7 Card sorting examples to inform your UX research

While it’s invaluable to learn about each element of card sorting, there’s nothing quite like getting a look at what an actual card sorting session looks like, before you jump into creating your own.

To help inspire you, we’ll be looking at seven different real-life card sorting case studies:

You’re probably familiar with card sorting by now, but in case you need a refresher, let’s remind ourselves:

Card sorting is a UX research method which involves people grouping different information cards—physical or digital—into various categories, in a way that makes sense to them.

The benefits of card sorting are varied and depend on your research goals. They include:

Now we’re caught up on the what and why of card sorting, let’s get into some examples of card sorting in action.

Mattel, creators of the Barbie doll, wanted to review the IA of the Doll Showcase section of their Barbie collectors’ website. With a comprehensive gallery of 2,000+ items, the Showcase was popular in user research, but user visits were declining.

The problems facing the Barbie collectors’ website was largely two-fold:

Understand collectors’ mental models regarding the Showcase, and gain ideas for how to reorganize the Showcase using card sorting.

Working with nine participants—from new to dedicated collectors, to represent the range of Showcase users—the team ran two card sorts. The first was an open sort, with ~130 cards/dolls from a fashion-focused subset of the Collection. Next up, a closed card sort was conducted, where participants reviewed ~100 cards representing the existing collections/series and were tasked with organizing them into themes.

Both the card sort and prior usability tests directed the team to the same conclusion: all collectors browse for dolls and organize their collections in unique ways. This meant a single method of search wasn’t an option. The outcome was a faceted navigation system that allowed users to browse the Showcase by doll name, release year, serial numbers or themes. As a result, use of the Showcase noticeably increased.

In 2005, Eurostar set about to streamline the usability of its website, introduce new booking features, and incorporate a new global brand identity. This required redesigning Eurostar’s global IA, which serves the company’s main website, as well as its country and region-specific websites, too.

An ambitious project, with a limited timeframe of just six months, Eurostar was faced with several challenges:

Card sorting would be used to identify the navigational elements and IA of Eurostar’s website.

After inventorying existing content, spreadsheets were used to capture the details and position of the website’s IA. Stakeholder workshops determined which items were deemed most important to be sorted and well-defined areas like About Us, Contact Us etc. were omitted.

With the goal of uncovering how users instinctively grouped Eurostar’s content, the team decided on unmoderated, open card sorting. This took place through online card sorting software, featuring ~80 cards and 180 participants—split into 20-user groups—from three different countries.

With 14,400 results items to catalog, the use of an online platform was invaluable. Analysis of the results showed users broadly grouped content either chronologically, or according to established online conventions/expectations. Additional cluster analysis of the card sort resulted in Eurostar’s new IA, which combined a chronological layout with top-level groups for booking and planning-related information.

At a later date, a further card sort (closed) was conducted to verify the improved IA, and results were overwhelmingly positive, validating four-out-of-five of the top-level groups.

Maze is a product research platform that empowers anyone to test and learn rapidly—via intuitive, customizable user tests and surveys (called mazes).
While researching the value of notifications, the Maze product team used card sorting to gather information on which notifications, if any, users would like to receive regarding the status of their and their team members’ mazes.

Implementing new platform notifications or altering existing ones is a tentative process, as too many notifications may overwhelm or frustrate users, and editing existing notifications may cause confusion. The team had several key questions:

Use card sorting and a bank of existing research questions to understand the user’s experience and problems. The team also wanted to get initial feedback on potential high-level solutions.

Deciding on closed, unmoderated card sorting conducted via Maze, the team ran the session with a group of 20 participants. To gain feedback on possible solutions, respondents were given four categories (from ‘painful’ to ‘very helpful’) and a number of solutions, then asked to sort by ‘level of helpfulness’.

The results were organized via Maze’s report tool, before being further analyzed and put into a Notion document by the product team. The primary findings shared that a majority of users would find it helpful to receive notifications when their target number of responses have been reached, and the majority prefer email as a notification channel.

While the final outcomes of this research are still underway, the results from this card sort will be valuable to future developments. And if you’re a Maze user, you’ll even see the changes for yourself!

Tasked with organizing information for a young people’s activity portal, UX designer and founder of UXParis, Sylvie Daumal, needed to determine how users would like the IA to be structured. Card sorting seemed like the obvious solution.

With lots of information needed on the website, from school-focused content, to extracurricular and local activities, the team had many questions about how to organize the content, like:

Use card sorting to gain insight into users’ mental models and guide the structure and IA of the activity portal.

Deciding on an open card sort with a group of target users, cards were populated with different words representing everything from local concerts to regular school activities and educational courses. The participants were then asked to sort these into groups that made sense to them.

The results offered valuable and surprising insight—respondents didn’t differentiate between subjects, like culture and sport, but they did consider differences between regular activities (e.g. training, classes or clubs) and events they could opt into (e.g. a concert or one-off activity). As a result, the first layer of navigation was determined to be ‘Activities’ and ‘Going Out’.

Further results revealed that users focused on fields of study or potential vocations (e.g. tourism course), rather than diploma-related categories (e.g. three-year diploma). This determined that school-related sections of the portal should be organized according to field, with course/diplomas that related back to that field being displayed within the content.

The insight gained from the card sort overwhelmingly informed—and altered—the final portal and levels of navigation, and the team felt fully confident in their final site map.

As part of an ongoing project to evaluate the usability of Singapore Polytechnic’s current website, Hilary Chan, an Economic Architect based in Singapore, employed a combination of user research methodologies—including card sorting.

The Singapore Polytechnic team used card sorting to understand how users group pieces of content and information together on the site.

The team went for an open card sort via a digital card sorting tool, working with 24 respondents. They provided participants with existing content from the website and asked them to sort the content into categories as participants saw fit. This data would later be reviewed and narrowed down via tree testing.

Using the results of the card sort, it became clear that three major groups could be derived from the exercise: courses grouped by topic, student life, and admissions & financial matters. This then formed the basis of the IA for Singapore Polytechnic’s new site.

While working on a website redesign project, Donna Spencer, Independent Design Consultant, decided to use card sorting to inform the website’s IA and site map.

There were two key issues with the current website that needed to be addressed in the redesign:

By using card sorting, the website IA could be redesigned to ensure users could find information, understand the company’s purpose and areas of expertise, and allow the website to expand along with the company.

The team wanted to understand how users familiar with the content would organize it, so the card sorting session took the form of open, moderated card sorting with 21 internal staff members as participants. Main topics of content were used for cards, rather than specific pages, and a simple spreadsheet analysis matrix was used after the sort.

While some cards were sorted in similar ways by everyone, there were many outliers. Here, the team discussions that accompanied the sort were most useful, revealing why some topics were linked, what was understood—and what wasn’t.

Following in-depth analysis of the card sort and user research, it was determined that a hierarchical categorization scheme for this website would be near impossible. So, the final website was organized by rough topic groupings, with independent mini-sites as offshoots for each topic. As a result, the home page clearly showed all areas of the company’s responsibilities, while users—regardless of prior knowledge—could head straight to the topic they were focused on.

The marketing team of a Fortune 500 financial company were working with John Nicholson, Principal Consultant at Marketade, to create a microsite to hold brand-related and marketing content that had previously been spread across multiple channels. With a plethora of unsorted content and a new website IA to determine, they needed a site structure that would help visitors navigate and browse content.

Starting from scratch, the challenge was to design a website that worked for both the user and the company’s goals. The team already had ideas about how to structure and label the IA—but none of these had been validated with real users, and were based on a majority of guesswork and intuition.

Having already invested time into the ideas and pre-design of the website, the company was looking to validate these existing IA ideas, with minor tweaks where needed.

While the project team initially wanted to use hybrid or closed card sorting, as they already had proposed categories, it was decided that to ensure the site made sense to users, an open sort would be most effective.

And so, an open, digital card sorting session took place, as stage one of a four-week project. The card sort would then be followed by a series of IA tree tests, between design iterations.

The card sort featured 100 cards out of a potential 125 content pieces, all of which were at the same content ‘level’ (e.g. all articles). First, 55 participants were selected from the target market, who took part in qualitative, moderated card sorting as part of five user groups. Next, 50 users ran through quantitative, unmoderated sessions with the same cards.

Using an online tool to sort the quantitative data, the team used data visualizations—a dendrogram and similarity matrix—to get an overview of standardized categories.

After capturing observations from the moderated sessions, all the results were combined and reviewed to form three IA options. These were then tree tested (more on tree testing vs. card sorting next chapter), and ultimately the final IA resulted in a 75% increase in content findability.

We’ve reached the end of this chapter, and you’re ready to go forth and run your own card sorting sessions! Ultimately, whether you’re redesigning a website, building a new product, or gathering ideas—these examples show how varied the application of card sorting can be, and how valuable the results are.

While all the examples above used card sorting in different ways, it’s overwhelmingly clear that card sorts are made far more manageable with the assistance of a digital card sorting tool.

Card sorting is a complex process with many considerations needed, and many chances for mistakes or missing UX best practices. Removing the risk of human error—and relieving yourself of the manual workload—can be a big support, giving you room to breathe and focus on the bigger picture.

Next in our card sorting guide, we’ll be delving into tree testing, how this differs from card sorting, and how it works in concurrence with card sorting. Read on to find out more.