Using Card Sorting to Create Stronger Information Architectures

January 23, 2017

Card sorting is an information-architecture technique that enables a group of subject-matter experts or users to either

provide input to the definition of a new information architecture for a Web site or application
evaluate and provide feedback on a Web site’s or application’s existing information architecture

During a typical card-sorting exercise, participants organize a set of cards comprising navigation items for a particular context into categories or groups that seem logical to them. Participants can name these groups and, thus, create a folksonomy, or user-defined taxonomy.

Champion Advertisement

When an information architect or UX designer is designing the navigation structure for a Web site or application that has a variety of content and functionality, conducting a card-sort exercise helps in understanding users’ expectations and mental models of the information space. Card sorting is a useful approach for designing information architectures, workflows, and menu structures. The understanding of the structure of an information space that you gain from a card sort provides the foundation for an information architecture. A key objective of a card-sort exercise is to identify and adopt a standardized taxonomy that is useful in organizing an information environment and creating a navigation structure that is obvious and natural to users.

Knowing how your users group information can help you to

structure a Web site or application
decide what to put on the home page
label categories
define a syntax or taxonomy that helps maintain consistency

How to Conduct a Card Sort

You can conduct a card-sort exercise in a variety of contexts—for example, in person during a one-on-one session or workshop or remotely via email or using a card-sort application. The basic process is as follows:

Print the navigation items the participants must categorize on cards. These cards should be large enough to accommodate the names of the items in a font that participants can read easily when they’ve spread the cards out on a desk or table—that is, at least 14 points.
Ask participants to group items in any way that makes sense to them. You may also ask them to name the resulting groups.
Once all participants have completed the exercise, record their data in a spreadsheet and examine the groupings. There will be general agreement about many items, whose groupings will be fairly apparent. For example, all participants may group Technical Support with Complaints and Product Assistance.
Generate a pictorial representation of the resulting groups or categories as a dendrogram, or category tree, by doing a cluster analysis on the combined data set.

Benefits of Card Sorting

Card sorting can be especially useful in any of the following circumstances:

If there is a huge variety of items that need to be organized.
If no syntax or taxonomy currently exists for organizing items.
If items are difficult to group into distinct categories because of their similarities.

Types of Card Sorts

There are two basic types of card sorts:

open card sorts
closed card sorts

Open Card Sorts

For an open card–sort exercise, ask participants to organize navigation items for a Web site or application into categories, as they see fit. Then, ask them to name each category, choosing a label for each category that provides an accurate description of its content. An open card sort is useful when you want to learn how users group content and what terminology or labels they would use for each category.

Closed Card Sorts

For a closed card–sort exercise, ask participants to organize navigation items for a Web site or application into predefined categories. A closed card sort, or reverse card sort, works best when a predefined set of categories already exists. A closed card sort is useful when you want to learn how users would group content into each predefined category.

Combining Card-Sort Techniques

At times, it may be useful to try combining both of these techniques. For example, you might initially conduct an open card sort to identify categories, then conduct a closed card sort to validate the category labels and groupings of items. Recently, I worked on a project on which we took a similar approach.

Card-Sorting Tools

There are a number of tools on the market that are specifically for card sorting, including xSort, Optimal Sort, and UX Sort. Plus, you can use Microsoft Excel’s pivot tables for card sorting.

xSort

xSort is a free card-sorting application for Mac OS that is shown in Figure 1. Its user interface consists of a virtual table top and a set of cards that participants can group. This application supports both open and closed card sorts and provides detailed statistical results, including dendrograms that are based on cluster analysis.

Optimal Sort

Optimal Sort is an effective Web application for remote card sorting. The free version of the application supports only open card sorting, but the premium version also includes support for closed card sorting. Optimal Sort provides comprehensive results, including a similarity matrix, dendrograms, and a Participant-Centric Analysis (PCA).

UX Sort

UX Sort is a free card-sorting application for Windows. Similar to xSort, the user interface consists of a table top and a set of cards that participants can group. The application supports open card sorting and provides results as dendrograms that are based on cluster analysis.

Microsoft Excel

The versatility of Microsoft Excel allows you to use the application as a simple tool for card sorting. You can conduct both open and closed card sorts using Excel, then analyze the data you’ve gathered from participants by creating a pivot table. More experienced Excel users can use features such as slicers for comprehensive results or PowerPivot for a visual representation of the results.

Physical Versus Digital Card Sorting

The key advantage of physical card sorting is the personal touch of interacting with participants face to face. You can gain a lot of insights into participants’ mental models by observing them during physical card—sort sessions. However, one of the biggest challenges of a physical card sort is data analysis. Conducting a physical card–sorting session is simple, but analysis can be cumbersome and time consuming. However, conducting a physical card sort is useful for a small group of participants with an information space that is not too large—that is, one with less than about 40 items.

Using software is more efficient for analyzing and generating card-sort results. Almost all card-sorting applications use cluster analysis and generate dendrograms. You can use applications for remote card sorting, which lets participants initiate a session at their convenience. However, such applications provide only statistical data and don’t provide any insights into the participants’ mental models. Use these applications when participants are spread across geographies or the number of participants is large.

References

Usability.gov. “Card Sorting.” Usability.gov, undated. Retrieved January 6, 2017.

Hudson, William. “Card Sorting.” Interaction Design Foundation, undated. Retrieved January 6, 2017.

In Card Sorts | Information Architecture