Improving Task Completion Through Tree Testing
How to help site visitors complete tasks by using labels that make sense to the audience.
As a strategist at Palantir, one of the implicit goals I address on projects is to improve people’s ability to find information and complete tasks on clients’ websites. Often the biggest area for improvement is the site’s main navigation. Main navigations are either crammed full of options or pared down to not enough, and frequently they use labels that are understood by the organization but not by the audience for whom the site is built.
The best way to improve task completion is to use the audiences’ own words as labels in the navigation. Using their own words shows commitment in meeting their needs, prioritizing how they view the information over how the organization views it. It also means eliminating jargon and keeping internal org ego at bay.
So how do you know what words to use? Together with top task identification and card sorting, the way to uncover audience-centric navigation is through tree testing. Tree tests are a way to evaluate a proposed classification structure: both the organization of the pages and subpages, as well as the labels used to describe them.
How Tree Tests Work
Tree tests ask participants to point out where in the navigation they would expect to go in order to complete different task scenarios. The tree structure represents the proposed parent-child relationship of the website’s navigation.
I conduct tree tests with a tool called Treejack from Optimal Workshop. With Treejack I can set up multiple studies to test which classification hypothesis leads to greater success, ask pre- and post-test questions to make sure I understand where participants may struggle to complete the task, and see the results in real-time, in case I need to recruit more people to take the test.
In order to draw conclusions, I wait for a minimum of 15 participants to complete the test. Patterns in the results stabilize after about 15 participants, so that’s what I aim for when starting to draw conclusions. I know a label is working if the success rate on a task is between 80-90%. After the first round of testing, the average success rate is around 60%, so multiple rounds of testing is always preferred.
Benefits of Using Tree Tests
Why does tree testing work? The secret is the simplicity and straightforwardness of the test. Tree tests are focused on the labels used in the navigation; you don’t need anything more than words, so it’s easier to get up and running compared to a usability test that could need anything from wireframes to functional prototypes. There’s no need for visuals, and you only need 15 participants to move forward in a meaningful way.
You don’t need to be redesigning your entire website in order to conduct a tree test. You can even do it with your existing navigation structure—which would be a great way to have a baseline understanding of how well people can navigate your site. When you combine tree tests with additional user research like top tasks identification and card sorting, you get a better understanding of your audiences’ expectations and mental models. Changing your navigation to fit those expectations will improve people’s ability to complete tasks faster. This better experience will lead to an increased level of trust in your organization.