Semantic search

Semantic search finds results by meaning rather than exact words, so "cheap CRM" can match a page about "affordable customer management software".

AI searchUpdated 1 min read

Also called vector search · meaning-based search

What it means

Keyword search looks for the words you typed. Semantic search looks for what you meant. It turns both the question and the pages into embeddings, lists of numbers that capture meaning, and finds the closest matches.

Most modern search engines and AI tools combine both: keywords for precision, meaning for everything else.

When an AI tool picks passages from pages to answer a question, it often uses semantic matching. The passage that best matches the meaning of the question wins, even if the wording is different.

What it means for writing

  • Stuffing keywords doesn't help. Clear meaning does.
  • Say what you are plainly: "invoicing software for freelancers", not "the future of getting paid".
  • Cover the real questions buyers ask, in the language they use.

Example

A buyer asks "what's an affordable way to send invoices as a freelancer?". A page titled "Invoicing app for self-employed people, from $5 a month" never uses the word "affordable", but semantic search still matches it.

Frequently asked questions

Is keyword research still useful?

Yes, to understand what buyers ask about. Just write naturally rather than repeating exact phrases.

Is semantic search the same as vector search?

Vector search is the technique most semantic search uses.

Does Google use semantic search?

Yes, for years. AI search tools use it even more.

Go deeper

All posts →

Make AI search your next revenue channel

Start free trial