Tree testing

Tag(s)

Research & testing

Cook(s)

Rebecca Blakiston

Nutrition profile

Tree testing helps you validate your website's information architecture, which is often reflected on menus and labels in a hierarchy of navigation. By asking users to complete tasks and click on labels to navigate, you'll evaluate how findable your content is. Since you're just testing menus and labels, you can run tree tests before developing complete content on your website.

Snippet from an information architecture outline for a library website, including main, tier 2, and tier 3 navigation

Cooking time

Tree tests are speedy! While it depends on the number of tasks you ask participants to complete, expect no more than 15 minutes per session.

The analysis of the results may take some time, and we recommend running subsequent tree tests to continue to refine your navigation.

Perfect for

Developing or improving upon a website's information architecture. Tree testing evaluates the hierarchical structure within your navigation, and works well as a next step after card sorting.

We recommend using card sorting as a first step to inform high-level categories of content, followed by tree testing to evaluate task-based navigation and test out deeper levels within your structure.

Prep work

Prepare your menu labels

You'll need to have a detailed information architecture ready for testing. You can put it in a spreadsheet or directly in the tool you're using for the tree test.

💡

Information architecture usually refers to a hierarchy of menu items and labels in a website's navigation.

Tree testing is most useful when you have a hierarchy where participants will need to click several menu items to complete the task being presented. For example, a library website's menu might be something like this:

Example of a draft website menu for an academic library, 2019.

Write out your tasks

You'll write out tasks similar to how you would write scenarios for a usability test. Avoid using leading language that's reflected in your menu labels.

We recommend testing no more than 10 tasks, and testing those tasks that are:

Most important to users (your "top tasks")
Potentially more difficult to locate

Example tasks for the academic library website above could be:

Learn how you can borrow out a laptop
Reserve a quiet room
See how long you can borrow books
Make an appointment with a librarian

Define the correct answers

Each tasks should have one or more correct answer where you expect users to go. By defining the correct answers, you'll be able to measure success rate and evaluate your information architecture.

Ingredients

Around 50 participants (for an unmoderated study) and incentives for those participants
An interactive prototype of your menu structure that allows participants to click or touch and land on different screens (try a free app such as Marvel POP) OR
A tree testing software such as Treejack, UserZoom, or UX Tweak

Directions

Similar to usability testing and card sorting, you can conduct moderated or unmoderated tree tests, and they can be conducted in-person or remotely. The benefit of moderating tree tests is that you can ask participants to talk aloud or explain why they chose a particular navigation route. Unmoderated, on the other hand, provides only quantitative data but allows you to reach a larger number of participants more quickly.

Unlike usability tests, instead of showing the participant a live website or screenshots of the website, you only need to present the participant with its navigation hierarchy. This can either be the "tree" interface generated by tree testing software, or an interactive prototype with the site's menu items and labels.

During the tree test, you'll ask participants to complete a task and then observe their behavior. Either manually or using software, track:

Navigation path taken
Success or failure
Time on task (optional)

Plating

Tree test findings can tell you how successful your navigation is in supporting users completing tasks. Your goal for each task should be:

high success rate (most important)
high direct path
low time on task

Ideally, you will have most participants use a direct path to complete the task successfully. If you aren't happy with your success rates, you might need adjust your menu labels, shift content around, and/or cross-link certain content in different places (i.e. positive redundancy).

💡

Positive redundancy refers to redundant content that benefits users. In the case of information architecture, you might have some content that should be accessible from multiple navigation points to improve its findability.

Color coding to describe navigation paths of participants

Treejack and similar tools can produce useful data visualizations that reflect navigation paths taken and success or failure rates. These can be especially helpful when sharing your findings with stakeholders.

Results from Tree Jack for the task on a library website, "See how long you can borrow books." In addition to the color coding, the thickness of the lines indicates how many people took those navigation paths.

Example of Tree Jack results for "Get help with a citation in a research paper" with three correct navigation paths.

Pro tips

Do a practice run to make sure your tasks are worded clearly and likely navigation paths are reflected in the study.
Prepare to conduct multiple tree tests. In response to your findings, iterate on your architecture based upon what you discover and conduct more tree tests until you reach a success rate you are comfortable with.
Depending upon your content and audience, you may want to do targeted recruitment of participants. For example, asking students to locate instructor-related resources might lead to less helpful results.

Resources

Tree Testing 101 by Optimal Workshop