Using tree testing to validate your information architecture
Developing a site structure is one of the most difficult and important parts of creating information architecture. And the larger the site, the harder it gets.
10 October 2016
How content is organised and how it’s labelled it can make a big difference to a site’s overall usability, so it’s important to get feedback from users as early as possible.
Tree testing tells you how easily people can find information on your website, as well as highlighting where they have trouble. By showing users your menu structure in its most basic form you can quickly and simply test whether it works for them before going too far down a particular route.
You’re not asking people to find particular pieces of content (as it probably doesn’t exist at this point), but instead you’re asking them to indicate where they may look.
Tree testing is great at answering questions like:
- Do the labels make sense?
- Is content grouped logically?
- Can people find content quickly and easily? If not, why not?
What do you want to know?
The clearer you can be about what exactly you want to find out, the more successful the test will be.
Are you looking to test a radical new restructure, or are you wanting to improve just one section? Do you want to test specific journeys for your audiences?
Who will you be asking?
To run a test, you need people to participate - ideally ones who form your sites audiences as they will be the ones using it. The more people you survey, the lower the margin of error , but you can still get fairly meaningful results with smaller numbers.
Face to face or online?
It’s possible to run tree tests manually, face to face, or online with a web-based tool.
Face to face. Doing it this way will probably end up with better results as you’ll have the opportunity to explore with users why they chose a particular option and get a greater understanding of their thought processes. Be prepared to use a lot of index cards.
However running the tests and compiling the results will take time, limiting the number you’ll be able to do and potentially negating the benefits of being able to get feedback quickly.
Online. Through a service like Treejack, you’ll be able to set up and run a test very quickly. They’ll even recruit testers for you if necessary, and the tools to analyse the results are pretty good.
The downsides are that you’re limited to the information users give you, and there’s little opportunity to get the deeper understanding that a face-to-face test might.
Setting up the test
Create your tree
The tree is a text-only version of your website structure. It doesn’t mater if it’s still in draft. You can create and test multiple versions, or if you’re not sure where something is going to go yet, you can put it in a number of places and see which one people find most.
Your tree should look something like this:
- About us
- Our mission
- Our vision
- Our team
- Our board
- Our staff
- What we do
- Service 1
- Service 2
- Solution 1
- Solution 2
- News and events
There’s no limit to the number of items you can add, but if your hierarchy is really large it may be a good idea to split it into a number of smaller tests to make it more manageable.
Creating the tasks
Next, put together 8-10 tasks you wish people to complete. This can include finding important pages, or completing a particular user journey. You’ll need to choose at least one correct destination for each.
Think of them as hypothetical scenarios. You’re looking to follow the sort of thought processes a site visitor may have. So instead of ‘Find the events listing’, it should be ‘Your colleague has told you about an upcoming event and you want to find out more.’
Be careful not to give away the answer in your question. Use different wording in the question as to what’s in the tree so you can be fairly confident that people have understood and are not simply matching phrases.
Stay tuned for part two: running the test and interpreting the results.