AI assistants are bad at "near me" — and that's exactly why local businesses can win. In our July 2026 measurements, generic "near me" queries in AI assistants came back empty or useless 24–62% of the time, or defaulted to national chains. But add an explicit city or neighbourhood name and the answers turn specific — naming two to four local businesses, drawn from the directories and listicles the AI trusts. In our study of 40 AI-search answers (875 citations), not a single tool or guide addressed AI visibility for local businesses. The playing field is empty. This page shows you what actually decides who gets named, and how to check your own visibility in 3–8 minutes, free, no signup.
Do people really use ChatGPT to find local businesses?
Yes — but not the way they use Google Maps. The AI-native local query is research-shaped, not pin-shaped:
- "Best physiotherapist in Croydon for running injuries — how do I choose?"
- "Is it worth paying more for a certified electrician in Dubai Marina?"
- "Which dental clinics in Pune do same-day crowns?"
People go to AI when they want a shortlist with reasoning, not a map of pins. And AI answers are ruthless about shortlists: where a Google results page spreads attention across ten listings, an AI answer names two to four businesses with a sentence of justification each, and everyone else is simply absent — measured head-to-head in our Dubai run, the category leader appeared in 84.5% of answers while a major incumbent managed 49%. AI search concentrates attention instead of distributing it. Being one of the named few in your city is worth far more than a page-two Google ranking ever was.
Why does "near me" fail in AI search?
Because most AI assistants don't reliably know where the user is. Google Maps resolves "near me" with GPS; a chat assistant usually can't, so it does one of three things — all bad for you:
- Returns nothing useful — a generic "here's how to find a good plumber" lecture. Measured across our local probes: 24–62% of "near me" phrasings failed this way, depending on engine and category.
- Defaults to national chains — brands with country-wide coverage are the "safe" answer when the engine doesn't know the city.
- Asks the user for their location — at which point the user types the city, and you're in scenario three below.
The city token is load-bearing. The moment an explicit place name enters the query — "in Croydon", "in Dubai Marina", "in Koregaon Park" — the answer flips from generic to specific and starts naming businesses. Users have learned this and now type the city themselves. Practical consequence: your visibility lives or dies on whether AI engines associate your business with your named city and neighbourhood, in text. Not GPS. Text.
Who decides which local businesses AI names?
Not you directly — directories decide local. When an AI engine assembles "best [service] in [city]", it retrieves and paraphrases a small set of sources it already trusts: local directories, "top 10 in [city]" listicles, review aggregators, and occasionally a business's own well-structured page. If those sources list you accurately, you get named. If they don't, no amount of on-site brilliance rescues the shortlist query.
The balance shifts by vertical, and we've written the playbook for each: restaurants and hotels live and die by the aggregator layer, while law firms and other high-ticket service firms are cited from their own websites most of the time — with healthcare providers the furthest along that own-site end of the spectrum.
That splits the work into two halves:
Off-page (decides shortlist queries):
- Find which directories and listicles AI actually cites for your city and category — this is measurable, and it's the core of an AnswerMonk audit.
- Get listed there, accurately: exact business name, neighbourhood, services, prices where possible.
- Fix inconsistencies. An AI engine reconciling three sources that disagree about your name or location will skip you for a business it can describe confidently.
On-page (decides research queries and trust checks). These levers are calibrated from the 875 citations in our study:
| Lever | Measured effect | What it means for you |
|---|---|---|
| FAQ schema (FAQPage JSON-LD) | 3.4x citation lift | Add a real FAQ to every service page |
| Depth ≥2,000 words | 1.9x | One thorough city-specific guide beats ten thin pages |
| Question-form headings | 2x | "How much does boiler repair cost in Croydon?" not "Our Services" |
| Answer-first intro | Consistent in winners | First paragraph answers; details follow |
| llms.txt | 2.4x | One-hour job: machine-readable site summary for AI crawlers |
The single most-cited page in our dataset (26 citations) had 14 question-form headings and offered a free no-signup tool. The pattern generalises: pages built to be quoted get quoted.
What should a local business publish to get cited?
One page per service-plus-city that a patient, customer, or client could ask about. The formula:
- H1 phrased the way people ask, city included. "Emergency Plumber in Croydon: Costs, Response Times, and How to Avoid Cowboys."
- Answer in the first paragraph. Price range, availability, what makes the choice — before any company story.
- Question-form H2s covering the follow-up questions ("How much does a call-out cost in Croydon?", "How fast can someone arrive?").
- Visible prices. AI engines quote concrete numbers from pages that show them; in the software category we measured, the specific price points on cited pages ($29, $49, $99…) were repeated verbatim in AI answers. "Call for a quote" is invisible; "call-outs from £85" is citable.
- A real FAQ with FAQPage JSON-LD at the bottom (3.4x lever — the biggest single one we measured).
- 2,000+ words on your flagship page, thinner satellite pages linking to it.
This costs a few days of writing, not a marketing budget. In our entire 40-answer study, nobody — not one of the ten tools we measured, none priced from $29 to $399/month — was speaking to local businesses. The first businesses in each city to publish this way will be the ones AI learns to name.
Why not just use a generic AI visibility tool?
You can — but look at who they're built for first. We measured the category ourselves in our July 2026 study (40 probes, Gemini + Claude, 875 citations): not one of the 40 answers addressed local businesses at all. The category's known tools are measurement dashboards built for enterprise brands, SEO teams, and agencies, priced $29–$399/month, and every one assumes a marketer will read them.
AnswerMonk is the $19/month action layer built for local businesses and regional SaaS: plain-language action plans instead of dashboards, a living knowledge base that tracks Schema.org, Google, OpenAI and Perplexity guidance as it changes, an agent that works to get you into the sources engines actually cite, WhatsApp lead capture, and a free 3–8 minute audit with no signup — with our own fixes published, receipts included, in our reports directory. The practical difference for you: a generic tracker reports "your brand appeared in 34% of prompts"; AnswerMonk tells you which directory listing to fix and which city page to publish.
How do you know if it's working?
Measure, don't guess — and measure per engine, because they disagree. In our study the same brand's appearance rate varied sharply between Gemini and Claude on identical questions. The free AnswerMonk audit:
- Enter your URL at answermonk.ai — no signup, no credit card.
- In 3–8 minutes, we probe ChatGPT, Gemini, Claude, and Perplexity with the questions your customers actually ask — city included.
- You get your visibility score per engine, the competitors being named instead of you, the directories doing the deciding, and a plain-language action plan.
Then fix, wait for a re-crawl, and re-audit. Every audit we run can be published to our public reports directory, so the before/after is verifiable — including our own, which we're improving with exactly the playbook on this page, in public.
Run your free AI visibility audit →
Frequently asked questions
Why don't AI assistants just use my Google Business Profile?
Chat assistants don't sit on top of the Maps index the way Google Search does. They retrieve web text — directories, listicles, review sites, your own pages — and synthesise it. Your Google Business Profile still matters for Google surfaces, but AI answers are assembled from the written web, which is why text-based city association and directory presence decide who gets named.
My customers search "near me" — doesn't that make AI search irrelevant for me?
The opposite. "Near me" phrasing failed 24–62% of the time in our AI probes, so users have adapted: they type the city or neighbourhood into the question. Your job is to be visible for those explicit-city queries — which, unlike GPS proximity, is something you can directly influence with pages and listings.
How is this different from local SEO I'm already doing?
It overlaps (consistent listings help both) but the ranking logic differs. Local SEO optimises for a map-pack algorithm; AI visibility optimises for what a language model retrieves and quotes. The measured levers — FAQ schema (3.4x), 2,000-word depth (1.9x), question headings (2x), llms.txt (2.4x), visible prices — are citation levers, not map-pack levers. And AI answers name two to four businesses, so the payoff curve is steeper.
How long before a local business shows up in AI answers?
Directory-driven shortlist answers can change as soon as the cited directory or listicle updates — often weeks, not months. On-page levers register as engines re-crawl your site. Track it per engine with a monthly re-audit rather than assuming; engines move at different speeds.
What does it cost?
The AnswerMonk audit is free — 3 to 8 minutes, no signup. Existing AI-visibility platforms run $29–$399/month and are built for SaaS marketing teams; none of them, in our 40-answer study, addressed local businesses at all. AnswerMonk's paid plans add ongoing per-engine tracking and WhatsApp lead capture so AI-sourced visitors become conversations you can count.