Where things stood
Manual catalog mapping between buyers and vendors was capping growth. Flower products have no global identifiers and enormous semantic variation — one large buyer alone had over 20,000 unmapped products against vendor catalogs exceeding 46,000 records, and new partners were abandoning onboarding under the sheer mapping volume.
What changed
Claro built semantic product matching combined with attribute penalties and flower-specific domain logic — color families, stem lengths, weights, grades and acceptable ranges all folded into the matching score — generating ranked vendor candidates with confidence tiers and a confirm-or-deny review workflow.
What it looks like today
Opportunity accuracy rose from 59% to 83%.
True-positive rate improved from 25% to 77% with vendor-product filtering, reaching 90% accuracy after filtering.
High-confidence results covered 53% of products at 83% accuracy, and the correct vendor appeared in the top ten candidates for 90% of products.
Unmapped inventory was commercially invisible. Catalog matching therefore directly affected order opportunities, onboarding speed, and transaction revenue.