Hybrid search runs both vector similarity search and keyword search (usually BM25) and combines the results — typically with a reciprocal rank fusion (RRF) merge. It consistently outperforms either approach alone because they're complementary: vector search handles paraphrase and semantic similarity; keyword search handles exact terms, product codes, names, and acronyms.
The practical setup: most vector databases (Weaviate, Qdrant, Elasticsearch, OpenSearch) have BM25 built in alongside vector search. Enable both, use RRF to merge results, and run a reranker on top. This three-layer pattern (hybrid retrieval → rerank → generate) is the current production baseline for quality RAG.
If you're only doing vector search, adding BM25 is usually the single highest-ROI RAG improvement you can make in an afternoon.
Bring this to your business
Knowing the term is one thing. Shipping it is another.
We do two-week AI Sprints — one term, one workflow, into production by Day 10.