A variant of card sorting where participants create and label their own grouping categories rather than sorting into predefined ones. Open sorting reveals the language and mental models users naturally apply to content, making it especially valuable in the early stages of an information architecture project.
Common contexts
- Discovering what category labels users expect for a healthcare portal navigation before any IA structure is drafted
- Running an open sort on 40 product features to find natural clusters before designing a SaaS settings menu
- Benchmarking how different user segments mentally group support articles to inform a knowledge base redesign
Use when
Run an open card sort at the beginning of an IA project — before you've committed to any category names — especially when your team's internal language doesn't match what users say in interviews and support tickets.
Avoid when
Don't use an open sort to validate an existing navigation structure; the open format will generate new structures rather than evaluate the one you've built — use tree testing for validation instead.
The most useful output of an open card sort isn't the groupings — it's the category labels participants write, because those are the exact words your navigation and search should use.
Real-world examples
- Shopify ran open card sorts with 50 merchants to design their admin navigation, discovering that users mentally grouped 'orders', 'customers', and 'shipping' together — a grouping their previous IA scattered across five menus.
- The BBC conducted open card sorts with 200 participants to restructure bbc.co.uk, finding that users created 12 consistent mental categories where the organisation's internal structure had 40+ departments.
- A healthcare system used open card sorts to understand how patients labelled their own symptoms, discovering that 'chest tightness' was used instead of 'angina' — a finding that reshaped their symptom-checker terminology.