Menu

IA

Card Sorting

An information architecture method for learning how people group content and what they call it, in open, closed and hybrid forms.

How to use card sorting to understand user mental models and shape information architecture, and how to choose between the open, closed and hybrid variants.

5 min read

What it is

Card sorting is a method for finding out how users would organise your content, and what language they would use to describe it.

Participants are given a set of items, usually pieces of content or , and asked to group them. Depending on the , they either invent the categories themselves or work with ones you supply.

The output is not an . It is evidence about how people expect one to work, which is the input to designing it.

Card sorting tells you how users think content should be grouped. It does not tell you whether they can find anything once you have grouped it.

Open, closed or hybrid: which one you need

The three answer different questions, and the one you want depends on whether you already have a structure.

Open. Users group the items and invent their own labels. Use it when you have no structure yet and need to learn how people think about the content.
Closed. You supply the categories and users place items into them. Use it when you have a proposed structure and want to know whether it holds.
Hybrid. You supply categories but users can add their own. Use it when you have a structure you are reasonably confident in and want to find what it is missing.

Key takeaway

Open discovers a structure, closed tests one, and hybrid does both at the cost of a messier result.

When to use it

Card sorting earns its place when the question is about grouping and naming.

It is most useful when:

You are designing or restructuring an information architecture
You need navigation or menu structures that match how users think
You want labelling driven by users rather than by the org chart
You are working with a large or complex content set
It is the wrong method when:
You need to know whether people can find things, which is tree testing
The content set is small enough that the grouping is obvious
You need behavioural or interaction insight rather than structural insight
Card sorting and tree testing answer opposite halves of the same question, and most information architecture work needs both.

How to run it

The mechanics are simple. Almost all of the quality comes from the preparation.

Set up properly

Decide which items are in scope, how many participants you need, and which answers your question.

  • Around 30 to 60 items. Past that, people stop sorting and start filing.
  • Around 15 to 30 participants, which is where grouping become legible.
  • Item names in user language, not language. Otherwise you are testing , not grouping.

Run the method

Give participants the items and ask them to group them in a way that makes sense to them. In a closed or hybrid sort, supply your categories at this point.

Ask them to label each group in their own words, and to talk through their reasoning as they go. The reasoning is usually worth more than the grouping.

Encourage people to sort naturally rather than logically. You want the intuition, not a considered .

Capture and make sense of it

Look across participants rather than at any one of them. The value is in the .

  • Items that were grouped together consistently
  • The words people reached for when naming a group
  • Items that landed somewhere different every time
  • Differences between participants that suggest more than one audience

Where it goes wrong

Most of the damage is done before anyone sorts anything.

Too many cards, so people stop sorting and start filing
Items written in system language, which tests comprehension rather than grouping
Reading significance into small differences between participants
Naming the clusters before they have finished forming
Treating the groupings as an architecture rather than as input to one
Running a closed sort when you do not yet have a structure worth testing

What you get from it

Done properly, this method gives you:

Groupings that emerged from users rather than from your org chart
The words people reach for when naming a category
Items nobody could place, which are usually mis-scoped or mis-named
Differences between participants that reveal more than one audience

Key takeaway

It helps you structure content the way users expect.

Get in touch

If this sounds like something you need, we can help you structure your content in a way that makes sense to your users.

No guesswork. No assumptions. Just clarity you can build on.

FAQ

Common questions

A few practical answers to the questions that usually come up around this method.

What is card sorting in UX?

It is a research method where users group content into categories, so you can see how they expect it to be organised and what they call it.

What is the difference between open, closed and hybrid card sorting?

In an open sort users create their own categories and labels. In a closed sort you supply the categories and they place items into them. A hybrid sort supplies categories but lets users add their own.

How many items and participants do you need?

Typically 30 to 60 items and 15 to 30 participants. More items than that and people start filing rather than sorting.

Is card sorting the same as tree testing?

No, and they answer opposite questions. Card sorting tells you how people would group content. Tree testing tells you whether they can find it once it has been grouped. Most information architecture work needs both.

Does card sorting improve UX?

It does when the problem is structural. It ensures navigation and grouping match how users think, rather than how the organisation is arranged.

Quick take

If you want to know how users group and name your content, run a card sort. Which of the three variants you run depends on whether you have a structure yet.

LET'S WORK TOGETHER

Ready to improve your product?

UX, research and product leadership for teams tackling complex digital services.

Previous feedback

I had a fantastic experience working with Andy. One of his most impressive achievements during our time at NHS HEE was masterminding a deeply complex information architecture for a new platform that brought together a large number of legacy websites.

Will Parkhouse

Senior Content Designer