Home > Planning > Site Content > Card Sorting

Card Sorting: Card Sorting process

What it is:

Card sorting is a common usability technique that is often used to discover how users group items, so that we can develop a structure that maximize the probability of users being able to find items.

Advantages of card sorting:

• Easy and cheap to conduct
• Enables us to understand how “real users” are likely to group items
• Identifies items that are likely to be difficult to categorize and find
• Identifies terminology that is likely to be misunderstood

What is Needed:

• Identification:
  Before conducting card sorting exercises, make sure you have identified all items that need to be categorized.
• Location:
  Make sure testing area has adequate space for spreading out the cards.
• Participants:
  Set up meeting times partner and participants. Make sure the participants are actual users of the site. You can have as many as a room full of participants or just two. The point is to gather a large amount of data from ALL your targeted users. One data sample from one audience type and three samples from another audience type will distort the outcome. Try to keep the groups evenly proportioned.
• Creating the Cards:
  Write each of the identified items on a note card. (We use laser labels when creating multiple decks of cards.) Vary your writing into statements, questions, one-word descriptions, multiple word descriptions, etc.

Helpful Hints:

• Do Not create more than 75 cards and make sure there are no duplicates.
• Ensure each term is as clear and unambiguous as possible.
• Make sure you have included all the items you need to categorize.
• Shuffle each deck thoroughly before handing them out.

Prior to Meeting:

• Print out two card stacks more than requested (just in case).
• Bring extra pens, post-it notes, evaluation forms and blank index cards (20+).
• Bring script of instructions so all participants have the same understanding of the process.

Beginning the Exercise:

• Introduce yourself (if not done already).
• Give a brief background of the Web site.
• Explain the type of users you are targeting with this site (this puts the participant into that mindset, considering many users wear more than one hat).
• Briefly explain the purpose of this exercise.
• CLARIFY that the exercise is not to test them, but is feedback for us on how they group information.
• Do NOT allow more than one participant per card stack if at all possible.
• Tell the participants to just relax and take their time.
• Estimated time: 45 minutes with 50 cards

Scatter the cards around a table in random order. The users should sit down at a table and sort the cards into piles according to similarity. Discourage users from producing piles that are too small or too large. Remind them that the stacks don't need to be even.

After a user has sorted the cards, ask them to group the piles into larger groups that belong together and to invent a name for each group. Have participants write these names on Post-It notes and place them on the table next to the groups. The time it takes to complete this can vary depending on the amount of note cards used.

Questions participants may ask:

1. What if I have no idea what a link name represents?
   Feel free to set aside cards that have no meaning for you and leave them out of the exercise.
2. May I put cards in more than one category?
   Yes. Make an extra card by copying the name on one of the blank cards and put one card in each category. Make a note on the cards that says "also in [category x]."
3. I have some ideas for related or new content I'd like to see on your site. May I include those somehow?
   Yes, just use the blank cards to make up new links of your own choosing and add them to the piles.
4. What if some of the links don't seem to fit into any category?
   Feel free to make a category called "other" or "general" or whatever.
5. Are you really going to allow offer( name of site or service yet to be offered) ?
   (or a similar question from a link name for something we don't have yet)
   Some of these items are from ideas that people gave us for what they would like to see on our site. We haven't made final decisions yet about all the content we will include, but we would love to get your input about which features are most important to you.

Analyzing the Cards:

This exercise will yield a list of names for each group of cards with subgroups corresponding to the original piles. Based on this information, it is possible to calculate similarity ratings for the various titles given. Give a pair of titles one similarity point for each time both titles occur in the same larger cluster and two points for each time they occur in the same pile. Sticklers for exact statistics can feed this similarity matrix into one of the many standard statistics packages for a cluster analysis.

However, it is not necessary to be a statistician to use card sorting. Navigational structure can be created based on "data eyeballing" and not on formal statistics. If it is necessary to apply a discount usability testing approach by testing only four users, the statistical methods would not be reliable.

For manual clustering, work bottom-up and group concepts that most users sort together. Expand these small groups into larger clusters. Do this by using concepts that some users sorted with most of the concepts in the group, but only if the grouping makes sense.

This is subjective and could be considered questionable if the objective "truth" is desired. However, if the goal is a coherent design, it is justified. Sometimes user data is too sparse for clear conclusions to be drawn on the basis of the numbers. Eyeballing the data gives you a list of top-level navigation items and initial names for the categories. The names should be inspired by the labels provided by the users. Since the users will not use exactly the same terms, feel free to apply judgment in choosing appropriate names.

Tool Kit