Information Architecture: Why and how to base it on real data

One of the more complicated concepts in User Experience (UX) is Information Architecture or IA. At its most basic level, an information architecture is like a multi-dimensional map of how you store and give access to information. It’s how you decide to arrange your information so that people can understand what you have and how to get to it.

People find what they’re looking for on your website by using your information architecture, so you want the arrangement to match your users’ context and knowledge base (their “mental model”) as much as possible. You want navigating your website to feel like common sense by making each click decision easy and intuitive for your users.

For example, if you arrange your website around your organizational chart, you’ll likely find that a lot of people are unable to find what they need because they don’t know which department handles their request. They are more likely to know what task they want to accomplish or what topic they want to explore.

If you’re creating the website for your state’s Department of Transportation, you wouldn’t organize the site around the DoT’s departmental structure. You’d instead prioritize information about construction projects, road closures, public transportation opportunities, toll pricing, etc., because your research shows these are some of your users’ primary tasks. Organizing your content according to your users’ mental model allows them to easily complete their task. Common sense, right?

Much of our work at Anthro-Tech starts in researching what users need, what their tasks are, what words they prefer, and how they think about content. Taking users’ preferences into account early helps you build experiences that meet both business and users’ needs. A data-driven IA defines the terms, labels, and locations for content that aligns with your users’ top tasks and preferences.

But how, exactly, do you use data to create an effective IA?

Understand your users

Methods such as interviews, observing people in the field, and surveys help you to understand a broad picture of your user groups and build tools, like personas or archetypes, to bring life to your users and build empathy for them within your organization.

Bringing clarity and definition to your user groups aligns your content with what your users need to know to complete their tasks. The depth of information needed to develop a tool like a persona depends on the project, but generally, they should provide details about each user group and what makes them unique, their motivations and top tasks, and their pain points or frustrations.

The personas and other user research insights help guide a data-driven information architecture. Mostly, they remind you that you are not the user. You have to think like your users and learn how they see the world in order to create the right paths for them through your content.

Start with a card sort

Now that you know more about your users, you can start to learn how they would organize your content. Through a method of testing called a card sort, you can invite representatives from your primary user groups to organize, or sort, your content into categories and give them labels. These categories help you understand how your users think of your content.

In a card sort, people aren’t choosing what links will show up in the navigation bar of your website; they are just telling you what things they think go in the same grouping.

Validate with a tree test

The card sort results are the starting point for developing the first draft of your information architecture. You’ll need to find or create a place for all your content within the boundaries your users’ card sort groupings. In doing this, you’ll undoubtedly need to make some assumptions, so to ensure your IA is data-driven, you’ll need to test these assumptions. We do that through a method of user testing called “tree testing”.

In a tree test, you set up the IA in a hierarchical, folder-like structure so that test participants can expand or collapse categories and drill down into sub-categories. You then provide them with a set of tasks to tell you where in the structure they would expect to find a piece of content.

Participants click through your draft IA, showing you where you made good assumptions and where you were off base. If a significant number of participants weren’t able to complete a specific task – or if participants were able to complete the task but not in a reasonable amount of time – it’s a good indicator the assumptions you made after the card sort were incorrect. You can then adjust your IA based on this data you collected and test again if necessary.

Personas, card sorting, and tree testing set up a user-focused and data-driven structure for your content. Not only will this help users find your content, it also gives you a common-sense strategy for how your content fits together, eliminating internal opinion-based conflicts and decisions over where content belongs.

We’ve created user-centered digital experiences for dozens of governmental and non-profit clients. Let us know if your organization is looking for help to tackle a complex information architecture.