Card sorting

Have a user sort a series of cards which each have one item written on them. This as traditionally used to help work out the right information architecture or navigation structure – but can also be used to sort list of features they into 'must have', 'nice to have' and 'don't want' groups. Card sorting can be done remotely or in person. See the notes below for some examples of how this has been used.

Good for:

  • getting really clear feedback on a large number of items
  • making user interviews a bit more interactive and therefore more engaging

Bad for:

  • some users find the drag&drop controls of remote card sorting tools to be hard to use

How to use

  • Write out cards (or virtual cards on an online service) for each item. These need to be concise and clear; you want every tester to have the same understanding of what they're sorting
  • Explain to the user what you want them to do. You can either give them the cards in a stack (good if there are lots of cards and if each item requires a little effort to understand) or place all the cards in front of them on the table.
  • Get the user to move the cards into the groups you've asked for. You can ask them to:
    • name the groups themselves, at the start, as they go or at the end
    • suggest the categories yourself (it can be helpful to write a category label cards in a different colour so they're clear and testers don't have to remember them)
    • leave the groups unnamed
  • Record the output. Online tools will do this for you. In person, you could take a photo of the card layout, or stack the cards from each group together.


Ideas/items sorted into groups, which may be identified/named.


  • What software or services could people use to do this?


How long does this take to do – both in elapsed time and active time.

What you need

  • people
  • equipment
  • space & wall-space




Kat, Katy


  • We used a three-round card sorting exercise when researching what alert content a news feed style product should have. In the first round, testers were asked to sort around 40 cards into two categories: "I want to know this" and "I don't want to know this". A photo was taken and all the "I don't want to know this" cards were discarded. In the second round, the remaining cards were sorted into "I'd want to do something about this" and "I wouldn't do anything about this". Again, the results were photographed and the "wouldn't" cards discarded. For the final round, testers sorted the remaining cards into "I'd know what to do" and "I'd need advice on what to do". The results helped us answer questions like "are there alerts that people would want to get but wouldn't prompt action?", and "what sort of advice would people want/need".
  • In the digital migration project, we asked customers to list the advantages of offline banking and the disadvantages of online banking, and then wrote each on a card. We then asked people to sort these cards into the order of which was the biggest influence on their experience. Doing this with cards made it much easier to get that ranking, as when you ask people to just say they usually forget or skip over some items, or are unclear.