Card Sorting and AI: How Artificial Intelligence Is Transforming Information Architecture

By Philip Burgess - UX Research Leader

By Philip Burgess – UX Research Leader📧 phil@philipburgess.net | 🌐 www.philipburgess.net

Understanding the Foundation: What Is Card Sorting?

Card sorting has been a cornerstone of UX research and information architecture (IA) for decades. It’s a deceptively simple method—participants group related concepts or “cards” into categories that make sense to them.The outcome?A clearer understanding of how users mentally organize information, helping teams design intuitive navigation structures, content hierarchies, and labeling systems.

Traditional card sorting—whether open, closed, or hybrid—provides valuable insights but can be time-consuming and subjective to analyze, especially at scale.That’s where AI is beginning to transform the landscape.

🤖 The New Era: AI Meets Card Sorting

Artificial Intelligence is no longer just a back-end tool for automation—it’s becoming an active research collaborator.In the realm of card sorting, AI now supports UX researchers in three key ways:

1️⃣ Automating Data Clustering

AI can instantly identify relationships between cards and categories that would take hours to detect manually.Instead of relying solely on frequency analysis or dendrograms, AI models use natural language processing (NLP) to understand context and meaning—grouping related items even if participants used slightly different labels.

Example Prompt (AI-Assisted Clustering):

“Analyze this open card sort dataset and identify 5–7 logical content groupings with rationale.”

This allows researchers to focus more on interpretation and validation, not manual cleanup.

2️⃣ Predicting Optimal Information Architectures

By training on large datasets of user behavior and navigation patterns, AI can predict likely information architectures that maximize findability.It can suggest:

• Category labels aligned with cognitive load principles.

• Ideal menu depths for usability and accessibility.

• Potential gaps or overlaps that may confuse users.

Think of it as a “smart IA co-pilot” that helps you validate early hypotheses before testing with users.

3️⃣ Cross-Referencing with Behavioral Data

AI can connect card sorting results with analytics, heatmaps, or clickstream data to triangulate insights.For instance:

• Do the card sort groupings align with actual navigation behavior?

• Are users labeling content differently than how they interact with it?

This fusion of qualitative structure and quantitative behavior gives UX teams a more holistic view of how users perceive and use information.

⚙️ How Researchers Are Using AI in Practice

Here’s how forward-thinking UX researchers are integrating AI into their card sorting workflows:

Step	Traditional Approach	AI-Enhanced Approach
Setup	Define cards & categories manually	Generate card ideas and taxonomy suggestions via AI prompts
Analysis	Cluster categories manually	Use NLP and clustering models for pattern detection
Interpretation	Subjective synthesis	AI-assisted summaries with rationale and confidence levels
Validation	Follow-up tree test	AI predicts success rates and potential confusion points

Tools like OptimalSort, Maze, and Miro are starting to explore these capabilities, while researchers are experimenting with ChatGPT, Claude, and Gemini for early synthesis and data interpretation.

💡 Best Practices for Using AI in Card Sorting

1. Use AI as an Assistant, Not a Decision MakerAlways validate AI groupings with real participant data and domain context.

2. Check for Bias in ClusteringAI models may overgeneralize or prioritize frequency over meaning. Ensure diverse participant samples to offset bias.

3. Refine Through IterationUse multiple AI prompts to compare alternative taxonomies before committing to one structure.

4. Document the Process TransparentlyNote where AI contributed (e.g., clustering, labeling) so stakeholders understand the provenance of insights.

🌐 Why This Matters for the Future of UX Research

Card sorting has always been about understanding mental models.AI extends that goal—helping us see connections faster, spot inconsistencies earlier, and scale analysis without diluting quality.

But it doesn’t replace human empathy, intuition, or domain expertise.Instead, it amplifies them—freeing researchers to focus on strategy, storytelling, and alignment with business outcomes.

🚀 Final Thought: AI as the Next Research Collaborator

As AI continues to evolve, the UX researcher’s role will shift from analyzer to orchestrator—guiding human insights and machine intelligence toward a shared goal:clarity for users, efficiency for teams, and strategy for organizations.

The next time you run a card sort, consider letting AI sit beside you—not as a shortcut, but as a collaborator in uncovering meaning.

By Philip Burgess – UX Research Leader📧 phil@philipburgess.net🌐 www.philipburgess.netFollow for more insights on AI + UX Research Integration