Hybrid search (retrieval)

Hybrid search combines dense vector retrieval with sparse keyword (BM25) retrieval, then fuses the two ranked lists — the production retrieval default for RAG in 2026.

Pure vector search misses exact-match queries (product SKUs, error codes, named entities). Pure keyword search misses semantic intent. Hybrid combines them — typically retrieve top 50-100 from each, fuse with reciprocal rank fusion (RRF) or weighted score, then re-rank the top 20-30 with a cross-encoder. Empirically beats both single approaches on production corpora. By 2026 hybrid retrieval is the default in serious RAG stacks (Cohere Rerank pipeline, Voyage AI Rerank, Pinecone hybrid, Weaviate hybrid, Vespa).

When to use hybrid search (retrieval)

• Any production RAG over diverse corpora.
• Mixed query types (some keyword-heavy, some semantic).

Common mistakes

• Fusing scores directly across normalisation regimes — use RRF or rank-based fusion.
• Skipping the re-ranker — first-stage fusion is noisy.

FAQ

Related terms

• Retrieval-augmented generation (RAG) — Retrieval-augmented generation (RAG) injects relevant documents into the prompt at query time so the model answers from your data instead of its training memory.
• Semantic search — Semantic search finds documents by meaning rather than keyword match, using embedding similarity in a vector space.
• Embeddings — Embeddings are dense numeric vectors that represent the meaning of text, images, or other data, allowing similarity to be measured as vector distance.
• Vector database — A vector database stores embeddings and performs approximate nearest-neighbor search at scale, the persistence layer behind RAG and semantic search.

Last updated: 2026-06-01. Raw markdown: https://promtable.com/glossary/hybrid-search.md.