You'll get the best card sort results when you carefully plan the content that is to be sorted. In this video, learn how to craft cards so that you receive the least-biased and most helpful results.
- Once you've decided what type of test to run, you need to spend some time constructing the cards that participants will be sorting and planning the overall session. First, you must ensure that the cards are of a similar granularity. You can test anything from individual components, whole pages, groups of pages or areas of a product, as long as you keep everything consistent within a test. For example, let's say you're working on any commerce site and want to run a closed card sort to see where a new individual product, like a pair of soccer cleats, should live.
In this case, one of the cards would be the new element, soccer cleats, and the other cards would also be other individual products, such as a pair of golf shoes, a basketball or a hockey stick. If you start mixing in cards that represent larger categories, such as shoes or fitness apparel, you can end up leading participants to specific groupings, or make it hard for them to create meaningful groups. Another very important component of preparing the cards is ensuring that you use clear, precise terminology that doesn't lead the participant in any way.
You want to make sure the users clearly understand what the item is, and you can provide a brief explanation if necessary. However, you also don't want to be so consistent that there's a natural bias. I'll borrow an example from Jakob Nielsen to demonstrate. Let's say you're trying to determine the structure of an agricultural website, and you could either organize by activity, such as planting, growing, harvesting or cooking, or you could organize by crop. If you label the cards something like plant corn, plant wheat, plant potatoes, grow corn, grow wheat, grow potatoes, you're very likely to get results that show participants organize by the activity.
Similarly, if you label the cards something like corn planting, corn growing, corn harvesting, wheat planting, wheat growing, wheat harvesting, you're likely to get overwhelming responses that support categorizing by crop. To get around these biases, use synonyms as it makes sense. For instance, one card could say harvest corn, and another could say pick wheat, since harvest and pick mean essentially the same thing. You just have to be careful not to go too far to find synonyms and end up confusing the results, because the synonyms aren't really close enough, or there really is only one correct label for something.
For instance, if you're talking about potatoes, you might be able to use a word like spud, but you may not be able to find another word for rice. To help avoid bias, you can also use something called non-parallel exposition structures, where you mix up the wording to be purposely inconsistent. For instance, one card might say picking corn, and the card for wheat would something like wheat picking. When you actually label the content in your product, you'll want to be consistent, but using this tactic in testing can help you avoid overly biased results.
You'll also want to consider the audience when choosing card wording. While part of the purpose of card sorts is to explore people's mental models, you want to try to match the terminology to your intended audience. At the very least, you'll want to avoid using terms that they never would. For example, if your target for the agricultural site is scientists, go ahead and use binomial names of crops, but if your target is consumers, most probably won't know the Latin names.
- What's card sorting?
- Open, closed, and hybrid card sorts
- Card planning
- Category planning
- Finding, selecting, and screening participants
- Scheduling and incentivizing participants
- Running sessions