What is hybrid search?
Quick answer
Hybrid search combines two kinds of search in one: keyword search, which matches exact words such as a policy number or product code, and vector search, which matches by meaning. The results are blended into one ranked list, so you catch both the exact match and the passage that says the same thing in other words.
Last updated
Take the free module →Module 14 of the free AI for Customer Support course, about 13 minutes. No account needed.
Updated · By Robert Breen
Why it matters for a small business
Vector search alone can miss exact terms. Ask about "form W-9" or "SKU 4471" and a meaning-based search may return something about tax forms in general. Keyword search alone misses paraphrases. Business documents are full of both codes and plain sentences, so blending the two often gives better answers.
It is not always needed. For a short handbook in plain language, vector search with good chunking is usually enough. Hybrid search earns its place when people search by codes, names, part numbers or legal terms that must match exactly.
Where you'll use it: in the vector database. Pinecone supports hybrid search by storing sparse (keyword) and dense (meaning) values, and other databases offer their own versions. Check whether your automation tool's vector store node exposes the option before you plan on it.
In a real lesson: Answer Staff Questions from Your Own Docs using n8n + Pinecone
In the free n8n + Pinecone staff Q&A lesson, Cedar Ridge Partners' handbook is searched by meaning only: the questions and chunks are turned into numbers with the same OpenAI embeddings model, text-embedding-3-small, used when the handbook was loaded.
That suits plain questions like "how many vacation days do I get?". If staff often asked about expense codes or form numbers, a hybrid index that also matches keywords would help find the exact policy line instead of a related one.
Try this lesson free or read the step-by-step guide.
Common confusions
Hybrid search vs semantic search
Semantic search is the meaning half. Hybrid search adds keyword matching on top.
Hybrid search vs reranking
Hybrid search finds candidates two ways. Reranking reorders the candidates afterward.
Tips
- Test with questions that use exact codes and with ones in plain words.
- Start with meaning-only search and add keywords if exact terms get missed.
- Keep IDs and codes in the chunk text so keyword search can find them.
Related terms
More AI agents terms
Where you use it: free lessons
- Answer Staff Questions from Your Own Docs using n8n + Pinecone (n8n, 13 min)
- Make Your Handbook Searchable for an AI Helpdesk using Pinecone + n8n (n8n, 14 min)
Frequently asked questions
- Is hybrid search slower?
- Slightly, since it does two kinds of matching, but for office-sized document sets the difference is rarely noticeable.
- Does Pinecone support hybrid search?
- Yes. Pinecone's docs describe hybrid search using sparse and dense values in the same index.
- What does sparse and dense mean?
- Dense values are the embedding numbers that capture meaning. Sparse values record which exact words appear, which is how keyword matching works inside a vector index.