information-architecture

Card sorting asks 15-30 users to group content cards into categories they name themselves (open sort) or fit into your predefined categories (closed sort). Analyze results with similarity matrices and dendrograms to find clusters with over 70% agreement, then translate those groupings into your sitemap.

Tree testing validates a navigation structure by giving 20-50 users text-based tasks like 'Where would you find X?' without any visual design. Run it before development begins, targeting at least 70% success rate and directness under 1.5 times the optimal path.

Aim for 3-4 levels deep with 5-9 items per level, based on Hick's Law and Miller's 7±2 research. Deeper hierarchies cause users to get lost, while more than 12 top-level items overwhelm; supplement deep content with search and faceted filters.

A hierarchy gives each item one location in a tree, while faceted navigation assigns orthogonal attributes like category, price, and brand that users filter in any combination. Facets scale better for catalogs over 500 items where users need multiple ways to slice content.

Common causes include vague labels like 'Resources' or 'Solutions', navigation mirroring the company org chart instead of user mental models, and structures never validated with real users. Card sorting reveals the vocabulary users actually expect, and tree testing catches confusing labels before launch.

For content sets over 5,000 items or when users know exactly what they want, navigation alone is insufficient. Provide multiple access paths including search with autocomplete, faceted filters, tags, breadcrumbs, and related links to support both searchers and browsers.