When it comes to web design, one of the most challenging things to resolve is properly grouping and categorizing labels in a way that users can easily understand.
And in UX, you can apply a fast and effective technique called “card sorting”. However, for best results, the card sorting technique should be followed up by a tree test to properly evaluate the proposed menu or site structure on the web page.
In this article, Tree Testing 101: A Comprehensive Guide, we will further discuss what this tree testing is and how it can help us better organize website menus and labels.
In addition, we will also discuss the following sub-topics:
We have mentioned that one of the best ways to organize and group information on your website or app is the card sorting technique. This method is where users are given a list of content items to group and label as they see them fit.
We conduct card sorting to get a glimpse of how users think. However, this method does not necessarily mean you get the exact and proper organization scheme.
To explain this further, let us take an example scenario where participants are asked to hold on to items that they find unfit. If you were to apply the same logic on a website and place these unfit items into the “other stuff” category, then the same users may avoid this menu since it has found out that users are highly suspicious of vague labels and they avoid clicking on these items simply because they suspect to exert a lot of work understanding what the content is in the first place.
Therefore, to properly organize content items on a website or app, we recommend that you combine the card sorting technique with tress testing.
What is tree testing?
So, what is the tree testing method about? Tree testing, or sometimes known as reverse card sorting, is a research method that is made to find and assess the proper hierarchy of things within a website or app.
In tree testing, you only need to place a text version of the site’s hierarchy or site structure. This is laid out in front of the participants to highlight a category or page within that structure where they expect to find a particular item or piece of content.
Tree testing vs card sorting technique
In card sorting, participants are tasked to group the content items based on what makes better sense to them.
In tree testing, participants are already given an existing set of categories and all they need to do is to highlight a section where they believe an item is most likely to be located.
Ideally, the card sort method should come first before the tree test since tree testing is a good validation of site structure results from your card sorting exercises.
The benefits of tree testing
Easily identifies navigation issues
Tree testing identifies any navigation issues on the website or app. Like for example, you can analyze sections where users will likely find the information they are looking for.
Validates the effectiveness of the site or app’s organization, structure, and labeling
We know that card sorting is helpful in organizing content items and labels, but this is not the end-all solution when it comes to a proper organization, hierarchy, and labeling of items. And as suggested, this is where you want to test your site structure and conduct a tree test.
Presents realistic scenarios
Compared to other information architecture tests, tree testing offers participants a more realistic environment and thus, can produce real-world behaviors.
Quick and easy
Tree testing sessions are short and easy; thus, it is quicker to conduct test variations of a website or app structure and compare the site structure results.
Cost-efficient and can be done remotely
You can remotely conduct tree testing, which reduces the tree tests costs
The disadvantages of tree testing
Focuses only on the structure
The focus of tree testing is the structure and therefore may not be an effective method for other areas of navigation.
Very basic in form
And as such, the tree test method disregards the visual elements that may help users in their navigation.
Often conducted remotely, and thus unmoderated
Since most tree tests are conducted remotely and tree tests are often left unmoderated, this may not capture the full form of user behavior like user comments that are made while taking the test and you cannot make follow-up questions as well when it comes to remote tree tests.
What tree testing looks like
Tree testing has two main elements:
The tree is a text-only version that is like a sitemap. The tree test starts by asking participants to click through the tree and pick the content item they think is correct.
The results of this test will tell you the following:
- The percentage of users who got it correct
- The percentage of users who got it wrong
- The paths that participants took before they picked an answer
- The time it took the participants to complete the task
An example of a tree looks like this:
The task looks like the images below, where participants are asked to pick a content item and see the top level of the tree.
Once the participants completed the tree test task, you can see the results like the one below:
When do you use tree testing?
As we have previously discussed, tree testing helps and is useful when you want to find out if the labels, structure, and information architecture of your content items are properly organized and most of all, easy to understand for your users, a better user experience success rate indicator.
You can get some good user research insights at all stages of the design process, whether you are starting from scratch or making some changes to your website or app.
You can conduct tree testing on big websites with large structures or you can do tree testing for smaller structures.
You can even test any size in between, you can set up one or multiple studies or different versions, or you can set 1-10 tasks at a time. Tree testing can be done in varied ways, different versions and sizes and the potential can be endless when you want to get user insights through tree test.
How to conduct tree testing?
Now we know that tree testing shows a basic menu structure to users without any added layout and design. The users are asked to complete a task or series of tasks, where they look for items using the site/app structure.
The typical tree test can run from 15-20 minutes. On average, real users are given 15-20 tasks per person. This is done this way since users tend to lose focus if tasks are given for too long.
Generally, tree testing is initiated early in the design process. And as mentioned, you can conduct card sorting first so you can validate the user research results when you conduct your a tree test.
Tree testing is conducted remotely. Users take the test on their own computers, unmoderated.
You may check some useful tree test tools that can help you run unmoderated UX research testing.
Analyzing the results
It is easier to analyze the tree testing results compared to the card sorting results. Furthermore, available tree test tools provide you with clear UX research visuals for each task you have assigned. You can quickly see the problem areas when it comes to the site’s structure.
When it comes to interpreting tree test results, you should look for the following:
Directness the percentage of participants who completed the task without hesitation and getting the correct answer the first time
Success rates- is the percentage of participants who complete the task vs the participants who failed
Time- the time it took participants to complete the task
You may also look into the number of participants tries to complete the task. This data indicates the items that are difficult to find.
With these results, you can easily identify the problems with your structure and labeling.
Once you have analyzed the tree test results, it is now easier for you to make some needed changes to your site or app structure.
This test research helps you make changes to make a better structure, navigation, and labeling of items on your site or application.
If in some cases, the insights you have gathered are not enough, then you can conduct a tree test the second time, and this time it should be moderated so you can ask the participants some personalized follow-up questions after.
Conducting a moderated testing tree can provide you with more qualitative insights. We suggest, as much as possible, that you conduct one-on-one testing for more quality results.
Since tree testing only presents a text-only version of your website or app, this removes everything that may become visual destructions. Users can focus more on the navigation and it can bring you good user behavior results.
Thus, it is recommended that you have a complete list of all your main content categories and subcategories. Even if you are only interested in specific navigation, you should not exclude the other items since this will assume that users already know which sections to go to.
Another recommendation when it comes to testing tree is to present 3-5 levels deep. This allows users to go down to the lowest levels. This will provide you with insights that you need in the “real world”, where your audience will be exploring.
Remember to keep your testing within 15-20 minutes only. You also want to test and limit users to 10 tasks. Any more can drain or bore them, which can affect user behavior. To exclude users, while maintaining good results.
We have learned that tree testing is exclusive only to evaluating category labels. This can be both a strength and a weakness.
As a weakness, the text-only version completely devoid of any visuals and design that also serve as significant factors in helping users interact better with the website or application.
As an example, a menu with a bigger and readable font style can provide quite a different browsing experience compared to the one tested in a tree test.
But even with these limitations, you can easily overcome or minimized these with careful data analysis.
Like for example, you can focus more on the item that your participant selects, rather than getting the success rate of sites with better, bigger font menus.
With that, we can say the tree testing limitations are just a small thing to even be bothered about as long as you know how to properly get your results.
Overall, tree testing is a good research method to iterate and evaluate major structural changes within your site or app’s information hierarchy early in the design process quickly and efficiently.