The Relevance tab is where you define exactly how your search engine interprets and matches queries against your catalog—letting you strike the perfect balance between AI-powered understanding and classic keyword matching. Every control here shapes the core “relevance signal” that drives which items rise to the top of search and category listings. Toggle on Enable AI Search to sse the advanced multi‐model search powered by Gen AI to unlock deep semantic matching across both text and images.

​
Weightage for Multi-Model Searching

Fine-tune the balance between visual and textual signals in your Gen AI results. Adjust the two sliders so their sum is always 100:
ControlWhat It Does
Image WeightagePrioritizes visual similarity i.e. color, pattern, shape thus boosting items whose images most closely match the query.
Text WeightagePrioritizes keyword and language-based relevance acrossproduct titles, descriptions, synonyms thus boosting text match.
Total Weightage (100)Ensures a fixed combined influence so that image + text importance always equals 100%.

​
Out-of-Stock Products

Determine how unavailable items should appear to shoppers:
  • Exclude in search results and category pages
  • Include in search results and category pages
  • Bury in search results and category pages
This lets you manage customer expectations—either hiding sold-out items completely, keeping them visible, or pushing them to the bottom.

​
Semantic Understanding

  • Natural Language Processing (NLP) Toggle to enable deeper understanding of conversational queries (“red shoes for running.”).
    Only for AI Search; Gen AI Search manages NLP automatically.
  • Semantic Understanding for Price Toggle to interpret price-based queries and ranges (e.g., “gifts between $500 and $1,000”).

​
Enrichments

Enable or disable specific text‐processing features to refine your results:
EnrichmentDescription
SynonymsRecognize alternate terms (e.g., “sneakers” ↔ “shoes”) to return all relevant items.
PhrasesTreat multi-word queries as a single unit (e.g., “iPhone case” vs. separate “iPhone” + “case”).
StopwordsAutomatically strip inconsequential words (a, and, the, for) for cleaner query parsing.
Spell-checkApply industry-specific or custom dictionary entries to correct or suggest spellings.
StemmingReduce words to their root form (e.g., “running” → “run”) so variations all match together.
The above settings are only for AI Search; Gen AI Search handles these behind the scenes.
By carefully tuning each of these settings, you shape the core relevance signal that determines which products your customers see first. Adjust weightages, toggle enrichments, and decide how to handle edge cases like out-of-stock items—then watch your search results transform to match your exact priorities.