AI Visibility Statistics From 1,352 Real AI Answers
AI Visibility Statistics From 1,352 Real AI Answers
These AI visibility statistics come from 1,352 real answers about 17 local businesses, and the headline number is 18.1%. That is how often the AI engines named the business we were asking about, which is all AI visibility means in practice: whether an assistant says your name when a customer asks it for a recommendation. We put one question to five engines, "What [service] company do you recommend in [city]?", across 79 cities between August 29 and September 22, 2026. In 162 of the 272 service and city combinations we tested, no engine named the business at all.
Key takeaways
- AI named the business being measured in 18.1% of answers (245 of 1,352). The typical business in the group appeared in 17.1% of its own answers.
- 14 of the 17 businesses were named in fewer than half their answers. Two were never named once, by any engine, in any city.
- In 162 of 272 service and city combinations, no engine named the business. That is 60% of everything we tested.
- When AI did name a business, it named it first 51% of the time. Being mentioned at all is the hard part.
- AI named about 3.5 businesses per answer, while BrightLocal's 2026 consumer research found 72% of buyers look at three or fewer before they decide.
How we measured this
Every number here traces back to one question, asked the same way every time: "What [service] company do you recommend in [city]?" We ran it for each service a business actually sells, in each city it actually serves.
Five engines answered: ChatGPT, Google AI Mode, Gemini, Claude and Perplexity. Grok appears in our report template but was never run, so every figure on this page reflects five engines, not six.
That worked out to 1,360 queries. We got 1,352 answers back. The eight that returned nothing were dropped rather than counted as failures, because an engine that times out has not made a judgment about anyone.
An engine that answered and named no businesses at all is a different thing, and those count as a miss. That distinction matters when you set our rates beside other published AI visibility data, so we are stating it up front.
One of the 17 reports used an earlier version of our format, covering a single service across eight markets with earlier model versions. We left it in and are telling you it is there.
If you want to see what the underlying websites looked like, we published audit findings from 41 local business sites separately. This piece measures what the AI said. That one measures what it had to work with.
GetLocalLeads.AI, an AI visibility agency for local service businesses, runs this same test every month.
How often does AI recommend local businesses?
Across all 17 businesses, AI named the business in 245 of 1,352 answers. That is 18.1%. The median business, the one sitting in the middle of the group, landed at 17.1%, so the average is not being dragged around by one outlier.
Underneath that, the spread is wide and mostly bad. Fourteen of the 17 were named in fewer than half of their answers. Ten of the 17 came in under a quarter. Two were never named at all, in any city, for any service, by any of the five engines. Eleven of the 17 had a competitor that AI named more often than them, and ten of the 17 reports came back with a Critical verdict.
It is fair to ask whether 18.1% is a bad number. Nobody expects to be named in every answer, and there are only so many slots. But flip it around for the businesses we measured.
For every ten customers who asked an assistant for a recommendation, roughly eight heard a list that did not include the business. Those eight never bounced off its website or ignored its ad. They were handed three or four names, none of them its own, so it was never in the running. There is no impression count anywhere that would show an owner that happening.
Here is the part that changes how you should read all of it. Fourteen of the 17 businesses were ranked number one by at least one engine, at least once. The knowledge is in there.
These are not businesses the internet has never heard of. They are businesses the engines know inconsistently, which is a different problem with a different fix. If you have wondered why you're missing from answers where you clearly belong, that gap between "known" and "named" is usually where it lives.
Seeing where you land is the first thing we do for a new client.
The same business, different city, different answer
Geography moved these numbers more than anything else we looked at.
We tested 96 business and city pairs. In 44 of them, 46%, the business was never named for any of its services in that city. Not ranked low. Absent.
The pattern inside that was consistent across the group. For businesses measured in more than one city, the strongest city averaged 45% of answers while every other city combined averaged 16%. Twelve out of twelve had a strongest city that beat the rest of their footprint. Eleven of 13 multi-city businesses had at least one city sitting at zero.
One Georgia business shows the shape of it clearly, and it is one case rather than a rule. Across its two home cities it was named in all 30 answers, every service, every engine, without a miss. In the largest expansion market it serves, it was named in none of the 15.
Businesses measured in a single city averaged 40.4% against 17.1% for multi-city businesses, but that comparison rests on only four businesses and 57 answers, so treat it as a direction, not a finding. The useful version is simpler: for the businesses we measured, AI visibility thinned out fast away from wherever the business is physically rooted, which is the same problem that makes cities without offices hard to win in ordinary local search.
We map this city by city before recommending anything.
Not every engine answers the same way
Gemini named the measured business in 28.8% of its answers. Google AI Mode managed 17.1%, Perplexity 16.2%, Claude 14.7%, and ChatGPT 14.0%. Google AI Mode is the conversational version of Google search, the one that answers in paragraphs instead of a page of blue links.
Before reading too much into that, here is the caveat that belongs in the same breath. Gemini names about 4.7 businesses per answer. ChatGPT names about 2.7. A longer list gives everybody more chances to appear, so some of Gemini's lead is generosity rather than knowledge.
Claude did something none of the others did at any scale: in 11.8% of its answers it declined to name a single business. An answer with no recommendation in it is still an answer a customer walked away from.
Each engine, taken alone, missed about six of the 17 businesses completely. Only seven of the 17 were named at least once by all five engines. Five of the 17 were named by one engine or none.
The practical read on all this is that engine choice is not a strategy decision you get to make. Your customer makes it, based on whatever app is already on their phone, and you have no visibility into which one they opened. Treating any single engine as the target means accepting that you are invisible to whoever picked differently.
The number that surprised us most: the average business had a 35 percentage point gap between its best engine and its worst. Same business, same question, wildly different answers depending on which assistant the customer happened to open. Anyone optimizing for one engine is reading a single column of a much wider table, which is why we keep saying beyond ChatGPT is where the work is.
Our audit covers all five engines, not just the one everyone names.
Being named is a switch, not a ladder
This is the finding we would keep if we had to throw out the rest.
Of the 245 times AI named one of these businesses, 124 of them, 51%, put it in the first position. Eighty-six percent landed in the top three. The average position across every mention was 2.0.
Now set that beside the misses. In 162 of 272 combinations, nobody was named at all. Only 10 of those 272 combinations had every engine name the business, which is 9% of the combinations where even one engine did.
Put those two facts together and the shape is clear. There is almost no middle. A business is either absent from an answer or sitting at the top of it. Very few land at four or five.
The engines do not even agree with each other on who belongs first. Across all 272 combinations, every engine picked the same number one business only nine times, 3.3%. We matched on exact business names, so spelling variants counted as disagreement and the true figure is a little higher. It is still tiny.
Traditional search trained everyone to think in ladders. You are on page two, you work to page one, you work into the top three. AI answers do not have a page two. There is no ranking below the answer that you are slowly climbing out of, because a customer never sees a fourth screen of ChatGPT results. Understanding how AI chooses a business matters more than any ranking report, because the whole game is getting inside the answer at all.
Why position four does not exist
Across 1,352 answers, the engines named 4,715 businesses. That works out to about 3.5 per answer.
BrightLocal's 2026 consumer research found that 72% of consumers look at three or fewer businesses before deciding, and only 1% look at more than ten.
Line those up. The assistant hands over three or four names. The buyer checks three or fewer. If you are the fourth name in an AI answer, you are already outside what most people will look at, and you are competing for a slot that shrinks every time the model writes a shorter list.
Of all 4,715 business names the engines produced across our tests, 245 belonged to the business we were measuring. That is 5.2%. The other 94.8% were competitors, and reports averaged 107 different companies named apiece.
One more thing worth knowing: 12 of the 49 service lines we tested were never named in a single city. A business can look fine on its headline rate while one of the services it sells is invisible everywhere it operates, which is something a single-number score will never surface. Getting cited has to happen service by service, not once for the whole company.
If you want the shortlist worked out for your market, book a call.
How this compares to the bigger studies
We are not the only ones measuring this, and the larger studies are worth knowing about. Most local SEO statistics you will find still measure Google: map pack appearances, review counts, click share. Very few measure what the assistants actually say, and the ones that do tend to study large chains.
SOCi's 2026 Local Visibility Index, published in March 2026, analyzed roughly 350,000 business locations across 3.2 million AI queries. It found brand locations appeared in AI recommendations 6.5% of the time, with Gemini recommending brands 11% of the time, against 36% for Google's local 3-pack. Coverage of that index also reported ChatGPT at 1.2% and Perplexity at 7.4%, though those per-platform splits were not on SOCi's own public page.
Our 18.1% is much higher than their 6.5%, and the reason matters. We asked about each business using its own service, in its own city. That is the most favorable prompt a business could get. SOCi sampled far more broadly. A targeted question is the ceiling, not the average, so read our figure as the best case and theirs as the wider picture.
The two datasets agree on something neither of us set out to prove. Gemini names the most businesses, ChatGPT the fewest. Different methods, different scales, same order.
The universe is different too. The average brand in their index had about 67 locations. The reader we built this for has one.
We publish our method so you can hold us to it.
What these numbers cannot tell you
Seventeen businesses is a small group. These are AI visibility statistics from the businesses we measured, and nothing here supports a claim about local businesses in general. Where we break the group into segments the samples get smaller still, down to four businesses in one comparison.
Every report is a baseline, captured once. None has been re-run, so nothing on this page measures improvement, or shows any work we did having an effect. For what a change over time actually looks like, our client's climb is a separate piece with its own evidence.
These businesses were also not chosen at random. They were measured for sales conversations and client work, which is its own kind of selection.
And engines vary between runs. Ask the same question tomorrow and some of these numbers move, so part of every rate here is noise rather than signal.
We publish it anyway, because a number you can check beats a rounder number you cannot. Ask us for the figures behind any claim we make.
Frequently asked questions
How often does AI recommend local businesses?
In our testing, AI named the business being asked about in 18.1% of 1,352 answers. SOCi's 2026 index of multi-location brands put the figure at 6.5%. Where a given business lands depends heavily on how the question is asked, and a targeted prompt about its own service in its own city is the best case.
Why doesn't ChatGPT recommend my business?
Usually because it does not have enough consistent information to be confident naming you. In our data, 14 of 17 businesses were named first by at least one engine, so the information often exists but is not reaching every engine. You can check your business in a few minutes.
Which AI engine recommends local businesses most often?
Gemini, at 28.8% of its answers, followed by Google AI Mode at 17.1%. ChatGPT was lowest at 14.0%. Gemini also lists more businesses per answer, so some of that lead comes from longer lists rather than better knowledge.
Is being named second or third good enough?
Third is usually fine, fourth usually is not. AI names about 3.5 businesses per answer and most buyers look at three or fewer, so the bottom of a list gets very little attention.
How do I check whether AI recommends my business?
Ask each engine which company it recommends for your service in your city, then write down every business it names and where you appear. Run it for each service and each city separately, because both change the answer. That is exactly the method behind this page, and AI visibility work starts with running it properly.
Run it on your own business this week
The method on this page is not proprietary. Open all five engines, ask about your own service in your own city, and write down who gets named. Ten minutes will tell you which of these numbers is yours, and the result is usually more specific than anything a monthly report has told you. If you would rather see the whole picture across every service and city you serve, GetLocalLeads.AI does this for a living.







