Retrieval Quality
How accurately and completely a RAG or search system pulls the right supporting content before an AI generates its answer, the step that determines whether a citation is even possible.
What it means
Retrieval happens before generation: an AI system searches its available sources, live web, indexed catalog, connected database, for content relevant to the query, then generates an answer grounded in what it retrieved. If retrieval quality is poor, missing the right product, pulling outdated data, retrieving an irrelevant page, the generated answer will be wrong or incomplete regardless of how good the underlying model is. Retrieval quality issues are often invisible to a merchant since the failure happens upstream of anything visible on their own site.
Why it matters for Shopify
Shopify merchants improve their odds of being retrieved accurately by keeping product data current, structured, and free of contradictions, poor retrieval quality often stems from stale or inconsistent source data, not model limitations.