Hybrid search

Hybrid search runs full-text search and vector search in one request. For the vector part, you can use managed semantic search workflows or set up vector fields yourself. Either way, you return one ranked list that combines keyword matching with similarity search.

On Elastic Cloud Serverless, use an Elasticsearch Vector Database project when embeddings and similarity search are the primary workload, even if you also run full-text queries. If search is mainly lexical or mixed with time series and analytics, use an Elasticsearch project instead. Compare Elasticsearch and Vector Database projects to choose the right project type for your use case.

The recommended way to use hybrid search in the Elastic Stack is the semantic_text workflow. Check out the hands-on tutorial for a step-by-step guide.

We recommend implementing hybrid search with the reciprocal rank fusion (RRF) algorithm. This approach merges rankings from the full-text and vector queries, giving more weight to documents that score well in either one. The final list balances exact keyword matches with similarity-based matches.