Semantic Search
Understand what shoppers mean, not just what they type.
Embeddings let the search engine understand 'office chair that won't kill my back' is the same query as 'ergonomic mesh task chair'. Scouty runs semantic alongside keyword search so you get both intent matching and exact-identifier precision.
Synonym lists run out before shopper vocabulary does.
Use-case and natural-language queries dominate the long tail. No synonym list, however large, will catch them all.
- Shoppers say 'for sweaty feet'; the catalog says 'moisture-wicking sock'
- Shoppers say 'for sensitive skin'; the catalog lists ingredients
- Shoppers say 'office chair that won't kill my back'; the catalog says 'ergonomic mesh task chair'
- Long-tail vocabulary is impossible to enumerate as rules
Vector retrieval that maps meaning, not surface form.
Scouty Semantic adds an embedding-based retrieval surface alongside the keyword index. It's metered by query and vector volume so cost scales predictably.
- Embeddings across products, documents, and assets
- Hybrid keyword + semantic retrieval merged automatically
- Optional reranker for noisy result sets
- Per-surface tuning for products, documents, or assets
- Transparent metering on queries and vector objects
- Layers on top of synonym rules without replacing them
What ships with semantic search.
Intent matching
Use-case queries find the right product even when no keyword overlap exists.
Hybrid retrieval
Keyword precision combined with semantic recall, merged and ranked.
Query expansion
Expand short queries with related concepts to surface long-tail products.
Reranking
Optional reranker model for noisy candidates and large catalogs.
Semantic analytics
See where semantic search recovers demand keyword search would have missed.
API + components
Same retrieval surface for storefronts, headless apps, and AI assistants.
Where it matters most.
Beauty / skincare
'Hydrating' finds products described as 'moisture repair' or 'rehydration complex'.
Outdoor / lifestyle
'For wet trails' finds GORE-TEX boots even when titles only list materials.
B2B technical
'Compatible with 480V three-phase' finds spec sheet content and the right SKUs.
Home / fashion
'Boho rug' or 'cottagecore lamp' surfaces visually-relevant products by style.
Often turned on alongside semantic search.
Get started
Ready to put one search bar across your products, documents, and assets?
Start with a free expert-led Search Audit. A Scouty specialist will review your storefront and recommend a clear next step.