Tree Testing (Reverse Card Sort)

After you ran your card sort and got some better ideas of how your users think about categorizing the topics at hand, you probably created a sitemap based on your research. Once you are feeling pretty good about the structure you created, it’s time to test that structure with real users. This is called tree testing, also known as reverse card sorting.

It’s important to have a solid understanding of the required information architecture of a site before even getting started on designing a visual navigation system. Tree tests are useful for evaluating the information architecture of a site without the influence of a visual design.

Our favorite way of running tree tests is through a tool called TreeJack within the Optimal Workshop suite of products. You can set up a test for free (there are some limitations) if you don’t have a lot of available budget.

Instructions

1. First, you need to set up the test. You’ll need to plug in your sitemap structure. You can include as many nested items as you need to, though we recommend limiting the test to two or three levels of nesting. If you are working on a very large site, you may want to consider setting up different tests for different sections of the site to get deeper insights.
2. Create your tasks. We often use task-based scenarios to provide a little more context for the participant. For a flower shop site, the scenario might be: Imagine that you are shopping for daisies to send to your mom for her birthday. Where would you go to find daisies?
3. Assign the correct answers for each task. You’ll need to identify which nodes in the tree are considered ‘correct’ so that you can evaluate how successful users were at completing each of the tasks.
4. Figure out how you’ll recruit your participants. We often try to get a link to a tree test posted on the client’s existing site with a simple CTA to “Help us improve our website!”. We have also sourced participants using email newsletter lists, Facebook Ads, and hallway testing if it comes down to it. Obviously, your participants should meet the criteria for your target audience. If you need to add a screener question to your test to ensure you are capturing the correct participants, you can do so when setting up the test in TreeJack. For a tree test, we recommend targeting at least 15 completed responses.
5. Preview your study. Have a few colleagues try to take it to see if they run into any hiccups. Figure out roughly how long it will take so you can give participants a time estimate upfront.
6. Launch the study. Watch the results roll in! Once you have enough responses, close the study.
7. Analyze the results. TreeJack has some useful data visualizations to help you make sense of the study results. Find out which tasks participants succeeded on, and which ones they struggled with. Dig deeper into the tasks where participants struggled to find out where they went wrong. Is a label causing confusion? Are users split between two options? Identify a few key areas where the site structure can be improved.
8. Make changes based on your insights. Revise your sitemap in the areas it needs to be improved.
9. Test again. Did success rates improve? Verify that your changes improved findability. Don’t expect to get 100% success rates on your tasks, though. We generally aim for 70-80% success as a benchmark for a good, solid site structure. Remember that visual design and navigation elements will also help to increase findability for key content.

Learn more about ETR