Yes, a small local business can get recommended by ChatGPT, and it does not take a national budget to get there. It takes being checkable: one identity that agrees with itself everywhere, a specific answer to the situation your buyer is actually in, and sources other than you confirming both. There is no listing fee, no submission form and no setting that turns the recommendation on. The assistant assembles its answer from whatever it can find and check about the businesses in your category. So the question is not how big you are. It is how much of you is on the record.
Most owners arrive at this question already braced for a no. You picture the franchise groups, the private-equity roll-up down the road, the firm with an actual marketing department, and you assume the shortlist got sorted by size long before anyone typed the question. It is a fair guess. It is also the wrong shape of the problem, and the published numbers are the fastest way to see why.
The largest look at local AI visibility so far measured nearly 350,000 locations belonging to 2,751 multi-location brands. ChatGPT recommended 1.2 percent of them. Those are brands with marketing budgets and, on average, more than a hundred locations apiece, and the assistant walked past almost all of them. Scale bought them shelf space. It did not buy them the sentence.
So the room you were afraid of is emptier than it looks. Small is not the disqualifier. Vague is. This guide covers what the evidence actually says about size, what an assistant has to be able to check before it will name anyone, where a small local specialist has a real advantage over the chain, whether the different assistants behave differently, and where to put your first month of effort.
What the Evidence Says About Size
Start with the study, because it is the closest thing to a scoreboard the category has. SOCi's 2026 Local Visibility Index, reported in January 2026, analyzed nearly 350,000 locations across 2,751 multi-location brands and counted how often each was recommended by an AI assistant. ChatGPT named 1.2 percent of locations, Gemini 11 percent and Perplexity 7.4 percent. The same locations appeared in Google's local three-pack 35.9 percent of the time.
Read the caveat before the conclusion, because it matters. That is vendor-produced research, and it measured chains and franchise groups rather than independent practices, so it cannot tell you how often a two-person firm gets recommended. What it can tell you is what happened to the businesses you were comparing yourself against. Size, spend and store count were all present in that sample. The recommendation mostly was not.
Google's own description of local results points the same way. Its guidance for business owners names three factors behind local results, relevance, distance and prominence, and it is explicit that there is no way to request or pay for a better local ranking. Relevance is how well your profile matches what someone asked for. Prominence includes how many other sites talk about you. Neither one has a revenue threshold in it. Nobody has that switch.
None of this proves a small firm gets named instead. No published study has tested that yet, and anyone telling you otherwise is guessing. What it does is shrink the fear that brought you here, because the businesses you assumed had already taken your place are, in the only large sample published so far, mostly absent from it too.
What an Assistant Has to Check Before It Names You
When someone asks an assistant who to call, it has to produce a name it can stand behind for a stranger who will act on the answer. That is closer to a verification job than a popularity contest. As we put it in our own work: your buyer's AI already knows their situation. It recommends whoever it can verify will look after exactly that person. Fame is not the test. Fit is, and fit has to be provable.
Nobody outside the labs can see how that decision is actually made. What follows is a working model built from what these systems visibly cite and what the platforms publish about themselves, not a peek inside the machine. It runs along three plain lines, and each one is buildable at any size.
First, are you a real, single, coherent business. That means the name, address, phone, hours, categories, people and credentials agree with each other on your site, your Google Business Profile and everywhere else a machine looks. Contradictions are not a small mistake. They leave a retrieval system holding two versions of you with no way to tell which is current, which is exactly what entity identity work resolves. Unglamorous, and the cheapest item on this list.
Second, does your own site say what you actually do, for whom, in what situations. Your website carries more of this load than most owners expect. In BrightLocal's December 2024 sample of 800 local searches across 20 US cities and 20 verticals, business websites supplied 58 percent of the sources ChatGPT Search displayed, ahead of mentions at 27 percent and directories at 15 percent.
That counts which sources were displayed rather than which ones decided anything, and it is a dated snapshot. It still points somewhere useful: your own pages were the most common source on show, and a page of adjectives gives a machine nothing to repeat.
Third, does anyone besides you say so. Reviews, citations, local press and independent listings are the corroboration layer. You can manage your own listings, and you should, but the reviews and the coverage are the part you cannot write yourself. Google says more reviews and positive ratings can help your local ranking, and its own local guidance treats prominence as partly a function of what the rest of the web says about you. Foundation, corroboration and knowledge worth citing: those three together are what turns a good business into a recommendable one.
Where a Small Local Business Has the Edge
Here is the part nobody tells you about being small. The advantage is specificity. Specificity is cheap.
Buyers rarely ask assistants generic questions. They ask loaded, situational ones: an estate attorney for a blended family, a dentist who is good with an anxious adult, a bookkeeper who can clean up three years of neglected books before a bank meeting. A national brand usually meets those with a category page covering every market and every service at once. You can answer one of them completely, in the words your clients actually use, from cases you have personally handled.
The assistant is not choosing the biggest business. It is choosing the one it can describe without guessing.
Google's guidance on optimizing for its own generative AI features lands on the same instruction, and it is worth reading in the plainest terms. It tells site owners to create content themselves based on what they know, and to consider what in-depth experience they can bring. It also says structured data is not required to appear in generative AI features, and that no special markup exists for it. The special files you were told to add are not the lever. Your judgment is.
The same page draws a hard line on the tactic small firms are most often sold: producing many near-identical pages for query variations, primarily to influence rankings or AI responses, violates Google's scaled content abuse policy. Fifty thin suburb pages is not a strategy. It is exposure. Four pages that genuinely answer four real situations will do more for you, and you can write them from the last month of your own consultations.
- Name the specific service and the specific situation on the same page, not the category.
- Say where you actually serve, in the words locals use, and keep it consistent with your profile.
- Put your reasoning on the page: how you decide, what you would do differently, what you would not take on.
- Publish the boring specifics too, including price ranges, timelines and what happens at the first appointment.
Do the Assistants Behave Differently?
They do, more for local questions than for almost anything else, and the differences are about where each one gets its business data rather than about what makes a business worth recommending.
The same January 2026 index found the recommendation rate varied widely by assistant, from 1.2 percent on ChatGPT to 11 percent on Gemini. It also reported business profile information at around 68 percent accurate on ChatGPT and Perplexity in that sample, against 100 percent on Gemini, which is grounded in Google Maps. Read those as figures from one vendor's test rather than fixed properties of the products. The useful part is the shape: your business details reach the assistants through different pipes, and the pipes are not equally clean.
The supply lines keep moving too. On July 23, 2026, Yelp and OpenAI announced a licensing arrangement putting Yelp reviews, ratings, photos and business information into ChatGPT's local responses, with Yelp branding and links appearing when its content is used. The practical read for an owner is not to chase whichever platform signed a deal last month. It is that your listing and review presence beyond Google now feeds a second assistant you were not thinking about.
Which is why we test five engines rather than one. We test the five engines your buyers actually ask, ChatGPT, Gemini, Perplexity, Claude and Grok, with the questions they actually ask. One screenshot from one assistant on one afternoon is an anecdote. The evidence underneath is one body of work. It is the same work for all of them.
Where to Start This Month
Start by looking, not buying. Ask two or three assistants the exact question a good client would ask, in your city, in their words, and write down what comes back and who it names. That reading takes an afternoon. It will tell you more than any vendor pitch.
Then fix the checkable things in order. Complete and correct your Google Business Profile, since Google says complete and accurate information makes a business more likely to appear in relevant local results, and its generative AI guidance says the profile can help your services show up in AI responses as well as ordinary ones. Reconcile your details everywhere else they appear.
Then ask recent clients for reviews that describe the specific situation you solved, and write the pages only you can write. If your SEO agency already covers some of this, good, and the honest question is which parts it does not touch.
If you would rather see the whole decision from the machine's side first, the free Recommendation Check scores your visibility across 109 checkpoints and shows what five AI engines answer when buyers ask who to hire in your category. You get a plain map of what is missing and what order to fix it in, before you spend anything. For the full walk-through of the work itself, the step-by-step version lives here, and what an assistant needs before it will name anyone is worth a read too.
One honest boundary, since you will hear the opposite from somebody this quarter. Nobody controls what a model says, and nobody serious will promise you a mention. The work makes you easier to verify and harder to summarize without, and then you measure what changes. Readiness is buildable. Outcomes are earned.
Want to Learn More?
Probably Genius builds the public record around expert-led businesses so that machines and buyers can both check it. It starts with a conversation that gets your expertise out of your head, and turns into 12 done-for-you thought-leadership articles every month, planned to produce approximately 60 clear Knowledge Entries, each scored 0 to 100 through the Integrity Gate, where nothing publishes under 80. If your market knows exactly how good you are and the machines do not, small was never the problem. You're probably a genius at what you do. We make sure AI gets the memo.