Attribute-Based Filtering

A recommendation technique that retrieves products or content by matching structured attributes rather than by analyzing the behavior of other users. If a running shoe is tagged with "stability: high" and "arch: low," the system can find insoles, socks, or second-pair shoes that share those biomechanical properties. Attribute-based filtering is especially powerful in cold-start situations where little or no user interaction data exists — a new product, a new customer, or a niche category. It requires a catalogue with high data granularity, but it delivers relevance immediately, without needing a history of collective behavior.

Example: Zappos used over forty attributes per shoe to power recommendations that suggested specific insoles for specific running gaits rather than generic water bottles.