What is tree testing?

You may have spent countless hours designing the look and feel of your website, from colors and fonts to logos and value props. But have you considered how the different pages of your site work together? When a customer arrives on your home page, will they be able to find the information they need? A tree testing solution is your ticket to designing websites that are easy for users to navigate and make finding relevant information a breeze.

What is tree testing?

The tree testing method—sometimes called ‘reverse card sorting’ or ‘card-based classification’—is a research technique that helps you understand where people get lost finding content or information on your website. In general, it provides actionable feedback about site navigation and content labels, so you can make more informed decisions about how information is labeled and organized.

Tree testing software shows contributors a text-based version of your website or app, with no navigation aids or design elements, and asks them to indicate where they would find specific items or topics. This means you can spin up a test in minutes, making it as simple and quick or as robust and iterative as you need—without having to rely on multiple teams to get it done.

UserTesting customers can complete this course on tree testing. 

Why tree testing matters

Imagine you’re visiting a friend’s house and need to find something—like a spoon, for example. You head to the kitchen because that’s where we’ve all been trained to find things like spoons. But which drawer do you look in first? We use context clues and other points of reference (like familiarity and habits) to help us find what we’re looking for. 

The same goes for navigating a website or app.

However, unlike the spoon example, there’s much more variation in website structure than kitchen layout. Websites vary from industry to industry, ranging in purpose from ecommerce to digital portfolios and everything in between. So when someone lands on your website to find something or complete a task, is it obvious where they should go to get things done? 

Your gut might tell you that it is. Your design team might tell you it is, too. But what really matters is what happens when actual users interact with your site. Whether you’re building a new website or optimizing an existing one, it’s important to have quality user insights. That starts with user experience (UX) research, and includes tree testing.

Note: UserTesting information architecture (IA) features, such as tree testing, are included in our Ultimate subscription plans and can be purchased as add-ons for Professional, Premium, and Advanced plans.

How does the tree testing process work?

Tree tests provide extremely powerful results and are fairly simple to set up. To conduct a tree test, you don’t need to sketch any wireframes or write any content. All you need to do is prepare the tree , or hierarchical menu, and the tasks—instructions that explain to the study contributors what they should attempt to find.

To begin, we recommend most users start with the Tree Testing template available in the Template Gallery. 

1. Build your tree

Your tree is a text-only version of your website hierarchy (similar to a sitemap). If you don’t already have access to a sitemap, you can build the tree from the labels and structure of your website.

When building your tree, it’s important to include a complete list of all your main categories and all their subcategories. Even if you’re only going to test a specific category, excluding others will bias your results. 

To go one step further, your tree may need to be three or more levels deep, depending on your site’s architecture. That’s ok, but it’s important to include all of your page’s subcategories, or you’ll risk eliciting unrealistic behavior. Users often evaluate link labels by comparing them with nearby alternatives.

Pro tip: Exclude labels like ‘Contact us’ and ‘Search.’ They may provide easy ways for people to find information on your site outside its natural hierarchy.

2. Write your tasks

Your tasks should reflect what you want to improve. Perhaps your support team has received feedback that people can’t find a specific form on your website. In this case, you’ll want to create a task instructing contributors to find where the form is located within the tree you’ve provided.

3. Recruit contributors

The quality of contributors taking part in your test will directly influence the quality of your data, so it’s important to take the time to parse out who it is fully you should be testing with. You want people who are as close to the right demographics as possible and willing to take the activity seriously. See how the UserTesting Contributor Network ensures high-quality feedback.

To get qualitative feedback when running a tree test, it’s best to opt for a larger sample size than you would when conducting a standard test. This is because you want to derive meaning from the number of contributors who complete each task in a particular way. We recommend distributing the test to 30–50 contributors to gain statistical confidence in your data. Additionally, you’ll get quite a bit of value from watching the videos of just a handful of those contributors.

Analyzing the results of tree testing

UserTesting’s tree testing tool provides you with both qualitative and quantitative results.

At a glance, here are the most relevant metrics you’ll encounter when analyzing the results of a tree test:

• Task success rate measures how often the right category was selected by contributors for that task. Arguably, this is the most important metric because you want customers to find what they’re looking for successfully.
• Path taken measures the paths taken to the right or wrong selection—especially useful for understanding if there are patterns in the wrong responses.
• Time spent on task can be very telling of where users struggled.
• Video feedback helps you understand the source of confusion, struggle, delight, and other sentiments felt during the test—great for identifying actionable improvements.

Although tree testing provides a wealth of quantitative data, the inferences you may draw from them are inconclusive. Task success rates and time spent on tasks help you to identify what’s going wrong, but they don’t explain why it’s happening. This is where the video feedback comes in handy.

Suppose you use analytics as a starting point for identifying problem areas and pair them with the moments they happen in the video. In that case, you’re much more likely to gain a holistic understanding of the problem. From there, you should be able to draw some conclusions for improvements or set yourself up for further testing.

Next steps

Check out our full tree testing tutorial for a complete process walkthrough.