What is GEO for a local business — and does yours actually need it?

Some time in the last year a new line item started appearing on marketing proposals: GEO, for Generative Engine Optimization. Depending on who is selling it, it is either the most urgent thing you are not doing or a rebranding of work you already pay for.

Both descriptions are partly true, which is why this is confusing. Here is the plain version — what the term means, what the work actually consists of, and a test for whether your business needs to care right now or can reasonably wait.

What GEO means, minus the jargon

Generative engine optimization is the practice of improving your chances of being named or cited when someone asks an AI assistant a question your business could answer. Search engine optimization aims at a page of ten blue links. GEO aims at a paragraph that names two or three businesses and stops.

The name is worse than the idea. “Generative engine” is a phrase nobody outside the industry uses, and the acronym competes with a dozen others — you will also see AEO (answer engine optimization) and AI SEO describing roughly the same thing.

More importantly: for a local business this is not a separate discipline from local SEO. Whitespark’s 2026 Local Search Ranking Factors survey asked 47 local search experts to score 187 factors, and added AI visibility as a formal category for the first time. What emerged was convergence, not divergence. The signals that get you into Google’s local pack are largely the signals that get you into an AI answer: a complete Google Business Profile, consistent business information across platforms, real reviews, and a site that says plainly what you do and where.

Whitespark, 2026 Local Search Ranking Factors — 47 contributors, 187 factors scored 0–5. The category weights are published as a chart rather than as text, so no percentages are quoted here. Whitespark notes the AI Search column was “a bit ambiguous” to score, since the platforms have different levels of data access.

So if you have been doing local SEO properly, you are not starting from zero. What is genuinely new is not the work. It is the measurement — you can no longer assume that ranking in Google means appearing in the answer, because a business can hold the local pack and be entirely absent from AI recommendations for the same query.

What actually happens when someone asks ChatGPT for a dentist

This determines what you can influence, so it is worth being concrete.

The assistant does not have a list of local businesses memorized. When someone types “best family dentist in Scottsdale”, it runs a search behind the scenes, reads a handful of pages, and writes a short answer from what it found. Those pages are usually a mix of directory listings, review sites, “best of” roundups, local publications, and business websites.

Three consequences follow, and they are the whole game:

1. You cannot optimize the model. You can only influence what it reads. The assistant is summarising sources. If your name, address, hours and specialties are accurate and present in the sources it reaches for, you are a candidate. If they are not, you are not.
2. The answer is short. One to three names, typically. There is no page two.
3. The answer is unstable. SparkToro and Gumshoe.ai had 600 volunteers run 12 brand-recommendation prompts a combined 2,961 times, and found less than a 1-in-100 chance that ChatGPT or Google’s AI would return the same list of brands on any two runs of the same prompt. A single screenshot proves almost nothing about where you stand — yours or a competitor’s.

SparkToro / Gumshoe.ai — across ChatGPT, Claude, and Google’s AI search. For the same list in the same order, it was closer to 1 in 1,000.

Why this became a real question in 2026

It is worth separating what is genuinely happening from the urgency being manufactured around it.

The shift is real. BrightLocal’s 2026 Local Consumer Review Survey found the share of consumers who had used an AI tool to find a local business went from 6% to 45% in a single year. That makes AI the third most-used local discovery channel, behind Google and Facebook and ahead of Yelp and TripAdvisor. A separate Yext survey in March 2026 found 47% of US adults had used AI for a local search in the previous month. Adoption skews toward 30–44 year olds, which for most home services and healthcare practices is the buying demographic.

BrightLocal, Local Consumer Review Survey 2026 — 1,002 US adults, published March 2026 — Yext, 2026 Consumer Search Behaviors Report — 1,120 US adults, March 2026. The same BrightLocal survey found the use of Google reviews slipping from 83% to 71% year over year. That is review-reading behaviour specifically, not Google’s overall share of local discovery — a distinction worth keeping, because the looser version of this stat gets quoted a lot.

Set against that, one number people quote too casually. SOCi’s 2026 Local Visibility Index found ChatGPT recommended only 1.2% of the roughly 350,000 business locations it studied, against 35.9% appearing in Google’s local pack. It is a striking gap, but that study looked at multi-location brands, so 1.2% is not a solo practice’s odds. Treat it as a direction, not a forecast for your business.

SOCi, 2026 Local Visibility Index — 2,751 brands, roughly 350,000 locations — via Search Engine Land. Perplexity recommended 7.4% of locations and Gemini 11.0%. A single-office practice is not in that dataset.

The number worth anchoring on is smaller and older. BrightLocal found that 72% of consumers look at three or fewer businesses before deciding, and only 1% look at more than ten. That was true before AI assistants existed. What AI does is enforce it — it hands over a three-name shortlist instead of a page you could scroll past.

BrightLocal, Consumer Search Behavior — 1,227 US consumers who had searched for a local business in the previous three months. The pattern held across every channel they measured, including Google, review sites, social media and AI tools.

So does your business need it?

Four questions, in order.

1. Do your customers research before they contact you?

If people choose you by walking past, or by whoever is closest in the moment, AI recommendations are a marginal channel. If they research — read reviews, compare two or three names, check credentials — then you are in scope, because that research increasingly starts with an AI answer instead of a results page.

Dentists, lawyers, med spas, specialists, roofers, HVAC replacement, remodellers: high consideration, in scope. Convenience retail, walk-in food, businesses that live on proximity: much less so, at least for now.

2. Is a customer worth enough to justify the work?

This work costs either money or hours. If a new customer is worth a few hundred dollars in lifetime value, the arithmetic is tight and you should prioritise elsewhere. If a new patient, client or job is worth thousands, then appearing in a three-name answer instead of being absent from it is worth real effort.

3. Is your market contested?

In a dense metro with a dozen established competitors, three slots is a hard fight and foundations take months to compound. In a smaller market you may already be the answer and simply not know it — in which case your job is monitoring, not construction. These are genuinely different situations needing different amounts of investment, and you cannot tell which one you are in without checking.

4. Are your fundamentals already in place?

This is the question that disqualifies more businesses than the other three combined. If your Google Business Profile is not claimed and complete, if your hours are wrong in three places, if you have fewer than 20 reviews — do that first. Not because GEO is unimportant, but because it is downstream of exactly that work. BrightLocal found 47% of consumers will not use a business with fewer than 20 reviews, and an assistant reading the same sources reaches a similar conclusion.

BrightLocal, Local Consumer Review Survey 2026. Only 9% said they would use a business with five or fewer reviews.

The honest “not yet” cases

A vendor will not tell you this, so: you can reasonably defer this work if

• you are booked to capacity and turning work away,
• your business runs on referrals and repeat customers,
• you do not have a website worth crawling, or
• your listings are in a state that needs a month of cleanup first.

In every one of those cases the highest-return version of “doing GEO” is fixing the underlying thing, which was worth fixing anyway.

Check, do not guess

Before you buy anything, spend thirty minutes finding out whether any of this describes your market.

Open a logged-out session. Ask an assistant six versions of the question your customers would ask — with your city, and with a nearby suburb. Run each one three times, because of the variance problem above. Write down who gets named and, more usefully, which sources appear in the citations. That last list is your actual to-do list, and it is specific to your city and trade in a way no vendor’s template can be.

If you are named consistently, you are in a monitoring situation. If three competitors own every answer, you are in a foundations situation. If the assistants disagree wildly, your category is unsettled and there is room. The step-by-step version of that check is here.

What the work actually consists of

Assuming you decide it applies, the work is less exotic than the acronym suggests. Five buckets:

1. Accurate business information everywhere. Name, address, phone, hours and services, replicated identically across platforms. SOCi’s 2026 index found contact details in AI answers matched the business’s Google profile only about two-thirds of the time on ChatGPT and Perplexity. Unglamorous, and the single biggest lever.
2. Presence on the sources cited in your city. Not two hundred directories — the fifteen or so that actually show up in your citations. For lawyers that is often Avvo and Justia; for dentists, Healthgrades and Zocdoc; plus whatever local publication runs the “best of” roundup in your metro.
3. Reviews on more than one platform. The average consumer now consults six review sites, and assistants pull from a similar spread. A profile that is 200 Google reviews and nothing else reads as thin.
4. A site an assistant can reach and parse. Hours, service area and services stated in plain text, not baked into an image or a booking widget. Worth checking your crawler settings too — see below.
5. Pages that answer real questions. Costs, insurance accepted, what happens at a first visit. These get cited because they answer what people actually ask.

SOCi via Search Engine Land for the accuracy figure; BrightLocal for the six-review-sites figure. SOCi does not publish which fields the accuracy score covers, so this is stated as what was reported rather than as “one in three answers has a wrong phone number”.

On crawler settings specifically: on July 1, 2026 Cloudflare replaced its single “block AI bots” switch with three categories — Search, Agent and Training. As of September 15, 2026, Training and Agent crawlers are blocked by default on pages that display ads, while Search crawlers stay allowed. Those defaults apply to new customers, new sites added by existing customers, and all existing free-plan customers. Most local service sites do not carry ads and are unaffected, but an inherited default is easy to miss, so check which policy your zone is actually on rather than assuming.

Cloudflare, “Your site, your rules: new AI traffic options for all customers”. One trap in the fine print: Cloudflare evaluates multi-purpose crawlers under all of their behaviours, so blocking “Training” can also catch Googlebot, Bingbot and Applebot. A blanket block aimed at AI can quietly damage ordinary search rankings.

What you will get sold that does not work

Trust is scarce in this category, so a few things to push back on.

llms.txt files

A proposed standard for telling AI systems about your site. Ahrefs analysed 137,210 domains and found 97% of published llms.txt files received zero requests in a month. Nobody is reading them.

Ahrefs — of the traffic those files did receive, the largest identified slice was SEO audit tools checking whether the file existed.

Schema markup as a magic switch

Ahrefs tracked 1,885 pages that added JSON-LD schema against 4,000 control pages and found no meaningful citation uplift: +2.4% in AI Mode and +2.2% in ChatGPT, both statistically indistinguishable from noise. Schema is cheap and does other jobs — just do not pay a premium for it as an AI play.

Ahrefs. The study also found a −4.6% change in AI Overviews that was measurable against controls, but Ahrefs explicitly declines to attribute it to schema, and neither do we. Worth stating the scope: these were pages already being cited, so this tests whether schema lifts a visible page, not whether it rescues an unknown one.

“AI directory submission” packages

Largely a category invented to be sold. Fifteen right platforms beat two hundred generic ones.

Any guarantee of a ranking

Given the run-to-run variance above, nobody can promise placement in an AI answer. Anyone who does is either uninformed or counting on you not to check twice.

What to expect if you do it

Listings and technical fixes land in the first few weeks and are verifiable immediately. Movement in the answers themselves, if it comes, tends to show up over a couple of months as the corrected information propagates through the sources assistants read.

Be prepared for the honest starting point, too. In a market with three slots and three entrenched incumbents, zero mentions is the normal first reading, not a verdict on your business. Which is why the metric to watch early on is not “am I named yet”. It is whether the inputs are moving — listings claimed, information consistent, sources citing you even when they do not name you. Those you control. The answer, you influence.