Key Takeaways
- Assume the question moves the answer, not the buyer. An assistant reacts to the place written in the question far more than to where the person is sitting. Two buyers in different states who type the same words usually see much the same answer.
- Expect coarse location at best, and often none. When a tool does infer where somebody is, it is working from a rough signal like a city or region. There is no map ranking behind an AI answer, so nothing works the way a local search result does.
- Name your place in your own words, on your own pages. The assistant cannot infer that you serve Denver from a phone area code. Write the city, the neighborhoods, and the radius in a sentence somebody could quote.
- Test the phrasing your buyers actually use. Ask the question without a place, then with your city, then with your neighborhood. The rung where local businesses start appearing is the rung worth competing on.
- Do not read one answer as a measurement. The same question asked twice can come back different for reasons that have nothing to do with location. Ask it three times before you conclude anything about where you stand.
A coach in Denver asks ChatGPT who the best strength coaches are and gets three national names. A client two states away asks the same thing and gets the same three. Neither answer mentions anybody either of them could actually book.
The natural conclusion is that the assistant does not know where anyone is. That is close to right, and it is the more useful half of the story. The other half is that it mostly does not need to, because what moves the answer is the place written into the question.
This post is about which of those two things is doing the work, and what to publish once you know.
Do AI answers change depending on where the buyer is?
Sometimes, but far less than people expect. The place written into the question moves the answer much more than the location of the person asking, and several assistants have no reliable idea where you are at all.
That is worth sitting with, because it inverts the usual local marketing instinct. In map results, the buyer's position is the main input. In an AI answer, the buyer's words are the main input, and their position is a weak signal that may not be used.
Google is explicit that its AI features are built on the same systems that rank ordinary results, which means place can matter there in the way it always has. Chat assistants are a different shape: they answer from what they have read plus, sometimes, a live look at the web. Perplexity's help center describes a product that searches and then cites, which is why its answers move when the phrasing moves.
What actually moves the answer
Three things, in descending order of importance. First, the words in the question, and particularly whether a place is named in them. Second, whether the tool looked anything up rather than answering from memory. Third, a coarse location signal, where the tool uses one at all.
The first is doing most of the work. Add a city to a question and the whole character of the answer changes, because a place gives the assistant something to match against pages that mention that place. Leave the city out and there is nothing to match, so it falls back on whatever it knows generally, which is national and well-known.
The second matters because an assistant answering from memory alone cannot reflect anything recent or small. The third is real but blunt. A rough sense of city or region is not a ranked list of nearby businesses, and it is not stable across sessions.
A good strength coach
A good strength coach for a beginner
A good strength coach in Denver
A good strength coach near Wash Park in Denver
The answer starts naming businesses in a place at the point the buyer puts a place in the question. Nothing about the buyer's own position moves it that far.
Why this is nothing like a map result
A map result is a ranked list of places, sorted partly by how far they are from you. There is no equivalent behind an AI answer. The assistant is assembling a sentence from documents, and a document either describes a business as serving somewhere or it does not.
So proximity does not help you. Being the closest strength coach to the person asking is worth nothing if no page says which city you work in. A competitor forty minutes further away whose site names three neighborhoods is the one that can be named.
This is also why the usual local signals carry less weight here than they do in search. Google's local business structured data is genuinely useful and worth having, but it is machine-readable metadata, not a sentence. What an assistant quotes is prose. The full picture of where that prose gets read is in our piece on where ChatGPT gets information about local businesses.
| What the buyer types | Place in the question | What the answer can name | Worth competing for? |
|---|---|---|---|
| A good strength coach | None | National names and categories of coach | No, it never resolves to a local business |
| A good strength coach for a beginner | None | Approaches, program types, occasionally an author | Only as proof you can explain your method |
| A good strength coach in Denver | A named city | Gyms, directories, sometimes individual coaches | Yes, this is the main one |
| A good strength coach near Wash Park in Denver | A named neighborhood | Named local businesses, where a page describes them that way | Yes, and it is far less contested |
What it means if you serve one city, and if you do not
If you serve one city, your job is to make the place unmissable in prose. Not in a footer, not in a title tag alone, but in a sentence a model could lift: we coach strength athletes in Denver and work with clients across the Front Range. Then the neighborhoods, by name, where you genuinely work.
Say the radius too. An assistant asked about a suburb has no way to know whether a business in the city center covers it unless somebody wrote that down. This is the cheapest fix on the list and the one most often left to inference.
If you work remotely, the same logic runs in reverse and is easy to get wrong. Do not stuff cities you do not serve. Say plainly that you work online with clients anywhere in the United States, and then compete on the specifics that are actually yours: the kind of client, the problem, the method. The same pattern for solo practitioners is covered in how local consultants show up in AI recommendations.
How to test it in ten minutes
Open two assistants. Ask your buyer's question with no place in it, then with your city, then with your neighborhood or suburb. Write down the names that come back at each rung, and which rung is the first to name any local business at all.
Ask each version three times. Answers vary between runs for reasons that have nothing to do with you, and a single response is an anecdote rather than a reading. We wrote about that failure mode in why the same question asked twice gives different answers.
Then look at what the named businesses have in common. In almost every local test we run, the answer is that some page other than their own describes them as being in that place: a directory entry, a roundup, a local news piece, a gym's coach roster. Being local is not a signal. Being described as local is.
One honest limit. None of this makes an assistant into a map. If a buyer wants the nearest option right now, they will open a maps app, and they should. What is winnable here is the buyer who is choosing rather than navigating, and who asks in a sentence rather than a search box.
If you would rather see the whole picture at once, the free check asks ChatGPT 25 real buyer questions from your niche at $0 and emails a plain-English readout. 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
Does AI give different answers depending on where the buyer is?
Sometimes, but far less than people expect. The place written into the question moves the answer much more than the location of the person asking, and several assistants have no reliable idea where you are at all.
Will ChatGPT recommend a local business to someone in my city?
Only if the question names a place and some page describes a business as serving that place. Without a named city or neighborhood in the question, the answer usually stays general and names categories rather than businesses.
Does using a VPN change what AI recommends?
Rarely in a way that matters. Where a tool does use location, it is a coarse signal, and the words in the question override it. Testing through a VPN mostly tells you about the tool rather than about your visibility.
Why did a client in another state see a different answer than I did?
Usually because the wording differed, or because the same question simply varies between runs. Model updates, whether the tool searched the web that time, and small phrasing changes all move an answer more than distance does.
Should I make a separate page for every city I serve?
Only where you genuinely serve them and can say something specific. Thin pages that swap one city name for another read as filler to both search engines and assistants, and they give a reader nothing worth quoting.
How do I find out which local questions I am missing from?
Ask them yourself, in the exact words a buyer would use, with your city in the question. Do it three times in two or three assistants and write down who gets named instead of you.
What does the AnswerHalo free check include?
It asks ChatGPT 25 real buyer questions from your niche and emails you a plain-English report: your score, who got named in your place, and the one fix to start with. It costs $0 and it runs as soon as you ask for it.
Does the free check ask location questions?
Yes, when your niche is local. The 25 questions are built for your niche before the run, so a business that serves one city gets asked the way its buyers ask, with the city in the question.
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 →