Friday, July 24, 2009

Card Sorting: Strategies

Recently, I had lunch with Adam Polansky and Stephen Thomas to discuss a card sort of our corporate Intranet. With the opportunity to design more usable categories, a cross-functional team has volunteered to do a card sort within the next few weeks. Our lunch was about the card sorting strategy for this project. We talked about a book, broke some bread, and created our strategy. Here is what we talked about.

Card Sort Strategies

We started our lunch with a brief chat about Donna Spencer's Card Sorting: Designing Usable Categories from Rosenfeld Media. In my humble opinion, this book is the definitive guide on how to perform a card sort, especially in various strategies outlined for each stage of a card sort project. Adam knows Donna, too. He told a funny story about Donna's IA Summit card sort.

Anyway, Spencer's book is refreshing in its step-by-step approach to card sorting. Tips and tricks are speckled in its chapter. At each step of the card sort process, Spencer provides real-world examples of card sort projects, where a lesson was learned (sometimes, the hard way).

My favorite real-world example was a card sort project, where Spenser prepared the project, but she handed it off to another person, who was ill prepared to execute the project. The results are both comical and tragic. The results were not usable. They had to repeat the card sort.

According to Spencer, a card sort strategy revolves around three key things: type of card sort (open or closed), participation strategy (individual vs team), and execution strategy (manual, software, or combination). We used this framework to make decisions on our project.

Type of Card Sort

As we all know, card sorts basically come in these two flavors:
  • Open card sort--where users group conent, then name the content bucket.
  • Closed card sort--where users place content into an already named content bucket.

The advantage of an open card sort is that you get to see how people would actually group the content together. The advantage of a closed card sort is you get to see how people place the content. It is an important distinction.

A frequently used strategy is to perform an open card sort to see how people group the content, then you analyze the results. In a few weeks, you can perform a closed card sort to verify if your content categories are correct.

Optimal Workshop's Treejack software provides a nifty alternative to doing a closed card sort. Users can actually drill into the category buckets seeing other choices, menu items, and groupings. It gives alot of context that you would not see in a closed card sort.

For our project, we decided very quickly to do an open card sort. The current Intranet categories are very shallow and old. As content had been added, it was placed in arbitrary places as the company expanded. Adam and Stephen already knew they wanted an open card sort. We still are not sure on whether to follow-up with a closed card or to use Treejack.

Participation Strategy

Basically, you are going to need to choose on how to execute the card:

  • Team sort--3 or 4 people discuss, decide, group the cards, and name the categories.
  • Individual sort--each person (working alone) peforms the card sort.

Team sorts can be very important in a number ways. For example, collaboration areas are ideal for matching content organization and category names used by a group. Spencer recommends that you take alot of notes during team sorts because the discussion of the group yields alot of insights (sometimes, the discussion is better than the card sort iteself).

Individual sorts are great when you want a large number and variety of results. In other situations, it is just impossible to get everyone together to perform a team sort. For the card sort of the Intranet, the team decided to use an individual sort (striving to get global locations included for a greater variety of input).

For the Intranet project, we decided to perform an individual sort, as the scope is to focus on the high-level content areas. Plus, we are not really looking into any of the collaborative areas of the site. Team sorting did not make sense for this project.

Execution Strategy

You can execute a card sort in one of three ways:
  • Manual sort--where you create index cards that your partipants manually use
  • Software sort --where you load the content into a software program that people use
  • Combination sort--where your sample includes a manual and software component

Manual sorts are low-tech, fun, and interactive. Spencer includes a wonderful Excel template for compiling results on the Rosenfeld Media site. Spencer provides a useful tip to number each content card. On the category label, you just enter the number of the content card. You can put the group to the side, if you need to continue working with your users.

Software card sorting tools have really improved. Some low-cost tools are now available that will help you to crunch the numbers. Most of the software exports results to Excel, where the number junkies and statisticians get to work their magic. Software-based card sorts are great because you can distribute quickly to large number of people. Plus, the software-based card sorts really do not take that long.

Combination card sorts occur when you send out an online card sort to one sample of the population and you perform a manual sort with another part of the population. You can get statistical significance with the online card sort, while you have the opportunity to listen and/or ask questions to manual card sort participants.

For our Intranet project, we decided to strive to do a combination card sort. We need to reach global locations, but we also want to hear some of the conversations. A software-based card sort does not give us the interative compenent of listening and watching. Plus, we may want to ask follow-up questions.

Conclusion

Spencer's book is an invaluable book to include in your library. We often think that card sorts are very to execute. Any card sort project needs to develop a strategy around the type of card, team/individual execution, and manual/software-based/combination style. Each card sort is going to be different. Let Donna Spencer's book help you to develop your strategy for your next card sort project.





1 comment:

Mabli said...

Very interesting post and blog. I appreciate your book recommendations also, since I'm working on reading/re-reading books on user experience subjects. Any advice for a seasoned web designer looking for a user experience designer job?