Key Takeaways
- Write for the meaning of the question, not its exact words. Semantic search matches a page to what a buyer means, so a page can be found for a question that shares none of its words.
- Name the problem the way buyers feel it. A buyer types "I keep putting things off", not your program name. Say the problem in plain words on the page and the match has something to hold.
- Stop repeating a keyword to be found. Saying the same phrase ten times does not move a meaning match. One clear passage that answers the question does.
- State who you serve and where, in words. Meaning still needs facts. The buyer, the problem, the place and the price have to appear on the page for any search to match them.
- Test with questions, not keywords. Ask an assistant the full question a buyer would ask. That is the kind of query semantic search was built to answer.
A buyer rarely types the words on your website. They type what is bothering them: "I keep putting off the work that would grow my business", or "who can help me stop procrastinating on sales". Your page says "business coaching for owners". None of the words match.
Search used to need the words to match. Now it often does not, because Google and the AI assistants match what a question means against what a page means. That is semantic search, and it decides whether your page is even in the running when a buyer asks.
Below is what semantic search is, how it turns words into meaning, how it differs from keyword search, why AI assistants rely on it, and how to write a page it can match.
What is semantic search?
Semantic search is search that matches the meaning of a question to the meaning of a page, rather than matching the exact words. A page about "putting off important work" can come back for a question about "procrastination", because the two mean nearly the same thing.
The word semantic just means "to do with meaning". The term matters here only because it explains a change in what gets found: a page is no longer judged mainly on whether it contains the buyer's words, but on whether it answers what the buyer meant.
How semantic search turns words into meaning
Semantic search turns each question and each passage into a list of numbers that stands for its meaning, then finds the passages whose numbers sit closest to the question's. Close numbers mean close meaning, whatever words were used.
That list of numbers is called an embedding. OpenAI's embeddings guide describes it plainly: an embedding is a list of numbers, the distance between two of them measures how related the texts are, and search is one of the main uses. A small distance suggests the texts are closely related; a large one suggests they are not.
You never see these numbers, and you do not need to. What matters is the consequence: two sentences with no words in common can sit side by side, and two sentences that share a keyword can sit far apart if they mean different things.
Google has worked this way for years. Its 2019 announcement, understanding searches better than ever before, said its BERT system would help Search better understand one in 10 searches in the US in English, especially longer, conversational ones where small words such as "to" change the meaning.
Keyword search vs semantic search
Keyword search asks "does this page contain these words?" Semantic search asks "does this page mean what the buyer meant?" The two answers often disagree, as this illustrative example of one buyer question and five sentences from coaching pages shows.
The buyer asks"Who can help me stop procrastinating on growing my small business?"
- Business coaching for owners who want more focus. Words: yes Meaning: no
- I help founders who keep putting off the work that grows the company. Words: no Meaning: yes
- Stop procrastinating with our free productivity checklist. Words: yes Meaning: no
- Weekly accountability calls for solo business owners in Denver. Words: no Meaning: yes
- Coaching for small business owners who procrastinate on sales and planning. Words: yes Meaning: yes
The two columns agree on one sentence out of five. Keyword search picks the first, third and fifth; semantic search picks the second, fourth and fifth.
The example is illustrative, and the marks are our judgment, not the output of a real system. The two readings agree on only one sentence of the five: keyword search picks the first, third and fifth, and semantic search picks the second, fourth and fifth. The checklist sentence shares the buyer's key word and answers a different need, while the accountability sentence shares almost no words and answers exactly what was asked.
| Question | Keyword search | Semantic search |
|---|---|---|
| What does it match? | The words in the query | The meaning of the query |
| Does a synonym count? | Only if the page uses it | Yes, close meanings sit together |
| A long, conversational question | Often matched on a few words, losing the point | Read as a whole, including small words like "to" |
| A page that repeats one phrase | Can look more relevant | Gains little; repetition adds no new meaning |
| A page that describes the problem in plain words | Missed if the buyer used other words | Matched if the meaning is close |
| A word with two meanings | Matches both | Uses the rest of the sentence to pick one |
In practice the two are blended. Google's guide to its ranking systems lists exact words as one signal among many, alongside neural matching, which it describes as an AI system that matches representations of concepts in queries and pages. Where search engine work and AI visibility work meet is covered in AEO vs SEO: what overlaps.
Why AI assistants match meaning, not keywords
AI assistants match meaning because buyers ask them full questions in their own words, and keyword matching does badly on full questions. When an assistant searches the web before answering, it looks for passages whose meaning fits the question, then writes its answer from them.
That step of pulling passages in is called retrieval, and it is covered in what retrieval is in AI answers. Semantic search is the matching inside it: it decides which of the passages on the web count as relevant to what the buyer meant.
The damaging admission: no assistant publishes exactly how it ranks passages, and meaning is not the only thing that counts. Whether your site can be crawled, whether other sites mention you, and how recent the page is all play a part. Semantic search gets a good passage considered; it does not guarantee the passage gets used.
How to write so semantic search finds you
To be found by semantic search, write passages that say plainly who you help, with what problem, where and how, in the words buyers use about themselves. Meaning cannot be matched if it is not on the page.
- Describe the problem as the buyer feels it. "Owners who keep putting off sales calls" carries more meaning than "productivity solutions".
- Give each buyer question its own section. One heading phrased as the question, with the answer in its first sentence, makes a passage that matches one meaning well.
- State the facts in text. Your buyer, your city, your format and your price have to be in words, not only in a graphic or a booking widget.
- Drop the repetition. Say a phrase once where it belongs. A second, clearer sentence adds meaning; a tenth repeat of the keyword does not.
- Pick the right category words. If you call yourself a "growth partner", nothing tells a search what you do. How to find the right words is covered in how to describe what you do so AI categorizes you right.
Then test the way buyers search: with whole questions. To see what ChatGPT says when buyers in your niche ask, and whether you appear, start with the free check. It asks ChatGPT 25 real buyer questions from your niche at $0 and emails a plain-English readout: your score, who got named in your place, and the one fix to start with. It runs as soon as you ask for it, and the report lands within the hour between 7am and 5pm Central, or by 10am the next morning outside those hours.
Questions we hear the most
What is semantic search?
Semantic search is search that matches the meaning of a question to the meaning of a page, instead of matching the exact words. It can return a page about "putting off important work" for a query about "procrastination", because the two mean nearly the same thing.
How is semantic search different from keyword search?
Keyword search looks for pages that contain the words you typed. Semantic search looks for pages that mean what you meant, so it handles synonyms, plain descriptions of a problem and long conversational questions that keyword search would miss.
Do ChatGPT and Google use semantic search?
Both match meaning as well as words. Google says it uses AI systems such as BERT and neural matching to understand concepts in queries and pages, and OpenAI sells embeddings, the tool behind meaning-based search, to developers who build search.
Do keywords still matter if search is semantic?
Yes, but differently. The words still have to describe your service, your buyer and your location, or there is nothing to match. What no longer helps is repeating one phrase many times; meaning comes from clear sentences, not from counts.
What is an embedding, in plain English?
An embedding is a list of numbers that stands for the meaning of a piece of text. Texts that mean similar things get lists that sit close together, so a search system can find related passages by measuring the distance between them.
How do I write a page that semantic search can match?
Write one section per buyer question, answer it in the first sentence, and describe the problem in the words buyers use about themselves. Name who you help, where and at what price in plain text, not only inside images.
How do I find out whether AI matches my business to buyer questions?
Ask the questions your buyers ask and see whether you come back. The AnswerHalo free check asks ChatGPT 25 real buyer questions from your niche at $0 and emails your score, who got named in your place, and the one fix to start with.
What does the $500 audit show that the free check does not?
The full audit runs 125 checks across ChatGPT, Claude, Gemini, Perplexity and Google AI Overviews and lands by this time tomorrow, so you see which phrasings of a question name you and which do not. Ongoing watching is $399 a month, or $300 a month added with the audit.
Your buyers are already asking. Find out what AI tells them.
The free check asks ChatGPT 25 real buyer questions about your niche and sends you the report within the hour, at $0. If it shows you are already getting named everywhere, we will say so in plain words.
Get my free check →