AI Visibility Check for a Local Business: City, Service, Reviews

Why a local check is not the general check

A local business is checked by place and by job together. A store that ships nationwide asks, "who sells X." A plumber, a dentist, a restaurant or a pest control company asks, "who should I call for X, near here." The answers are shaped by where the assistant thinks the person is, and by sources that barely matter elsewhere: map listings, review sites, local forums and local press.

That changes the method in three ways. The questions must carry a place. The sources you look at are different. And the people named in your place are a smaller, more specific group, which makes the competitor reading sharper than in most categories.

This page adapts the general method for that situation. If you have not read the basics yet, the basic method for running a check covers what a single run can and cannot prove. Here we assume you have a few hours, a notebook, and a business with a service area.

Start with your real service area

Before writing a single question, write down where you actually work. Not where you wish you worked. A common mistake is testing for the biggest nearby city, because it feels more important, while most of your customers come from three smaller towns.

Make a short list.

• The town or neighborhood where your business is based.
• Two or three other places where most of your customers live.
• One bigger city nearby that people may use as a search term.

Each of those becomes a place name you will put in questions. If you serve a wide area, do not try to test every town. Pick the three that bring you the most jobs, and add more later.

Also list the specific jobs you want to be known for. A plumbing company may fix leaks, replace water heaters and clear drains. They are different questions, and an assistant may name different people for each. Our page on AI search optimization for a local business covers what to do when the list shows gaps. For now, simply write the list down.

Building the question grid

Combine your places and your jobs. The diagram above shows the four kinds of local question. You want most of your set in the top right cell, where a place and a job are both named, because that is how most local buyers actually ask.

For a first run, aim for about twelve questions.

• Six place and job questions. Two jobs across three places. "Who is a good plumber for a leaking water heater in Ocala."
• Three job-only questions. No place, to see what the assistant assumes. "Who should I call when my water heater leaks."
• Two near-me style questions. "Plumber near me open on Sunday." These depend heavily on your location, so run them from where your customers are.
• One comparison. "Is it better to repair or replace a 12-year-old water heater, and who can do it in Ocala."

Write them the way a stressed or hurried person would say them. Add the constraint that comes up in your real calls, such as "same day," "emergency," "under a certain price," "licensed" or "family-owned."

Keep the exact wording fixed. How to write and then freeze a set is covered in the question set guide.

Running the check from where your customers are

Location shapes local answers more than anything else. An assistant may use your device's location, your internet connection, or only the place you type. So the cleanest approach is to put the place in the question and also to run the check from the area you serve.

Use a private window for each assistant, and check at least ChatGPT, Perplexity, Claude and Gemini. Our free manual check lays out the basic routine and the logged-out rule in full, so we will not repeat it here. The additions for a local run are these.

• Run the near-me questions on your phone, since that is where your customers search.
• Note whether the assistant showed a map or a list of nearby places. Some do. That tells you the answer is drawing on map data.
• Check Google AI Mode too, because Google owns the map data. See how to check Google AI Mode.
• If an assistant asks for your location, say what you did. Allowed or declined, write it down.

For each answer, record who was named, in what order, with what wording, and which sources were shown. Do this for every question. It is slow, and it is the part that matters.

The sources that matter for a local business

For national brands, assistants lean on a wide range of web pages. For local businesses, the sources tend to be narrower and more specific. When you read the source links, look for these types.

Map and directory listings. The business profile on a mapping service, with hours, categories, photos and reviews. If your listing is incomplete or inconsistent with your website, that is the first thing to tidy.

Review sites and review text. What customers wrote, and how often. Assistants often echo the phrases that appear in reviews, such as "responsive" or "great with older homes." Our page on reviews and AI recommendations goes into this link.

Local forums and community threads. A neighborhood discussion where someone asked for a recommendation and neighbors answered.

Local press and "best of" lists. Newspapers, regional magazines, chamber of commerce pages.

Your own website. Often the least decisive of the five, but where the specific facts live: services, areas, licenses, years in business.

We have written separately about why assistants often favor local operators over chains, in why AI recommends local businesses over chains. If you are a small business worried about big names, that page is worth a read. It rests on third-party mentions, which is where your check will point you too.

Read the words used about you

Being named is only part of the picture. The words around your name tell you how you are being described, and for a local business, they often echo customer reviews closely.

When you are named, copy the sentence. Then look for the following.

• The reason given. "Known for fast response," "specializes in older homes," "highly rated." Is it a reason you would want a customer to hear?
• The facts stated. Hours, areas served, pricing hints, licensing. Are they right? Wrong facts here are worth fixing at the source.
• The framing. Is your business the default option, a budget choice, a specialist, or an afterthought at the end of a list?

If you are not named, read the sentences used for the people who are. Their reasons are the language the assistant is currently using for your category and your area. That is useful reading, and it is the beginning of the competitor reading we describe in checking competitors in AI search.

Do not take every sentence as true. Assistants sometimes get details wrong. A fact in an answer is a clue about what the sources say, not a verdict about the business.

Check your own listing alongside

While you are running the check, spend ten minutes on the places an assistant would find you. This is not an audit, just a sanity check that the basics match.

• Is your business name written the same way in your map listing, on your website and on review sites?
• Are your address and phone number identical wherever they appear?
• Do the services listed in your map profile match the jobs you want to be asked about?
• Does your website say, in plain text near the top of a page, which towns you serve and what you do?

Small differences matter more than they sound. If one source calls you "Smith and Sons Plumbing" and another calls you "Smith Plumbing LLC," a machine reading both has to guess they are the same business. Making them match costs little and removes the guesswork.

For a fuller walk through this, the page on the AI visibility checklist gives an ordered list of checks, cheapest first. Use it after your question run, not instead of it.

How often a local business should check

Once a month is plenty for most local businesses. Local answers do not swing wildly day to day, though they can shift when new reviews arrive or when a competitor gets a press mention.

A sensible rhythm is this.

• Now: one full run of your twelve questions across four assistants.
• A month later: the same twelve again, with identical wording.
• After a real change: if you collect a burst of new reviews, update your listing, or publish a major page, run the set again a few weeks later.

Keep the notes in one place with the same columns each time. We suggest a layout in an AI visibility tracking spreadsheet.

Remember that assistants vary from one run to the next. A business named once in a month is not reliably visible, and a business missing once is not reliably invisible. Two or three runs in the same week will tell you whether a result is steady. For more on that, see why AI answers change.

What to do with the result

Sort your results into three piles.

Questions where you are named. Good. Note the reasons the assistants gave and check that your pages and listings support them.

Questions where others are named and you are not. This is your to-do list. For each, ask what the named businesses have that you do not: more reviews, a clearer service page, a local article, a forum mention.

Questions nobody useful is named for. Rare, and sometimes an opening. If the assistants give vague answers, a clear, specific page from you may become the answer.

For the interpretation side, read reading your AI visibility results, and for the action side, what to do after a check. For the underlying reasons assistants choose the way they do, see how AI assistants pick businesses.

One word of caution. Do not run out and buy fifty reviews or fabricate mentions. The aim is to be genuinely known for your work in your area, in places other people write. That takes time, and it holds.

A worked example: a roofing company

Here is how the pieces fit together for an imaginary roofing company serving three towns. The names are invented, and so are the answers, which exist only to show how you would write things down.

The owner lists three towns and two jobs, leak repair and full replacement. That gives six place and job questions. She adds two job-only questions, two near-me questions and one comparison, which makes eleven.

On the first run she asks all eleven in four assistants, from her phone, in private windows. In her notes, three businesses keep appearing. One is a long-established firm with hundreds of customer reviews. One is a franchise whose own site lists every town. One is a small crew that a local newspaper had written up the previous spring.

She opens the sources. The established firm is carried by reviews. The franchise is carried by its own town pages. The small crew appears because of the one article, plus a neighborhood forum thread where a customer thanked them.

Her own business is named in two of forty-four answers. She finds that her website has one page for all roofing and does not mention two of the three towns. Her map listing says "roofing contractor" but not "repair." Her reviews are good but few.

The to-do list writes itself. Fast: a page per job and per town, and a listing that says what she does. Medium: a steady habit of asking happy customers for a review that says what was fixed. Slow: be worth writing about locally.

That is a small, honest plan from one afternoon of checking.

How RunOctopus handles the follow-through

RunOctopus does not offer a free tool that checks what AI assistants say about your local business. The manual method on this page is the free way to see it.

Where we help is what follows. RunOctopus is a done-for-you service. Ollie, our AI, researches your business and writes and publishes pages directly on your own website, alongside a team that plans, writes, reviews and publishes. It works with Shopify, WordPress, WooCommerce and Wix. You do not install anything, approve each page, upload anything or schedule anything. For a local business that means service and area pages that actually say what you do and where.

Paying customers receive reports on where ChatGPT, Perplexity, Claude and Gemini name their business. The promise we make is more visibility on Google and in AI answers within 90 days, measured against the 90 days before joining, or every dollar back.

You can see what Ollie would plan for your site with the free site read. Paste your address and it reads the site and shows the plan. No account, no call. It does not check AI answers.