Card sorting is a way to research your users' mental model: how they group different items based on their perception of them. It's a common technique when we're trying to figure out information architecture, navigation and taxonomies on a more complex product or website.
How to run a card sorting exercise
Who to test with
- Testing with 15 users should be enough to get pretty accurate perspective. It will give you a correlation of 0.9 between the results. 30 will give you 0.95.
- You must select users from your target audience, who match your persona.
A basic test:
- Create index cards with terms and descriptions
- Shuffle the cards
- Ask the user to group/pile the cards based on which ones they feel belong together
- Ask the user to give a 'name' or a heading to each of the different groups they have stacked up
- Repeat with other testers until you have 15 sets of results.
- Look for patterns while analysing the groupings made by different users
- Take the results of your card sorting tests into consideration during the IA design process
- This exercise can also be performed with pre-defined category names (buckets): we just ask the user to put the cards into the right bucket.
- Optionally you can also ask users to set a priority between different cards or categories.
A typical example
For an e-commerce site, write down each navigation item onto a card, and ask the user to group them. Card sorting will show you how the users want to see the different product categories grouped.
How will this effect the website performance? You'll hopefully end up with a more logical IA, which will help people to find items easier and result in better conversion rates.
Tools to use
- Remotely: Optimal Sort is our recommended remote card sorting tool.
- Fully manual: Card sorting can be done without any technical tools. Simply write the terms on index cards and ask the user to organise them. You can also use OptimalSort to print out and organise a moderated card sorting session, which may save you some time.