Case Study: Canals Achieves 96% Top-3 Product-Match Accuracy with Elasticsearch by Elastic

Canals boosts top-3 product-match accuracy to 96% with Elastic

Canals, a startup building AI for wholesale distributors, faced the challenge of automating product matching, a process historically reliant on human tribal knowledge. Their initial keyword-only search solution built on Elasticsearch failed to achieve the necessary accuracy for matching messy customer requests to the correct items in sprawling distributor catalogs.

By implementing a hybrid search solution on Elasticsearch, Canals combined keyword, dense vector, and sparse vector retrieval in a single platform. This approach, powered by Elastic Cloud, lifted their top-3 product match accuracy from 63% to 96%. The solution processes 400,000 queries daily across 90 million products, and now automatically selects the correct product with no user action required in roughly 90% of cases, leading to significant efficiency gains for their customers.

Canals