How Do I Get My Local Business Recommended by ChatGPT, Claude, Perplexity, and Gemini?
Most of the confident advice about which index each assistant reads is not documented by the vendors. Here is what is actually measurable: what gets cited, how much traffic it is worth, and the fundamentals that raise you across all of them at once.
The short answer is that every assistant recommends the business it can read, verify, and corroborate across the open web. The longer answer starts with a correction, because the version of this advice in circulation is mostly confident about the wrong things.
You will read that ChatGPT runs on Bing, Claude runs on Brave, and Perplexity runs on Yelp. Those claims range from partially documented to not documented at all, they change without notice, and building a strategy on them means rebuilding it every time a contract does. What is actually measurable is which surfaces recommend anyone, what they cite when they do, and which of your own assets they can read. That is what this covers.
How do AI assistants decide which local businesses to recommend?
They do not rank businesses. They assemble an answer: take the question, search, read what comes back, and put together a recommendation from whatever can be verified. Google ranks pages and shows ten. An assistant reads a stack of them and names two or three. There is no position to win, only a confidence threshold to clear.
Confidence is the mechanism worth understanding. When a model finds your business described consistently on your site, your profiles, a review platform and a local news mention, each source corroborates the others and your odds of being named climb. When it finds two addresses, three phone numbers and a website it cannot parse, it does not average them. It reaches for a competitor it is more sure about.
What is actually known about where each assistant looks
Less than the confident tables suggest. Here is the state of the evidence, at the strength the sources support.
What OpenAI saysThe launch post names no provider. The main search help page mentions third-party search providers and links to Microsoft's privacy statement. The Enterprise and Edu help page names Bing outright. OpenAI also runs its own search crawler, OAI-SearchBot.
What follows for youAssume it reads the open web broadly. Claim Bing Places, which is cheap and feeds Bing's local results, and make sure your site does not block OAI-SearchBot.
What Anthropic saysBrave appeared on Anthropic's subprocessor list in March 2025. Claude's help center names Bing for image search only, and names no provider for web results.
What follows for youBeing findable in more than one index is worth something. A Brave-specific strategy built on a subprocessor listing is not.
What Perplexity saysIt announced a Yelp integration for local searches in 2024, and Tripadvisor's January 2025 release says its one billion reviews and contributions feed Perplexity answers.
What follows for youYour Yelp profile is being read aloud. Fix it first if you are in food, travel or hospitality.
What Google runsIts own stack: Business Profile, Maps, reviews and the search index.
What follows for youThe local SEO you have already done carries over here most directly.
Two independent measurements do hold up, and both point at the same place. Across more than 28 million small-business queries in late 2025, Foundation Marketing and AirOps found Yelp holding about 62 percent of Perplexity's local citations, 4.9 times the next source (Foundation, 2026). And Whitespark, across 153 queries in 17 categories and 9 cities, found Facebook the dominant review source on Bing Places listings, appearing on nearly one and a half times as many businesses as Yelp, the next platform (Whitespark, 2025). The Facebook page you stopped updating years ago is doing more work than you think.
How much traffic is this actually worth?
Less than the hype, more than nothing, and the honest framing matters because it changes what you should spend.
When Ahrefs looked at roughly 35,000 sites in early 2025, AI sources sent 0.1 percent of total referral traffic (Ahrefs, 2025). A follow-up that June, across nearly 82,000 sites, put AI at 0.25 percent of total traffic. Growing, and still small.
The counterweight is intent. Seer Interactive compared conversion rates by source for one client and found visitors from ChatGPT converting at 15.9 percent, against 1.76 percent from Google organic (Seer Interactive, 2025).
Read that carefully, because it is routinely inflated in the retelling. It is one client, and the AI sample was just under 11,000 sessions next to almost 14 million from Google. Every assistant did convert better than Google organic for that client. Nobody has shown that it holds for yours.
So: tiny volume, unusually warm when it does arrive. That justifies fixing your fundamentals. It does not justify a retainer aimed at AI visibility alone.
The click-through claim everyone is still repeating
You will see that an AI Overview cuts organic click-through by about 61 percent. That figure was real, from Seer's September 2025 analysis (Seer Interactive, 2025). Seer themselves superseded it in April 2026, with a larger study across 53 brands, 5.47 million queries and 2.43 billion impressions, and the revision changes the conclusion rather than the decimal (Seer Interactive, 2026).
Organic click-through on AI Overview queries climbed from a low of 1.3 percent in December 2025 to 2.4 percent by February 2026. More importantly, Seer reframed what the earlier drop had been: not clicks disappearing, but impressions exploding on top of a roughly flat click count. For results where the brand was cited, clicks went from 398,000 in September 2025 to 400,000 in October, while impressions more than doubled, from 15.8 million to 33.1 million. The ratio collapsed. The traffic did not.
That is not a clean bill of health for AI Overviews, and Seer does not present it as one. On informational queries, being cited inside the Overview lifts clicks per impression by 120 percent over not being cited, and still leaves you 38 percent behind a result with no Overview at all.
A CTR figure with no impression figure beside it is not a measurement, it is a mood. This is the single most repeated statistic in AI search commentary, and its own author now explains it differently.
And how common AI Overviews actually are
Semrush, tracking more than 10 million keywords through 2025, found AI Overviews on 6.49 percent of queries in January, a peak of 24.61 percent in July, and 15.69 percent by November (Semrush, 2025).
Up sharply from January, down from the summer peak, and nowhere near the "half of all searches" figures that circulate.
Local is a different picture, and the headline number hides the useful part. Whitespark, across 540 queries in three cities and six industries, found AI Overviews on 68 percent of local business-type queries against 39 percent for the local pack (Whitespark, 2025). That 68 is an average across three kinds of search, and they behave very differently.
- Picking a business"personal injury lawyers in Phoenix"
- Asking what it costs"how much do personal injury lawyers charge in Phoenix?"
- Asking whether they need one"should I get a lawyer after a car accident in Phoenix?"
Show the numbers
| Kind of search | AI Overview | Local pack |
|---|---|---|
| Local intent | 15% | 93% |
| Informational intent | 92% | 6% |
| Hybrid intent | 97% | 17% |
On a plain local search, the local pack appeared 93 percent of the time and an AI Overview only 15. On the questions people ask before that, what it costs or whether they need one at all, AI Overviews appeared 92 and 97 percent of the time, and the pack only 6 and 17 percent. So the map pack still owns the moment someone picks a business. The AI answer owns the research that leads up to it. That is the number worth acting on.
What actually gets cited
BrightLocal ran 800 local searches in ChatGPT, across 20 industries and 20 cities, and recorded the sources behind each answer (BrightLocal, 2024).
- Business websitesThe business's own site58%
- Business mentionsPages naming the business. 39% of these were Wikipedia.27%
- DirectoriesListing sites15%
Two caveats worth carrying. The research dates from December 2024, which is old for this subject. And its "business mentions" category is not all editorial: Wikipedia alone is 39 percent of it.
Whitespark's 2026 Local Search Ranking Factors, polling 47 local search experts, puts three of the top five AI visibility factors in the citation family (Whitespark). The direction is consistent across both: being named on pages the model trusts matters about as much as your own site.
Reddit, and why single-platform strategies fail
Peec AI analyzed 30 million cited sources across ChatGPT, Google AI Mode, Gemini, Perplexity and AI Overviews, and found Reddit the most-cited domain overall (Peec AI). With an exception that matters here: for ChatGPT specifically, Wikipedia ranks first, not Reddit. In an article about ChatGPT, that is not a footnote.
Reddit is still worth your attention, because the "who do you recommend" threads in your city subreddit get read by machines for years after they are posted. Participate as yourself, say you are the owner, answer better than anyone else, and never astroturf.
But hold it loosely. Semrush found ChatGPT citing Reddit in close to 60 percent of responses in early August 2025, and in around 10 percent by mid-September, across 230,000 prompts and more than 100 million citations (Semrush, 2025). Wikipedia fell around the same time, from roughly 55 percent of responses to under 20.
Semrush does not give a confirmed cause. Its head of organic and AI visibility suspects ChatGPT was deliberately citing some sites less often, and says the popular theory, Google dropping its 100-results-per-page option, is not the whole story. Anyone telling you exactly why it happened is guessing. What it demonstrates is that source weightings rotate without warning, which is the whole argument for corroboration across many surfaces rather than depth on one.
Which profiles and signals to actually build
Three profiles carry most of the structured-data weight: Google Business Profile, Bing Places for Business, and Apple Business Connect. Structured profiles are the easiest thing for a machine to trust, which makes them the foundation everything else corroborates. Claim all three, make the name, address, phone and hours identical across them, and stop there before chasing anything cleverer.
What customers should actually say in reviews
The models read what reviews say, not just what they score. A review naming the service, the outcome and the neighborhood hands the model language it can repeat. "They re-piped a rowhome in Manayunk in two days and the quote held" is worth more than ten that say "great service." So the ask changes from "leave us five stars" to "mention what we did and where."
Then route by platform, because the review economies are genuinely different: Google feeds Gemini and AI Overviews, Facebook leads on Bing Places, Yelp dominates Perplexity, and Tripadvisor matters in food and travel. Keep them recent and reply to every one.
Editorial mentions
The best-of lists, local press roundups, chamber and association pages, and curated directories. One mention is a claim. Five consistent mentions across sites the model trusts is a fact. Corroboration is the currency, and it is the slowest thing on this list, which is why it is also the most defensible.
Make your site readable
Server-rendered HTML is the floor. Many modern templates render content with client-side JavaScript, which means the words a human sees are not in the raw HTML a crawler receives. View source and confirm your services and your city are actually in it. Here is what that looks like on this site.
<!-- trimmed --><title>Get Found on Google and in AI Answers | Pfender Marketing Co.</title><meta name="description" content="Joe Pfender helps local business owners and founders get found on Google and in AI answers, turn visits into calls and track every lead. Free site audit.">..."telephone":"+1-717-837-9265","address":{"@type":"PostalAddress","addressLocality":"Philadelphia","addressRegion":"PA","addressCountry":"US"} ...<h1 id="hero-title">Make your business easier to <span class="hero-turn"><em>find, understand, and choose.</em></span></h1><p class="hero-lead">I find where customers drop off between looking you up and calling you, then fix it: the website, the Google profile, reviews, follow-up and tracking. One person from diagnosis to done.Then add schema, and keep your business information identical to your profiles.
Do I need an llms.txt file?
No. Almost every AI visibility checklist tells you to add one. Ahrefs looked at 137,210 domains in May 2026, found 28 percent had published the file, and found that 97 percent of those files received no requests at all in the month (Ahrefs, 2026). Not zero AI traffic. Zero requests of any kind, human or bot. Nothing is fetching it. Put the hour into schema instead.
Can a tool put this on autopilot?
No, and the category is worth being precise about. Interest is real: Ahrefs' own keyword data shows searches for "ai search tracking" up 184 percent over the past year and "ai rank tracking" up 175 percent (Ahrefs, 2026). You will see figures ten times that quoted elsewhere. They do not come from anywhere.
The legitimate tools are measurement layers. They run prompts at scale and report how often you are named, and the better ones flag technical gaps. That is genuinely useful. What none of them can do is claim your Bing profile, earn a real best-of mention, turn a happy customer into a specific review, or make you a business worth naming. A dashboard watching the scoreboard is not playing the game.
And the guaranteed-placement version is the old "guaranteed page one on Google" pitch with a new coat of paint. I wrote about why you should run from guaranteed rankings, and it applies double here. The Reddit swing above is the proof: ChatGPT's citations of a single source went from about 60 percent of answers to about 10 in six weeks, and the people who measured it can only offer theories about why.
You cannot buy a position inside a machine you do not control. You can only become the answer it keeps finding.
How to measure it
Three layers. First, GA4 has had its own AI Assistant channel since May 2026, under Reports, Acquisition, Traffic acquisition. It picks up visits from the assistants Google recognizes by referrer. Many AI visits arrive with no referrer and land in Direct, so treat the number as a floor rather than a count. If you want to control the list yourself, build a custom channel group under Admin, Data display, Channel groups:
Channel name: AI Assistants
Condition: Session source matches regex
Regex:
chatgpt\.com|chat\.openai\.com|perplexity\.ai|claude\.ai|gemini\.google\.com|copilot\.microsoft\.com|grok\.com|meta\.ai|deepseek\.com|you\.com Second, and more useful than any tool for the first six months: a monthly prompt audit by hand. It costs nothing and tells you exactly who to study.
Third, if you want a continuous read, the tracker category exists and prices from free single-keyword tiers up into enterprise. For most local businesses, the manual audit plus GA4 is plenty to start.
What this actually comes down to
A local business is rarely invisible to AI because it missed one tactic. It is invisible because of fragmentation: an address that disagrees across a dozen tools, a website the models cannot read, reviews scattered and unanswered, no schema, a Bing profile nobody claimed. The machines meet a business that cannot describe itself consistently, and they move on.
That same fragmentation is usually why the software bill is too high, because every disconnected tool holds its own slightly different version of the truth and you pay for all of them to disagree. The two problems are the same problem, and that is the work I do: one audit covering where you are invisible and where you are paying twice, then a single system that addresses both.
Common questions
Does ChatGPT use Google to find local businesses?
OpenAI does not name Google. Its main search help page mentions third-party search providers and points to Microsoft's privacy statement, its Enterprise and Edu page names Bing, and it runs its own search crawler. Anyone stating which index answers your particular question is inferring. The practical answer is unchanged: be findable and consistent across the open web rather than optimizing for one presumed index.
Which assistant should I work on first?
None of them, individually. Most of the work overlaps: consistent profiles, a readable site, detailed recent reviews, and editorial mentions raise you across all of them at once. If you want a single starting point, it is your Google Business Profile, because Gemini and AI Overviews read it directly and it is also the surface that recommends local businesses most often.
Do I need an llms.txt file?
No. Ahrefs found 97 percent of published llms.txt files received no requests at all in May 2026. Spend the time on schema, server-rendered content, and consistent business information.
Can an agency guarantee my business will be recommended?
No, and the guarantee is the tell. Nobody controls model output, source weightings rotate without warning, and the honest version of this work is fundamentals plus measurement.
How long before assistants start recommending my business?
Weeks to months rather than days, and nobody can give you a date. The assistants have to recrawl your updated profiles and site, then find you corroborated across enough independent sources to clear their confidence bar. Recent reviews and fresh content help.