Card Sorting
In card sorting research, users are asked to arrange representative items of content (typically represented by a title and short description) in groups that make sense to them, and then suggest names for the groups they have created. If appropriate, they may be asked to create higher level categories for the groups. This can be done with specialist software, or with actual cards on a table. Both work well.
So far so simple. Sometimes card sorting is conducted remotely – users simply log in, do their card sort, and go. Sometimes it is done in groups. We don’t recommend either of those approaches.
If you do card sorting badly, you are simply asking people who know little about your site, or information design, to develop your site structure. Then averaging out the result. Not a recipe for success.
Understanding users’ choices
We believe the real value in card sorting lies in understanding why users make their choices. Have they understood your content labels correctly? Is there a particular need that leads them to place certain items together? Can they use their own category labels to find items? Where might they want to see cross links? What would be key content for them that they would want to see on a home page?
That’s why we recommend doing card sorting in one to one sessions. We always make sure we recruit respondents who match your target audience profiles. We take them through a seven step process that explores all the questions above.
We do the quantitative stuff – placement to labels, common words, cluster analysis. But then we overlay this with in-depth qualitative analysis in order to develop a recommended IA structure.
Done this way, card sorting not only delivers a draft IA structure which matches customers’ perceptions, language and needs, but also delivers rich data for cross linking, faceted IA development and quick links.
