Skip to main content

What is search in Layers?

Search is the system that connects your customers to your products. When a shopper types a query, Layers goes beyond simple keyword matching — it combines semantic understanding, keyword relevance, engagement data, and contextual awareness to rank and return the most relevant products. Layers supports multiple search modes: text search for keyword and natural-language queries, image search for visual discovery, similar product search for finding related items, and autocomplete for real-time query suggestions.

How search ranking works

Every text search result is scored using five signal groups. Each group contributes a percentage of the total score, and all group weights always sum to 100%:
Signal groupWhat it measures
SemanticHow closely product content matches the meaning of the query
KeywordHow well exact terms in the query match product fields
EngagementHow customers interact with products (views, sales, cart sessions)
FreshnessHow recently the product was published
InventoryWhether the product is in stock
You have direct control over how much each signal contributes to the final score through the Ranking Relevancy settings in the dashboard.

Query understanding

Layers interprets what your customers are searching for, not just the words they type. The query understanding pipeline includes:
  • Query expansion — Enriches searches with synonyms and related terms. A search for “sneakers” also considers “athletic shoes” and “trainers.”
  • Typo tolerance — Corrects spelling mistakes automatically so customers still find what they need.
  • Intent detection — Recognizes signals like price sensitivity, sale interest, and feature preferences within the query.
  • Contextual awareness — Uses cart contents, purchase history, and geographic location to personalize how queries are interpreted.

Ranking rules

Beyond global signal weights, you can create ranking rules that apply conditional adjustments to search results:
  • Boost — Promote products matching specific attributes higher in results
  • Bury — Demote products matching specific attributes lower in results
  • Pin — Lock specific products to exact positions for particular queries
Rules can be scoped globally or triggered by specific query conditions, including exact match, substring match, and semantic similarity.

Search rules

Search rules modify how search requests are processed before results are returned. They can replace or modify the search query, add additional filters, or override sort orders — all based on targeting conditions like the query itself, which collection the shopper is browsing, or shopper context such as location, channel, and customer data.

Next steps