Local AEO: what AI answers change for a local business

Some constraints have a field; some need page text.

What is local AEO?

Local AEO is answer engine optimisation applied to a business with a physical location or service area. Answer engine optimisation means preparing your information so an AI surface can use it as the answer. The concept itself sits in what AEO is and our guide to AI search. This page asks one narrower question: what changes for a local business now that AI surfaces answer local questions?

Three definitions, because the rest depends on them. AI Overviews are AI-written summary blocks inside a normal Google results page. AI Mode is a separate conversational surface you open yourself. The local pack is the three-business map block built from Business Profile data.

The honest headline: most of what you read about local AI search is inference dressed as documentation. So we have separated the two here, visibly.

What has Google actually published about local results in AI answers?

Less than the industry implies. Google has published no document explaining how local results are chosen inside AI Overviews or AI Mode, no statement that reviews feed AI answers, and no page documenting the local pack and AI surfaces as connected systems. What exists is a handful of sentences scattered across unrelated help pages. Here is every relevant one we could find.

That is the documented set. Nothing in it says reviews, Q&A, photos or posts are inputs to an AI answer, and we found no primary Google source for it. If someone says reviews feed AI Overviews, ask which Google page says so.

Anything below marked implication is our reasoning, not Google's statement.

How does a local AI answer differ from a local pack result?

A local pack result is a ranked list. An AI answer is a sentence. That is not cosmetic: a sentence carries conditions a list cannot — "open until 9pm", "wheelchair accessible", "no appointment needed" — and those come from structured fields, not your prose.

Click and position wording from "Impressions, position and clicks"; report scope from the generative AI performance report. Undated, accessed October 2026.

One consequence. AI clicks exist and are counted — they just sit inside the ordinary Performance report, mixed in with everything else. The report that isolates AI surfaces has no click column. So you measure impressions and say so plainly. Our note on measuring LLM visibility has the method.

The multi-constraint shift: one condition became four

This is the part that genuinely changes. "Dentist near me" carries one constraint plus a location. "Dentist near me open Sunday that takes walk-ins and has step-free access" carries four. Each condition narrows the field, and the business with all four answered is the one that can be named.

Google's own agentic booking example names the constraint set it works with: party size, date, time, location and preferred cuisine. Five conditions in one request. That feature is a US Labs experiment, not an Indian reality — but it shows what the surface is being built to do.

Implication. Classic "near me" SEO optimised for a short head query. Local AEO optimises for a long, conditional one — and most of that sentence is answered by fields, not content.

Three hypothetical Indian examples:

• A dentist in Pune. "Open Sunday" comes from hours. "Step-free entrance" from an accessibility attribute. "Takes card" from a payments attribute. Miss one field, miss the question.
• A salon in Thane during Diwali. Only special hours answer this. Regular weekly hours will say you are open on a day you are shut. Google names "special store hours" as information to provide.
• A driving school in Nashik. "Instruction in Marathi" has no field in Google's published attribute list. That constraint can only be answered in words, on your own page.

That last one is why this is not a profile job alone.

Which local inputs do you actually control?

Split them by where the answer lives. Some constraints have a structured field in your Business Profile. Some have no field and can only be answered in text on your site. Most businesses do only the first.

We are not re-teaching profile setup — our Business Profile optimisation guide and the local SEO checklist own that. The point is the sorting, not the steps. Worth knowing: Google now lets you manage hours, special hours, attributes, action links and review replies through the Gemini web app, for single verified profiles outside the EEA and UK.

What does not change about local search?

Most of it. Google's local ranking guidance still says results run on relevance, distance and prominence, and that businesses with complete, accurate information are more likely to appear. None of it has been rewritten for AI surfaces.

• Proximity still dominates. Distance is one of three named factors. No content beats being closer. See why a business is not showing on Maps if yours has vanished.
• NAP consistency still matters. Name, address and phone must match everywhere — see citations and NAP consistency.
• Reviews still matter — Google ties prominence partly to review count. What is not documented is reviews feeding AI answers. Keep collecting reviews for the documented reason.
• Maps ranking is its own discipline. Nothing here replaces ranking in Google Maps, where most local clicks still land.
• No special markup is required. Google states there is "no special schema.org structured data that you need to add" and no need for "new machine readable files, AI text files, or markup". Keep schema markup for rich results; structured data for AI draws the line.

What we do not know about local AEO

This is the most useful part of the page, because the gaps are large and most vendors fill them with unearned confidence.

• Whether Business Profile fields are used inside AI answers. Google says keep them current. It does not say how, or whether, they are read.
• Whether reviews, Q&A, photos or posts are inputs. No source found.
• How proximity works in an AI answer. AI Mode's help page does not mention location handling at all.
• Whether the local pack and AI surfaces share a selection system. Not documented.
• How any of it performs in India. Agentic booking and AI calling are US-only today.
• What an AI local impression is worth. Impressions without clicks cannot be valued.

Anyone presenting these as settled is guessing — us included.

What to do about local AEO this month

Short list, and nothing on it depends on an undocumented claim. Every item holds up even if AI surfaces never read a single field.

1. Fill every empty profile field, special hours first, before the next festival. Documented as helping local results, and named in Google's AI-features best practices.
2. Answer the constraints that have no field on your location and service pages — languages, walk-in policy, who you serve. Plain sentences, near the top.
3. Check snippet eligibility. A nosnippet or tight max-snippet tag makes a page ineligible as a supporting link — the one hard gate Google names.
4. Make each location page self-contained. A section that only makes sense after the one above it cannot be lifted. Same discipline as getting cited in AI Overviews.
5. Get named off your own site too — local directories, press, community pages. Our guide to brand mentions explains why.
6. Report impressions, not invented AI clicks. Set that expectation before someone sets a wrong one.

Also answer your phone. Google's AI already calls local businesses for quotes in the US. At some point, what decides whether you are in the answer is a human picking up.

Where this leaves a local programme

Local AEO is not a new discipline bolted onto local SEO. It is the same discipline with a higher penalty for incompleteness. The map pack forgave an empty attribute; a four-condition question does not.

The work shifts from ranking a profile to describing a business exhaustively — in fields where fields exist, in sentences where they do not. It is cheap, it is yours to control, and it pays off in AI Overviews, in Maps and in hyperlocal campaigns at once. When Google finally publishes something local-specific, the businesses that already filled everything in will not need to do a thing.