When should you combine keyword and semantic search?
Answer out loud or in your head first, then compare.
Answer
Semantic search helps with paraphrases, while text matching handles exact codes, names and numbers well. Combining them, i.e. hybrid search, can use the strengths of both. Then you can assess the selected passages more carefully.
But I don't blindly add two raw numerical scores, because they may have different scales. You can merge the positions in the lists, e.g. with RRF, which rewards candidates ranked well. Our test questions decide whether the combination helps. If a document was read incorrectly or isn't there, a more complex ranking won't fix it.
Why it matters: Explains the name and purpose before the details of the ranking algorithm.
Possible follow-ups: Why do an invoice number and a problem description need different strengths from a search engine?
How did it go? Be honest — “Not yet” puts the question on your review list.
Start with
- G-P01 · What is RAG? — RAG and retrieval
- G-P02 · What are embeddings? — RAG and retrieval
Further reading
- Azure AI Search: Hybrid search overview(opens in a new tab)Docs; accessed 22.10.2026. Vectors plus full text, with rankings merged by RRF.