Local businesses in AI search get recommended when their sources agree. When someone asks ChatGPT, Gemini or Google's AI Mode for "the best dentist near me", the assistant leans on map listings, review text, directories and your own website to decide who to name. Make those sources agree about what you do, where you do it and why customers rate you, then keep them fresh.
This playbook is for owners of physical or service-area businesses: clinics, studios, trades, salons, consultancies, restaurants. It skips theory (our guide to AEO and GEO covers that) and focuses on the jobs you can do this month.
Where do AI assistants get local recommendations from?
They mostly reuse the same local data that powers maps and search, then summarise it in words. An assistant rarely "knows" your business from training alone; for a near-me question it usually looks things up at the moment it answers.
The common inputs are:
- Map and listing data. Google Business Profile feeds Google's AI Overviews and AI Mode. Apple Business Connect feeds Apple Maps and Siri. Other assistants draw on whichever map and listing providers they partner with.
- Review text, not just star ratings. A review that says "they fixed our leaking roof the same afternoon in Parramatta" tells a model your service, speed and location in one sentence.
- Your website. Service pages, location details, opening hours, prices and FAQs that a crawler can read.
- Third-party mentions. Local directories, industry associations, news coverage, "best of" roundups and community forums.
None of these platforms publish a full recipe, so treat any claim of a guaranteed ranking factor with suspicion. What you can control is consistency and specificity across all four sources.
What makes an assistant name one business over another?
Assistants favour businesses whose details are clear, consistent and backed by specific evidence from customers. When signals conflict (different phone numbers, vague service lists, old hours), a model has less reason to trust you and an easy alternative in your competitor.
| Signal | What the AI reads | What to do this month |
|---|---|---|
| Business category | Primary and secondary categories on your listings | Choose the most precise primary category; add secondary ones only if you truly offer them |
| Name, address, phone | Whether details match across listings and your site | Audit the top 10 places you're listed and fix every mismatch |
| Services | Service lists on your profile and dedicated pages on your site | One page per core service, with plain descriptions and typical price ranges |
| Review content | Recency, volume and the words customers use | Ask happy customers to mention the service and suburb in their own words |
| Opening hours and availability | Hours, holiday hours, booking links | Update holiday hours before every public holiday |
| Proof of expertise | Credentials, licences, photos of real work | Add licence numbers, team bios and project photos with captions |
| Third-party mentions | Directories, associations, local press | Get listed where your industry body or city chamber publishes members |
The 30-day playbook
Work through listings first, then your website, then reviews and mentions, because each layer depends on the one before.
| Week | Focus | Concrete tasks |
|---|---|---|
| 1 | Listings | Claim or verify Google Business Profile, Apple Business Connect and Bing Places. Match name, address, phone and hours exactly. Pick precise categories. |
| 2 | Website | Publish or tighten one page per service. Add an address block, a map embed, hours and LocalBusiness structured data. |
| 3 | Reviews | Send a short review request to recent customers with a direct link. Reply to every review from the past year, naming the service where natural. |
| 4 | Mentions and testing | Join two relevant directories or associations. Run your test prompts (see below) and record which businesses are named. |
Our schema markup guide for AI search shows how to write the LocalBusiness and Organization markup for week 2. If you serve customers at their location rather than a shopfront, read our explainer on Google Business Profile for service-area businesses before you touch your address settings.
Why do thin city pages hurt rather than help?
Mass-produced "plumber in [suburb]" pages with swapped place names are a liability, not a shortcut. Google's spam policies treat pages made at scale mainly to rank, with little unique value, as scaled content abuse. AI systems that pick sources for quality have no reason to quote a page that says nothing new.
We learned this on our own site. grww.tech once carried 167 templated city pages. They looked like coverage but offered nothing a visitor couldn't get from the main service page, so we set all 167 to noindex to remove the scaled-content risk and pointed effort at fewer, stronger pages.
A location page earns its place when it contains something specific to that place:
- Projects completed there, with photos and a sentence about each.
- Local conditions that change the service (council approval steps, climate, building types).
- Who from your team covers that area and how quickly they can arrive.
- Reviews from customers in that location.
If you can't write at least a few genuinely local paragraphs, skip the page and mention the area on your main service page instead.
How do reviews influence AI recommendations?
Reviews matter as evidence: an assistant can only call you "great for emergency call-outs" if customers have said so somewhere it can read. Star averages separate good from poor businesses; review wording decides what you're recommended for.
Practical rules:
- Ask at the moment of delight. Right after delivery, handover or a successful appointment, not weeks later.
- Prompt for detail without scripting. "Would you mention which service we did and how it went?" is fine. Writing the review for them, offering discounts for five stars or gating requests to happy customers breaks platform rules.
- Reply with substance. Thank the reviewer and restate the service and area naturally. That adds readable context and shows prospects you're responsive.
- Handle negatives in public, briefly. Acknowledge, offer to resolve offline, then follow through. A calm reply often reads better than a perfect score.
For a full system to keep reviews coming in, see how to get more Google reviews.
How do you check whether AI assistants recommend you?
Run the same set of prompts each month across three or four assistants and log who gets named. It's manual, but it's the only honest measurement available to a small business today.
Build a prompt list of 8–12 queries that real customers would type, mixing styles:
- "Best [service] near [suburb]"
- "Who should I call for [specific problem] in [city]?"
- "Affordable [service] in [city] with good reviews"
- "[Service] in [city] open on Sundays"
Ask each one in ChatGPT, Gemini or Google AI Mode, Perplexity and Copilot, ideally logged out or in a fresh session, and note the businesses mentioned, the reasons given and any sources cited. Results vary by location and session, so look for trends over several months rather than reacting to a single answer. Pay attention to the wording too: the reasons an assistant gives tell you which evidence it found.
When an assistant gives a reason for picking a competitor ("known for same-day service"), that's a gap you can close with real evidence: a service page, a policy, reviews that mention it.
What if you're a service-area business with no shopfront?
You can still be recommended; your website and reviews simply carry more weight because there's no storefront pin to anchor you. Hide your home address if customers don't visit, set realistic service areas, and make your site explicit about the suburbs or regions you cover and how quickly you can attend.
Help models connect your business name to your trade and region with the same description everywhere: website "About" section, social bios, directories. Consistency is what lets a model treat all those mentions as one business.
What should you avoid?
Avoid anything that creates conflicting or fake signals; it can undo months of honest work.
- Keyword-stuffing your business name ("Best Plumber Sydney Cheap 24/7") on listings, which breaks Google Business Profile guidelines.
- Buying reviews or swapping them with other businesses.
- Listing services you don't offer to appear in more answers.
- Leaving old phone numbers or addresses live on abandoned directories.
- Publishing AI-written location pages with no local substance.
Frequently asked questions
Can I pay to be recommended by ChatGPT or Gemini?
Not in the organic answers. Some platforms are testing or running ads around AI answers, but the recommendations themselves draw on public information. The reliable route is to make your listings, site and reviews clear and consistent.
How long before changes show up in AI answers?
Listing updates can appear within days, while new reviews and website pages may take weeks to be crawled and reflected. Assistants also refresh at different speeds. Track monthly and expect a gradual shift rather than an overnight jump.
Do I need a separate page for every suburb I serve?
No. Create location pages only where you have genuinely local content such as projects, reviews or local conditions. Otherwise, list your service areas clearly on your main service pages and in your Business Profile.
Do star ratings matter, or only review text?
Both. Ratings help separate well-regarded businesses from poor ones, while the wording tells an assistant what you're good at. A steady flow of detailed, recent reviews beats a high score built on a handful of old ones.
Is local SEO still worth doing if people use AI assistants?
Yes, because most of the same work feeds both. Accurate listings, strong service pages and reviews help you in map results, classic search and AI answers at once. The fundamentals haven't changed; they simply matter in more places.
Make your business the obvious answer
Being named in a "best near me" answer comes down to clear, consistent facts and real customer evidence spread across the places assistants read. If you'd like help with service pages, structured data or a listings audit, see our services or ask for a fixed-price quote.