A reverse card sort, or tree test, is the best way to assess an existing structure. In this video, get a quick overview of planning and running tree tests.
- A tree test is basically the reverse of card sort. Used to verify structure, the team end up putting together or to assess an existing structure. In a tree test you show participants the elements of the site or product structure. Named exactly as you're proposing and give them tasks to perform. For instance, you might ask someone where in an ecommerce site they'd find soccer cleats. You want to be sure that you include the individual pieces of information that you're asking people to look for as elements in the tree.
In the searching for soccer cleat example you want to make sure that soccer cleats and a few other individual products are including in the tree setup. Rather than just the category or categories that the cleats live in. You'll also probably want to exclude wayfinding items like search or help from the test tree. These types of things are helpful when real users are interacting with your structure. But won't give you much information about how well the other parts of the structure work. The tasks in your tree test should be based on the real goals that users have when using your product.
You can glean these from other types of research such as interviews, and diary studies. You might also create tasks based on where you know people are currently having navigation issues that you heard about from support logs, or observed in a usability test. You structure the tasks much like you would in the usability test. Mapping to the goal that a person has. For example, with the same soccer cleat task, you might come up with something like, your son is about to start a new season of soccer, and grew out of his cleats last year.
Find a pair in his size. You have to be careful with language in the tasks just like with the cards or the wording of any other type of UX research task. You don't want to lead participants. You want to exactly map the language in the questions to the structure you're testing. For instance, if you're working on a clothing store website, you may have a structure labeled something like women, shirts, active, sleeveless. But in the task, you'd want to say something like, find a tanktop to work out in.
Rather than, find a sleeveless active shirt. Reviewing and analyzing the results of a tree test is much simpler than analyzing card sort data. Essentially, you look to see how many of the participants were able to successfully complete a task. And how long it took. If at least 75% of participants can find things in a reasonable amount of time you can conclude that things are findable. When participants can't complete the tasks you'll want to look at the paths that they end up taking and see where participants went.
If there are patterns there and people consistently went to the same wrong spot you'll likely want to move the item. If people were all over the place it's a clue you need to do some follow up research to understand the source of confusion. One other note. If you've moderated sessions, you'll also need to spend time gleaning qualitative insights from the discussion around the tree test, or card sort. To best analyze back data I'd recommend following the general qualitative research analysis process described in the UX Research Fundamentals course.
Create a spreadsheet with each participant and their pertinent information and notes. Then mine the data for individual facts, quotes, and points of interest or confusion. Once you've identified individual findings you can do a card sort, or affinity map of the insights to group like items. This will help you to see where there is lots of overlap, and where the insights diverge. This process is especially valuable when done as a group with all team members and stakeholders so that you engage all parties, help them see the value in the research, and don't miss any points of view.
- 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