Facets of Faceted Search
Source: https://medium.com/@dtunkelang/facets-of-faceted-search-38c3e1043592 Author: Daniel Tunkelang
Summary
Foundational overview of Faceted Search — what it is, how it works, and the design decisions that make or break a faceted navigation experience.
What Faceted Search Is
Faceted search lets users progressively narrow a result set by selecting values from multiple independent attribute dimensions (facets). Each facet selection filters results, and facet counts update dynamically to reflect the remaining set.
Constraints vs. Preferences
Key distinction (explored further in Tunkelang’s “Facets: Constraints or Preferences?“):
- Constraints: hard filters — “must have free shipping”
- Preferences: soft signals — “prefer red, but would accept blue”
Most systems implement only hard constraints. Preference-based faceting (using soft ranking signals) better models actual user behavior.
Dynamic vs. Static Facets
- Static: always show the same facets
- Dynamic: infer relevant facets from query context (e.g., “running shoes” → show surface/support, not color first)
Query-aware dynamic facets significantly improve the experience for non-trivial queries.
The “Which Facets?” Problem
Not all attributes make useful facets. Good facets:
- Have meaningful variance across the result set
- Reflect how users actually think about the category
- Have manageable cardinality (not 10,000 unique values)