Introduction to card sorting

I wrote a tweet on Saturday that got a lot of attention (I wasn’t expecting it to). It was in response to a tweet that had stuck in my mind about the accessibility of card sorting.

In amongst the replies there was a request to write up what I had said. So this is the thread in a lot more detail.

This is part one of two on the subject of card sorting. The second explores the problems with card sorting.

What is card sorting?

Card sorting is a method to group and label information.

Traditionally, it has been used to create structure for digital products and services.

Test participants are given a reasonably large number of cards with labels written on them.

Participants are then asked to sort the cards into groups relevant to the context of the task.

There are two primary methods of card sorting and this determines the outcome of the testing.

Types of card sorting

There are two primary methods of card sorting. Open and closed.

The different types are used in slightly different ways to produce subtly different outcomes.

Open card sorting

The premise of open card sorting is to create two things.

First, asking test participant to group the cards together in a structure that they understand.

Second, asking the participants to then label the clusters they have created with a meaningful label.

The goal of an open sort is to create human focused structures and labels.

Closed card sorting

Closed card sorting is subtly different, because you (as the tester) are presenting participants with a set of defined categories.

The test asks participants to sort the cards under the categories that you’ve presented them with.

The goal of this test is to understand if the labels you’ve proposed resonate with the test participants.

Outcomes of card sorting

The outcomes should give you guidance on how to structure and label information.

Possible uses

You can use a sort for a number of tasks throughout the design process stages. for a number of tasks at various early stages of the design process.

Problem identification

If you are working to rebuild or fix a problem with a digital product or service, it’s essential to understand what the problem is you’re trying to solve.

You can use a sort to test existing primary or secondary label structure. You do this by testing if people are grouping information in the same way it currently is.

This method of exploration can help you identify if there is a problem with structure or labelling.

You should use this as one type of testing method, with other types. These should include interviews or usability testing.

Stakeholder engagement

Using an open sort can be an excellent way to learn the language that your stakeholders are using to describe things.

Doing this early on in a design process allows you to build relationships and create an essential glossary of terms.

This approach allows you to see the mental model that your stakeholders have for something. It also gives you the ability to pinpoint what language you will need to test. Want to root out jargon and get it into early testing? this is how you can identify it and plan for managing it.

Solving labelling problems

You can use card sorting to solve isolated labelling problems.

This often happens after prototype testing has identified a labelling problem.

It usually means that the sub-structure labelling works well. Or the sibling labelling works well.

This will help you get out of a problem with a gnarly labelling issue.

Grouping and labelling of information

This is the primary use for card sorting.

I’ve explained above the differences between open and closed.

Card sorting is used to define categorised structures of digital products and services.

If you are going to use this method to do this, it must not be the only way to test the effectiveness of structure or labelling.

How not to use card sorting

I recently saw the results of an open sort for the account section of an app.

The results were an unreadable mess.

The goal was to validate stakeholder language.

If you’re going to structure something that either:

  • introduces new concepts to people, or
  • uses unavoidable language or jargon

do not use a card sorting exercise.

The problem is the people were not subject matter experts.

It is hard to label and structure account sections. This is because it often contains technical or subject matter terms.

You could, and should use an information architect or navigation design expert, and a subject matter expert. The two need to work together to create a logical structure and language lay people will be able to navigate.

More like this


Benchmarking is a bit like putting a stake in the