This article is re-posted from User Centric’s blog.
Best Practices for Card Sorting
When clients discuss their objectives and say things like “We don’t know what should go where on our website,” we suggest card sorting as a possible strategy to organize content. Card sorting allows companies to identify similarities in website content. However, we find ourselves explaining card sort methodology and, even more so how to create labels and descriptions used in the data collection.
We know card sorting results are based on participants’ individual answers, and that outcomes can be very subjective. Therefore, deciding on and perfecting the card labels, descriptions and (sometimes) category labels must happen before data collection starts. If the wording on these is not done just right, then the data that is collected will be not as valuable as its potential.
We have scoured the Internet for “card sorting labels do’s & don’ts” and “best practices for card sorting descriptions” with only limited success. There are plenty of overall guidelines available, but we feel it would be better if we could provide a reference in conjunction with our personal feedback.
With so much pressure relying on study design, we’ve had to develop our own guidelines. Below are recommendations we share with our clients when creating card labels, descriptions and category labels:
In General:
Cards should be comprised of a representative sample of the website. Consider using the most popular pages or those pages that represent a large percentage of the overall content.
Content cannot be too varied (or too similar) or participants will not be able to create natural groupings. Users must be able to easily sort the content into more than one category.
Be sure that labels and descriptions across cards have enough similarity, but do not repeat words, phrases or names too often between those labels and descriptions. When conducting a closed card sort, it is important that the categories also do not contain similar words or phrases as the cards. Participants will simply match words or phrases together instead of thinking about the overlying groupings.
Card Labels:
Card labels should be the primary description of the page. Do not count on participants to read the descriptions. Titles need to provide enough information for the participant to make an informed decision.
Labels should be as short as possible while still being descriptive. However, keep in mind that participants still need enough information to make an educated decision. If a restriction is needed, we suggest no more than 10 words, but preferably much shorter.
All cards must be either actions or descriptions, instead of a mixture of both. It is easier for participants to sort cards with a similar phrase structure.
Card Descriptions:
Again, be descriptive yet concise. Participants find it exhausting if they have to read long descriptions.
Card descriptions can be a little longer than the labels. However, continually look for ways to shorten the text—never more than one sentence.
Descriptions should provide additional details about the page or section of the website than given in the label. Do not simply restate or reword the card label.
Closed Card Sort Categories:
Titles should be as short as possible and should closely match the titles intended for use on the website.
Like cards, categories must be distinct enough for users to characterize the difference between them.
We use several methodologies to uncover users' mental models. We’ve found that card sorting is only effective when measures are taken to develop quality cards prior to data collection; methodology and preparedness of card sorting can unintentionally alter data outcomes.
Companies don’t want their hand held, especially if user-centered objectives are embedded in a product life cycle or website. However, the return on investment is too valuable to ignore expert advice.
What advice have you given a company looking to use card sorting research to redesign a Website? What challenges did you face?
Read about our User Experience solutions.
