QBiz Leads AI

Good News for Local Businesses: AI Search Almost Never Drops Your Town

Here is a worry we hear a lot from local owners. People used to type "emergency plumber in Leeds" or "family solicitor near me" straight into Google, so the town was right there in the search and you knew it counted. Now those same people ask ChatGPT, Google's AI answers or Perplexity, and the tool rewrites the question behind the scenes before it goes looking for anything. The fear is obvious: if the machine rewrites "best electrician in Bristol" into something shorter and cleaner, does "in Bristol" quietly fall off, taking your one real advantage with it?

It is a fair thing to worry about. It is also, on the best evidence available, not what happens.

When AI engines rewrite a question, they throw away a surprising amount: extra adjectives, price limits, long brand lists. But the one thing they hold onto, almost without exception, is location. For a business whose whole edge is being the right answer in one place, that is the most reassuring finding in the entire field.

This guide explains what actually happens when AI rewrites your customer's question, why your town survives the cut, and how to make geography work as hard for you as it possibly can.

What "the AI rewrite" actually is

Once you see how the rewrite works, the fear mostly evaporates. When someone asks an AI tool a question, the tool rarely searches for their exact words. Instead it rewrites and expands the question into one or more cleaner search strings, runs those against the web, reads what comes back, and then writes a single answer. The customer never sees this middle step. They type a sentence and get a reply, and everything in between is hidden.

Google is open about doing this. It calls the method "query fan-out", and describes it as "breaking down your question into subtopics and issuing a multitude of queries simultaneously on your behalf" (Google, AI in Search, May 2025)[2]. Ask "who is a good family solicitor in Cardiff for a divorce?" and the engine quietly runs several smaller searches at once: family law firms in Cardiff, divorce specialists, reviews, fees, and so on, then assembles one answer from the lot. The full mechanic is in our query fan-out guide.

So the rewrite is real, and it does change your customer's words. The only question that matters for a local business is a narrow one: when the words get changed, does the town stay in?

The finding: location is almost always preserved

Start with how the mechanism is built, because that alone points to the answer. Google's own description of query fan-out is that the engine issues "a multitude of queries" on the searcher's behalf to gather local, specific results. An engine designed to return useful local answers has every reason to carry the town through into those sub-searches and no reason to discard it. The behaviour you would predict from the mechanism is that geography stays in.

That prediction is borne out by the one study that has measured it directly. A caveat first: this measurement comes from Profound, a company that sells AI-visibility tracking, so read it as the vendor's own research rather than a neutral referee, and weigh it against the independent evidence below. With that noted, Profound tracked a random sample of 10,000 prompts across ChatGPT, Perplexity and Microsoft Copilot over a 14-day window, capturing every search string each engine generated from each prompt, then measured which parts of the original question survived the rewrite and which were thrown away. The pattern was consistent across all three engines:

That is the practical heart of it. Because geography is the detail the rewrite is least likely to discard, a business with genuine local relevance can treat its location as the one signal it can reliably count on carrying through.

Read that against the list above. The things the rewrite happily discards (your customer's adjectives, their budget, the competitor they half-remembered) are the things you cannot control anyway. The one thing it keeps is the one thing you can own completely: where you are. The filter most likely to survive is the filter you were always going to win on.

What backs this up beyond one study

A single vendor study is a starting point, not proof, so it helps that the finding sits inside a wider, independent pattern.

Google itself reports that AI Overviews are "driving over 10% increase in usage of Google for the types of queries that show AI Overviews" in its biggest markets (Google, May 2025)[2]. People are not searching less in the AI era; they are searching more, and a large slice of that searching carries local intent ("near me", "in [town]", "open now"). It would make little sense for an engine chasing helpful local answers to discard the single word that makes a local answer local. The location-preserved finding is, in plain terms, the engines behaving the way you would expect them to behave if they wanted to be useful.

Independent measurement of how people actually search locally puts a floor under all of this. BrightLocal's Local Consumer Review Survey 2026 found that the use of AI tools like ChatGPT for local business recommendations "has grown rapidly, rising from 6% last year to 45% and becoming the third most popular source of business recommendations" (BrightLocal, Local Consumer Review Survey 2026)[3]. AI answers are now an ordinary part of how customers pick a local business, which is exactly the setting where the town has to survive the rewrite. Nearly half of local searchers are already asking these tools, and the tools are keeping their location in.

The second gift: "best [thing]" is the safest question shape there is

There is a smaller finding alongside the big one, just as useful, and it shapes what you should actually write on your website.

Not every question shape survives the rewrite equally well. Prompts framed as "best [product or service]" tend to be the most stable across every engine, holding their shape more often than other phrasings. By contrast, "which" and "where" questions are among the least stable, collapsing into bare keywords more readily.

Now put the two findings together, because this is where it gets practical. The most durable question shape is "best [thing]", and the most durable detail inside any question is location. Stack them and you get the single most stable kind of query in AI search:

best [your service] in [your town]

"Best emergency electrician in Bristol." "Best family solicitor in Cardiff." "Best bookkeeper for a limited company in Leeds." These are exactly the questions a local customer asks, and they are the questions least likely to be mangled on the way to the retrieval layer. The way your customer naturally phrases the search and the way the engine actually searches line up almost perfectly. That is rare, and it is squarely in your favour.

A national chain cannot lean on this the way you can. "Best [thing] in [town]" forces a local comparison, and a brand with no genuine presence in that town has little to bring to it. You only have to be the clearest, best-documented, best-reviewed answer in one place. That is a far smaller hill to climb than competing everywhere at once, and it is the hill local businesses were built to own. We make the broader version of this argument in the pillar guide to AEO for local businesses.

What this does not mean (so you do not overclaim to yourself)

It would be easy to read all this as "AI loves local businesses, relax". It is more precise, and more useful, to be clear about the limits.

Surviving the rewrite is not the same as being named. Location staying in the search only guarantees that the engine looks for local answers. Whether your business is one of the names it returns depends on everything else: a complete Google Business Profile, recent reviews, clear service and area pages, consistent contact details across the web. Geography gets you into the right race. The rest of the work decides whether you place. The pillar guide lays out that work step by step.

The source mix is not uniform. Which sources an engine leans on shifts from one model to the next, and no single trick guarantees a mention. Different engines read different corners of the web, which is why a presence spread across your site, your profile, reviews and genuine off-site mentions beats betting everything on one page. We break down each engine's source preferences in how AI engines pick their sources.

The answer still varies day to day. Ask the same engine the same local question twice and the names can change. That is normal behaviour, not a sign you have been dropped. It is a reason to build durable signals rather than chase one perfect answer on one afternoon.

None of these caveats undoes the good news; they sharpen it. Location is the door that AI holds open for you. Walking through it is the rest of the job.

How to make geography work as hard as it can

If your town is the detail most likely to survive the rewrite, the obvious move is to make your town unmistakable everywhere AI reads. Here is the work, ordered roughly from highest impact down.

1. Give every area you serve its own real page

The most common mistake is one thin "Areas we cover" page listing twenty towns. To an AI engine running a local search, that page answers none of the specific "best [thing] in [town]" questions well. Build a genuine page for each town or district you actually serve, with real, specific content: the services you offer there, local landmarks or districts, response times to that area, jobs you have done nearby. One strong local page beats a list of twenty place names every time.

2. Put the location in the words, not just the map

AI reads text. A page that mentions your town once in the footer is weaker than one that names it naturally throughout: in the heading, in the service descriptions, in the questions and answers. Write the way your customer searches. "How much does emergency boiler repair cost in Manchester?" is a sentence your page can be the direct answer to. Aim for the "best [service] in [town]" phrasing in your own copy, because it is the shape the engines handle most cleanly.

3. Make your Google Business Profile complete and current

For local questions, this profile is one of the most heavily read sources there is. Claim it, then fill in everything: exact business name, address, phone, opening hours, every service, every area you cover, photos and a clear description. The location signals here are about as authoritative as a local signal gets, and a half-finished profile is the quiet reason many businesses never appear in local AI answers at all.

4. Keep your name, address and phone identical everywhere

Your business name, address and phone number should match exactly across your website, your Google profile and every directory and listing. An old address on one site and a new one on another makes the engine less sure where you really are, and less likely to name you for a "near me" question. This is dull housekeeping that quietly decides a lot.

5. Earn reviews that mention the place

Reviews are the strongest trust signal AI has for a local business, and reviews that name the town ("brilliant plumber, sorted our leak in Headingley within the hour") double as location signals. Ask every satisfied customer for a review, keep them current, post them where your trade is judged, and reply. Steady, recent, local reviews are exactly the durable signal that survives the day-to-day wobble in AI answers.

6. Exist on the local web beyond your own site

AI reads the wider web, not just your homepage. Local directories, trade bodies, "find an installer" pages, community sites and genuine local press all reinforce where you are and what you do. Accurate and consistent everywhere is the rule. You are not trying to be everywhere; you are trying to be unmistakably, consistently here.

How to check it for yourself, today, for free

You do not need a tool to test any of this. Do it now:

  1. Open ChatGPT, Google's AI Mode and Perplexity.
  2. Ask each the question a real customer would, in the durable shape: "best [your service] in [your town]", and a couple of variations like "[your trade] near me for [the job]".
  3. Note whether the answer is genuinely local (it almost always will be, because the town survives) and whether you are one of the names.
  4. Repeat a few times. Answers vary, so look for the pattern, not a single result.

If the answers are clearly about your town but you are never named, you have learned the most useful thing possible: location is not your problem, visibility is. The door is open and you are not yet walking through it. Look at who is named and what they have that you do not, usually more recent reviews, clearer local pages, a complete profile, and you have your to-do list.

Frequently asked questions

Does AI really keep my town when it rewrites my customer's search?

On the best available evidence, almost always. The measurement described above found location "almost always preserved" through the rewrite on all three major engines, while adjectives, price filters and brand lists were frequently dropped. Geography is the detail most likely to survive, not the first thing cut.

So is location all I need to get named by AI?

No. Location surviving the rewrite only means the engine is looking for local answers. Whether you are one of the names depends on your Google Business Profile, recent reviews, clear local pages and consistent details across the web. Geography gets you into the right race; the rest of the work decides whether you place.

What is the best way to phrase my pages?

Around "best [service] in [town]". That shape is both the way customers naturally search and the question form least likely to be mangled in the rewrite, so your wording and the engine's actual search line up. Build a real page for each area you serve rather than one list of place names.

Will a national brand still beat me because it is bigger?

Not automatically, and less often than you would think for local questions. "Best [thing] in [town]" forces a local comparison, and a national brand with no genuine presence in your town has little to bring to it. A well-documented, well-reviewed local business is frequently the better answer.

Why does the answer change when I ask twice?

That is normal. AI answers vary from one run to the next, especially on ChatGPT, which rarely searches the same way twice. It is not a sign you have been dropped. Build durable signals (steady reviews, consistent information, real local presence) rather than chasing one perfect answer on one day.

Where to start

If you take one thing from this, take the reassurance: AI search is not quietly erasing your geography. Your town is the detail the machine is most careful to keep. The job is not to defend your location, it is to make it impossible to miss, with a real page for every area you serve, a complete Google profile, recent local reviews and consistent details across the web.

If you would rather have this checked for you first, a free QBiz Leads AI visibility check scans your website in about thirty seconds and returns a clear pass or fail on the key signals that decide whether AI tools can find and recommend your business, with a prioritized list of what to fix. It is the free first check before you invest in anything else.

Get your AI Visibility audit →

Sources

  • [1] Profound, "What AI Engines Actually Search For," 2026: https://www.tryprofound.com/blog/what-ai-engines-actually-search-for (vendor research; Profound is a company selling AI-visibility tracking, so treat its figures as the vendor's own; sample of 10,000 prompts across ChatGPT, Perplexity and Copilot over 14 days; location "almost always preserved" through the rewrite; "best [product or service]" prompts the most stable shape at 65% on Perplexity, 52% on Copilot, 39% on ChatGPT)
  • [2] Google, "AI in Search: Going beyond information to intelligence," 20 May 2025: https://blog.google/products/search/google-search-ai-mode-update/ (primary; query fan-out defined as "breaking down your question into subtopics and issuing a multitude of queries simultaneously on your behalf"; AI Overviews "driving over 10% increase in usage" for the query types that show them in the US and India)
  • [3] BrightLocal, "Local Consumer Review Survey 2026": https://www.brightlocal.com/research/local-consumer-review-survey/ (independent; use of AI tools like ChatGPT for local business recommendations "has grown rapidly, rising from 6% last year to 45% and becoming the third most popular source of business recommendations"; Google's review share slipped from 83% in 2025 to 71%; 97% of consumers read reviews)

Leave a comment

Thoughts on this post? Leave a comment below. Comments are moderated before they appear, so yours will not show on the page straight away.

Your email is used only to contact you about your comment if needed — it is never published.

Comments

No comments yet. Be the first to leave one above.