Ask an AI assistant for a good plumber in your suburb and you get three names. Not a map with forty pins and a filter panel. Three names, sometimes with a sentence explaining why each one.
For local businesses that is a harsher shortlist than the search results page ever was. Ranking eighth in the map pack still put you on the screen. Being eighth in an AI model's assessment puts you nowhere.
Local AI recommendations are decided mostly by whether your business facts agree everywhere they appear. Yext found 42% of AI citations come from business listings. Fix listing consistency and duplicate profiles first, get recent specific reviews second, then make your service and location pages answer questions directly. Content volume matters least here.
How do AI assistants decide which local businesses to recommend?
They assemble an answer from sources that describe local businesses, then name the few they can describe most confidently.
That confidence is the whole game. A model naming a business is making an implicit endorsement, and it will reach for businesses whose category, location and service details are corroborated across several independent sources. A business whose facts contradict each other across the web is a risk, and there is always another option whose facts do not.
This is why local AI visibility often does not match local search rankings. Ranking rewards relevance and proximity. Recommendation rewards clarity and corroboration. A business can be strong on the first and weak on the second.
Yext's study of 6.8 million AI citations found 86% came from brand-managed sources, with 42% from business listings and 44% from first-party websites (Yext, Oct 2025). For local businesses that figure is the strategy: the listings layer is close to half the opportunity and is almost entirely administrative work.
Bain & Company found 80% of consumers rely on AI-generated results for at least 40% of their searches (Bain, Feb 2025).
Local AEO versus local SEO
They overlap heavily, and the differences are worth being precise about because they change what you prioritise.
| Local SEO | Local AEO | |
|---|---|---|
| Goal | Appear in map pack and local results | Be named in a spoken or written recommendation |
| Typical output | A list of businesses with filters | Two to five names, sometimes with reasons |
| Proximity weighting | Very strong | Present but weaker |
| Decisive factor | Relevance, distance, prominence | Consistency and corroboration |
| Reviews | Count and rating matter | Recency and specificity matter more |
| Failure mode | Ranked low but visible | Not mentioned at all |
The practical consequence: local SEO work is largely transferable, but the ordering changes. Listing consistency moves from housekeeping to the single highest-value task, and review specificity matters more than review count.
Step 1: Make your business facts agree everywhere
This is unglamorous, it is most of the result, and it is the step people skip because it does not feel like marketing.
Find every profile that exists. Search your business name, your phone number and your address separately. You are looking for profiles you forgot about, ones created by former staff, and auto-generated entries from data aggregators. Most established local businesses have several they did not know about.
Kill the duplicates. A live profile carrying a previous address is actively harmful, because it gives a model two contradictory answers about where you operate. Removing or merging duplicates is often the highest-value hour in a local AEO project.
Reconcile every field. Business name, address, phone, category, service list, hours and service area should match your website exactly. Not approximately, exactly. Choose the same primary category on every platform that offers one.
Complete the profiles you keep. Empty fields give a model less to work with. Service descriptions, attributes, hours and photos all contribute to a fuller picture.
Your Google Business Profile, fully completed and accurate, with the same primary category and service descriptions as your website. It is the most widely referenced local data source, and getting it right also fixes a large share of conventional local search performance at the same time.
Step 2: Build reviews that are useful to a model
A star rating is a summary. What actually helps a model recommend you is review text that names services, locations and outcomes.
"Great service, highly recommend" tells a model nothing about what you do. "They replaced our hot water system in Brunswick the same day we called" tells it your service, your area and your responsiveness in one sentence.
The way to get the second kind is to ask for it specifically. A review request that asks which service someone used and what problem it solved produces materially more useful text than a generic request for feedback.
Recency matters too. A strong rating built entirely from reviews three years old says little about current service, and models weight recent evidence more heavily for exactly that reason.
Do not buy reviews, write them yourself, or ask only satisfied customers while filtering out everyone else. Platform policies prohibit it, penalties are real, and a review corpus that reads as manufactured undermines the corroboration you are trying to build. Selective gating is a policy violation on most major platforms even when every individual review is genuine.
Step 3: Make your site answer local questions directly
Once your facts are consistent and your reviews are useful, your website needs to give a model something clean to extract.
Answer first on every service page. Open with what the service is, where you deliver it and who it is for. Positioning language reads well to humans and gives a model nothing to quote.
Name your service areas explicitly. If you serve specific suburbs, regions or a radius, say so in plain text. A model cannot infer your service area from a map embed.
Give each core service its own page. A single page listing eleven services is harder to resolve than distinct pages that each answer one question thoroughly.
Add accurate LocalBusiness schema. Standard structured data matching the visible page, with the same facts as your listings. No special AI schema is required, and Google has stated as much. Our technical guide covers what holds up and what does not.
Publish the practical details. Hours, parking, accessibility, payment methods, response times, whether you offer callouts. These are the specifics that make a recommendation useful and give a model reasons to prefer you.
Step 4: Get corroborated locally
Independent local sources confirming what you claim are what separates a business a model will name from one it will not.
Local directories and chambers of commerce are low effort and specific to your area. Trade and professional registers matter more than general directories in licensed categories, and they carry credential verification a model can rely on. Local media coverage, community sponsorships and supplier or partner pages all create mentions on domains you do not control, which is precisely why they count.
None of this needs to be a campaign. For most local businesses it is a handful of genuine relationships that already exist and are simply not documented anywhere online. Our guide to earning AI citations covers the broader off-site approach.
What this looks like in practice
A trades business ranking well locally but never named by ChatGPT usually has the same three problems.
Two live profiles, one with the address from before they moved. Forty reviews averaging 4.8, of which thirty-five say some version of "great job, very professional" and none name a service or suburb. And a website whose services page lists eleven offerings in a bulleted grid with no page for any of them.
Nothing there is a content volume problem, and publishing weekly blog posts would not fix any of it. The sequence that works is: remove the stale profile, reconcile the remaining one against the site, start asking for specific reviews, then build out individual service pages with the service area named in plain text.
That is usually a few weeks of unglamorous work rather than a campaign, which is the point.
If you decide to hire rather than run this in-house, our comparison of the best AEO, GEO and AI SEO agencies scores 15 firms against a published rubric.
Measuring local AI visibility
Run the questions your customers would actually ask, not the ones you would like to rank for. "Best electrician in Fitzroy", "emergency plumber near me open now", "who repairs commercial coffee machines in Brisbane".
Record whether you are named, how you are described, and which sources the engine cites. Repeat the runs, because a model can answer the same question differently between sessions and a single check tells you almost nothing.
Watch for the two distinct failure modes, because they need different fixes. Not being named at all points at corroboration and consistency. Being named but described wrongly points at a specific source carrying bad data, which you can usually trace and correct.
Our guide to checking whether your business appears in AI search walks through building that baseline, and the metrics framework covers prompt sets and repeat-run cadence. If you are weighing whether the investment is justified at your size, is AEO worth it for small businesses works through the economics.
Want to know if AI assistants recommend your business locally?
Book a free AI Visibility Audit. We will run the local questions your customers ask, show you who gets named instead of you, and explain exactly what is causing it.
Book Your AI Visibility AuditFrequently Asked Questions
Does my Google Business Profile affect AI recommendations?
It is one of the most widely referenced sources of local business data, so keeping it complete and accurate matters for AI recommendation as well as conventional local search. What it is not is a direct control panel for AI answers. Models draw on many sources, and a perfect profile contradicted by three stale directory entries still leaves you ambiguous.
How many reviews do I need to get recommended by AI?
There is no threshold, and anyone quoting one is guessing. What is more useful than a target number is the pattern: recent rather than historic, specific rather than generic, and spread across more than one platform. Twenty recent reviews naming actual services and suburbs are more useful to a model than a hundred old generic ones.
Do AI assistants use my location the way Google Maps does?
Less directly. Proximity is a strong ranking factor in map results, while AI recommendations weight it alongside clarity and corroboration. That is genuinely good news for businesses slightly outside a dense centre: being the most clearly described specialist in your category can outweigh being the closest option, which is rarely true in a map pack.
Should local businesses target ChatGPT, Gemini, or both?
Both, because your customers are not standardised. ChatGPT reported 800 million weekly active users (Sam Altman via TechCrunch, Oct 2025) and Gemini passed 750 million monthly active users (Google, Feb 2026). The reassuring part is that the underlying work is the same for both: consistent facts, useful reviews, clear service pages and local corroboration. You are not running two separate programmes.
How long does local AEO take to work?
Faster than most AEO work, because listing platforms are crawled frequently and the fixes are concrete. Removing a duplicate profile and reconciling your details can show up within weeks. Review patterns take a couple of months to shift meaningfully since they depend on customer volume. Local corroboration is slowest, but it is also the part competitors cannot simply buy past you.
Ashur Homa
Built and scaled a digital brand to $100M+ in sales with zero ad spend. Has helped businesses generate millions through AI go-to-market strategy. Leads growth at Omni Eclipse.
Connect on LinkedIn
