Semantic search
Semantic search finds the seats article when the customer asks about adding another login.
Why it matters
Keyword search requires the asker to already know your vocabulary, which defeats the purpose of asking. Most failed help-centre searches are vocabulary mismatches rather than missing content.
- Search returning nothing for well-documented topics
- Support tickets about articles that exist
- Readers giving up after one search that returned noise
How it differs from keyword search
Keyword search scores documents by term overlap, so a question sharing no words with the answer scores zero. Semantic search compares embedded meaning, so paraphrases, synonyms, and differently-framed questions still match.
Where keyword search is still better
Exact identifiers — error codes, SKUs, part numbers, function names. Semantic similarity is unhelpful for a string that must match exactly, which is why serious systems combine both rather than replacing one with the other.
Semantic for questions
Natural language, paraphrase, differing vocabulary.
Keyword for identifiers
Error codes and part numbers need exact matching.
Common questions
- Is semantic search always better?
- No. For exact identifiers it is worse, because approximate meaning is precisely the wrong tool for a string that must match exactly. Hybrid approaches exist because both failure modes are real.
- Does it need my content in a special format?
- No, but structure helps. Content that answers a question directly under a clear heading is retrieved more reliably than the same fact buried mid-paragraph.
Related terms
From the blog
All posts- EvaluatingParaphrase testing: the five-minute retrieval checkA five-minute paraphrase test separates keyword matching from real retrieval. Same fact, three phrasings, one expected answer — including on Matter Chat.Read
- MeasuringDeflection rate is the most overstated number in support automationCounting bot conversations as deflected tickets overstates the result. The honest version is a before-and-after on ticket volume, read next to satisfaction.Read
- EvaluatingHow to test an AI support tool before you trust itEvery AI support tool demos well, because demos ask questions the content covers. Four questions that separate them, and what a good answer looks like.Read
Answer honestly. Capture the rest.
Point Matter Chat at your site and see what it can — and can't — answer. It's honest about both.
No credit card. 2 minute setup.
Every answer cites the source it came from. When there isn't one, it says so — and hands the visitor to you.
Installs on the tools you already run.