. They came from listening to how people talked about their own day, in calls and quick conversations between tasks. Nobody had to learn those labels, because they were already theirs.

Compare that with what usually happens. A team picks internal terms during a kickoff, those terms end up in the code, then in the interface, and then the company spends months teaching customers a vocabulary that only made sense in that first meeting.

The cheapest win I know

Borrowing people's own language is the cheapest usability improvement I know of. It costs almost nothing, and it fixes problems that no amount of visual polish can fix.

The words are already out there, in support tickets, sales calls, onboarding sessions and hallway conversations. Most of the time, the job isn't to come up with better names. It's to write down the ones people are already using, before someone in a meeting room invents new ones.

And when the room can't agree, that's usually a sign the room is the wrong audience. Take the question to the people who'll live with the answer.