Hybrid search
Hybrid search runs a keyword search and a vector search on the same query and merges the two ranked lists, so results can match on exact words as well as on meaning.
Also called: hybrid retrieval
By Divya Sharma · Checked Nov 15, 2026 · How we review
Vector search is good at paraphrases and weak at exact identifiers. Keyword search with BM25 is the reverse: it finds an exact error code but misses "the app crashes on login" when the document says "authentication failure". Hybrid search runs both and combines them, which usually covers more of the queries real users type.
How the lists are merged
The two searches produce scores on different scales, so they cannot simply be added. Two common methods are used. Reciprocal rank fusion (RRF) ignores raw scores and rewards documents that rank high in either list; it needs no tuning. Weighted score fusion normalizes each list's scores and blends them with a weight, often called alpha, that sets how much each side counts.
Example
A developer docs site gets a query naming a specific HTTP status code and the word "retry". BM25 finds pages with that exact code. Vector search finds a page on backoff strategies that never mentions it. Fusion puts both near the top.
Where it fits
Hybrid search is a first stage. Teams often send its top results to a reranking model for a final order before passing them to a retrieval-augmented generation prompt. Many vector databases, such as Weaviate, support it directly. Learned sparse retrieval can stand in for BM25 on the keyword side.
Related terms
- BM25
- BM25 is a ranking formula for keyword search. It scores each document by how often the query's words appear in it, giving more weight to rare words and less credit to long documents and repeated terms.
- Vector search
- Vector search finds items by comparing their embeddings to the embedding of a query and returning the closest ones, so results match on meaning rather than on shared keywords.
- Reranking
- Reranking is a second search step that takes a short list of candidate documents and reorders them by relevance to the query, using a more accurate but slower model than the first search.
- Sparse retrieval
- Sparse retrieval represents queries and documents as vectors over a vocabulary where most entries are zero and each non-zero entry is a word weight, so search can use fast inverted indexes like keyword search.
- Retrieval-augmented generation
- Retrieval-augmented generation (RAG) is a pattern where a system first searches a collection of documents for passages relevant to a question, then gives those passages to a language model so its answer draws on them.