Tree Testing to Evaluate Information Architecture Categories

Summary

This video explains the differences and complementary roles of card sorting and tree testing in user research. Card sorting helps uncover users' mental models and categorize content, making it ideal for new projects. Tree testing, on the other hand, evaluates an existing menu structure by having users complete tasks to find specific information, revealing findability issues. For redesigns, starting with a tree test identifies problem areas, and card sorting helps refine solutions. The video also covers practical details such as sample sizes, quantitative metrics, and testing multiple variants, providing a clear strategy for optimizing information architecture and improving user navigation.

Takeaways

• 🃏 Card sorting helps understand users' mental models and how they group content, but it doesn’t show if users can find content in others’ groupings.
• 🌳 Tree testing is the inverse of card sorting: users navigate the menu structure to find specific information without page content or design distractions.
• 🔍 Tree testing is ideal for revising menus because it identifies findability problems in the information architecture.
• 🎯 For redesign projects, start with a tree test to uncover problematic categories before making adjustments.
• 📝 Create tasks for tree testing based on main business priorities to ensure relevance and actionable insights.
• 📊 Tree testing is quantitative, requiring a larger sample size than typical qualitative usability tests.
• 👥 At least 50 participants are recommended for a single tree test to achieve a 10% margin of error at a 90% confidence level.
• ⚖️ For A/B testing of two menu variants, you need at least 100 participants, 50 per variant.
• 🤝 Card sorting and tree testing complement each other rather than compete.
• 🚀 Suggested workflow: For new projects, start with card sorting then validate with tree testing; for redesigns, start with tree testing then use card sorting to explore solutions.
• 📈 Tree testing can also be used to test multiple IA structures or even competitor menus to evaluate menu effectiveness.

Q & A

What is the purpose of card sorting in information architecture?

Card sorting is a user research method used to understand people's mental models and to determine how content should be grouped in information architecture.

How is tree testing different from card sorting?

Tree testing is the inverse of card sorting. While card sorting asks users to sort individual content items into logical groups, tree testing presents users with an existing menu structure and tasks them to find specific information, testing the actual organization of content.

What is the main benefit of starting with tree testing when redesigning a website?

Tree testing is diagnostic and helps identify findability problems in information architecture. It allows you to identify which categories or sections in the menu structure are difficult for users to navigate, giving you insights into areas that need adjustment.

Why is tree testing considered a better starting point for most redesign projects than card sorting?

Tree testing directly tests the usability of the menu structure, helping to identify specific issues with findability. Card sorting, on the other hand, focuses on groupings and doesn't offer direct insights into whether users can actually find the content within those groupings.

How does tree testing contribute to optimizing the navigation of a website?

By testing the actual menu structure, tree testing helps reveal which categories are problematic and which work well, allowing you to refine the navigation based on real user feedback, ensuring better usability.

What kind of data does tree testing provide?

Tree testing is a quantitative research method that provides data about users' success in finding the content they need. The primary metric is the percentage of users who can successfully find specific content within the menu structure.

How many participants should be involved in a tree testing study for reliable results?

For reliable results, at least 50 participants are needed. This sample size provides a margin of error around 10% at the 90% confidence level, which is sufficient for most usability studies.

Why would you need 100 participants for an A/B test in tree testing?

An A/B test compares two variants of a menu structure. To ensure statistical validity, you need at least 50 participants for each variant, making a total of 100 participants to test both structures effectively.

How do card sorting and tree testing complement each other?

Card sorting and tree testing are complementary methods. If you're starting a new project, you can use card sorting to identify categories, then use tree testing to refine those categories. In a redesign, start with tree testing to find usability issues, then use card sorting to explore potential solutions.

What kind of tasks should you create for participants during tree testing?

During tree testing, tasks should be designed to reflect core business goals. For example, asking participants to find specific information or products that align with the website's business priorities helps determine how well the structure meets user needs.