Getting Started : Card Sorting With Proven by Users
Card Sorting is a user-centered research method used to help determine how people categorize a set of items - referred to as 'cards' - to increase findability. The technique can be used to determine the both the naming of the categories and the grouping of the items within the categories. The most frequent use of Card Sorting is to establish an information architecture(IA) or navigation of a website.
Card Sorting can provide valuable insight on how a set of target users structure the content on a website. Among other things card sorting can:
- Identify the mental model of users and how they intuitively want to name and categorize the content on a website.
- Identify differences in the naming or structure by different user groups. (perhaps by age, or job function)
- Determine gaps in knowledge about areas of content within certain user groups.
Proven By Users is an Online Card Sorting Tool. The process of completing your card sort can be summarized in three steps
- Create your Card Sort Project. Creating your card sorting project is as simple as adding the cards you want sorted, adding the groups you want (if you want to perform a closed or hybrid sort), and adjusting the settings on the project.
- Email the projects' unique URL to your participants. Once your project is ready to go and set to be 'Live' it's time to have participants complete your card sort. On your home page you'll find a Participant URL to email. As Participants complete your test you'll see the results (page refresh is required).
- Analyze the Results. Once a participant has completed your test you can view, filter and download the results.
Let's outline the 3 types and when you might use each one:
In this type of card sort, participants create and name their own groups. Like the other types, the open card sort discovers how people group the cards into categories - how they structure the information. Additionally this type of card sort uncovers the terms people use.
When to use it: An open card sort is a very exploratory exercise. It is best used to discover the mental modal that users have of both the data and the terms they use to describe the information.
In a closed card sort, participants are provided the group names and they choose which of the pre-determined groups to place each of the cards into.
When to use it: A closed sort is good to use when there is confidence with the grouping and naming. The focus on this test is around which cards are in what groups. This is often done as a validation of the results of an open card sort.
As you might expect, the hybrid card sort is a combination of the open and closed sort types. In the hybrid, you create the categories, but the participant is allowed to modify the groups - adding, deleting or renaming them.
When to use it: A hybrid card sort is useful when there is confidence in the number and naming of the categories, but there is a desire to see where participants have a need to deviate from the pre-determined groupings.
A few considerations should be taken to determine how many cards.
- How much time you want the test to take. While it depends on the complexity of your cards, and the type of sort you are performing, a rough estimate is that 1-3 cards can be sorted in a minute. Keep in mind, this is just for the card sorting portion of the test - it doesn't include any survey, and larger tests do tend to take longer.
- The confidence in the results. It is frequently believed that the longer the test takes, the more burned out the participants become and the less effort they take to accurately categorize the cards. This can undermine the accuracy of the test.
- The size of the testing screen. While Proven By Users allows the participant to expand and collapse groups, and move groups around the screen, participants who are using smaller screens will need to actively manage the groups in large card sorts.
For this answer we will rely on the top experts in the research field.
Tullis and Wood recommend testing 20–30 users for card sorting
Jakob Nielsen recommends testing with 15 participants
These are great articles to read, and while exceptions to these recommendations exist ( ex: when not showing all users all the cards ) testing between 15-30 participants for your card sort covers most situations.