Generative engine optimization (GEO): the complete guide to getting named by ChatGPT, Gemini and Perplexity

What is generative engine optimization?

Generative engine optimization (GEO) is the work of getting your business named, described accurately and cited inside the answers that AI assistants write. It covers ChatGPT, Gemini, Perplexity, Claude, Microsoft Copilot and Google AI Overviews and AI Mode. The goal is not a blue link on page one; it is a mention in the sentence that answers the customer's question.

When a traveler asks ChatGPT for "a boutique hotel near Times Square with late check-in", the reply is a short list with reasons. You are on it or you are not. A page can rank fourth in Google and still get clicks. An AI answer that names three hotels has no fourth.

The term comes from a 2023 research paper, "GEO: Generative Engine Optimization", by researchers at Princeton and partners, later accepted to the KDD 2024 conference. It reported that content changes could raise a source's visibility in generative engine responses by up to 40% (arXiv). We cover what those changes were further down.

The engines do not rank you. They decide whether to mention you. That changes what you optimize.

GEO, AEO and SEO in one paragraph

SEO earns a position in a list of results. Answer engine optimization (AEO) makes your pages and profiles easy for an answer engine to quote. GEO is the wider goal of being named in AI-written answers, across engines, and checking that you are. In practice the three overlap heavily; the AEO vs GEO vs SEO comparison sets out where they differ. For a one-line definition, see the GEO glossary entry.

Why GEO matters for local and multi-location businesses

AI answers are now part of ordinary search. Google says its AI Overviews and AI Mode can run several related searches behind one question (it calls this "query fan-out") and show a wider set of links than a classic results page (Google Search Central). Microsoft reports AI citations for Copilot and Bing AI summaries in Bing Webmaster Tools (Bing Webmaster blog).

For a local business the answer often does not produce a click at all. It produces a call, a Maps search or a walk-in. That is why zero-click search and GEO belong in the same conversation: the customer decided inside the answer.

For brands with many branches, that means GEO is a multi-location SEO problem as much as a content problem. Each location needs its own facts, its own reviews and its own page in good shape.

How GEO and local SEO fit together

Local SEO decides whether you show in the map pack and the local results. Its three classic levers are relevance, distance and prominence (local SEO ranking factors). GEO uses the same raw material: the profile that proves relevance, the address that fixes distance, the reviews and mentions that build prominence.

The difference is the last step. A map pack shows three businesses in an order you can track with a rank grid. An AI answer writes a sentence about each business it picks, and that sentence can be wrong, out of date or unflattering even when you are named. So GEO adds two jobs to local SEO: checking the answers themselves, and correcting the sources behind any wrong description. If you are new to the basics, start with what local SEO is and come back.

How generative engines build an answer

You do not need to know how any model works inside. You need the outline, because each step is a place where your business can drop out.

Each engine builds its answer from sources it trusts, and the mix differs. The table is a starting point; the sources each answer actually used should correct it for your category.

Crawler names: OpenAI and Perplexity publish their crawler documentation. Sources vary by question and change over time.

OpenAI says sites that block OAI-SearchBot will not appear in ChatGPT search answers, though they can still show as navigational links; GPTBot is the separate crawler for model training (OpenAI crawler docs). Perplexity recommends allowing PerplexityBot for sites that want to appear in its results (Perplexity crawler docs).

What the research and Google say actually works

There is a lot of GEO advice online, and much of it is guesswork. Here is what primary sources say, as of November 2026.

The GEO paper: evidence, quotes and numbers help; stuffing does not

The original study tested content changes on a large set of queries. It found that adding citations, quotations from relevant sources and statistics could significantly raise a source's visibility, that results varied by domain, and that keyword stuffing performed worse than doing nothing (GEO paper). For a local business, the translation is simple: state real facts with specifics (opening hours, prices you publish, distances, service areas) and back claims with a source.

Google: AI search optimization is still SEO

In May 2026 Google published an official guide on optimizing for its generative AI features. It says optimizing for generative AI search "is still SEO", and lists what you do not need: no new machine-readable files or AI text files, no splitting content into tiny chunks, no writing in a special way for AI, and no special schema.org markup. It also says Merchant Center feeds and Google Business Profiles help products and local businesses appear, and that seeking inauthentic mentions is not helpful (Google AI optimization guide).

Google also clarified in May 2026 that its spam policies apply to generative AI responses in Search, and in June 2026 that Google Search does not use llms.txt files for visibility or rankings (Search Central changelog). Other engines may read llms.txt; our llms.txt guide covers when it is worth adding.

Where ChatGPT found local sources in one study

BrightLocal ran 800 local business searches in ChatGPT in November 2024 across 20 US cities and recorded the first ten sources shown. Business websites made up the largest share; Reddit and Quora did not appear in that study (BrightLocal). It is two years old and engines change, so treat it as a reason to fix your own site first, not as a fixed rule.

What the named businesses usually have in common

When you compare the businesses an engine named with one it skipped, the gap is rarely mysterious. It is usually one of four things.

Complete profiles

The named business lists the attributes the question asked about. If the question is "with a pool" and the pool is not on the profile, the engine has nothing to go on. Start with the Google Business Profile optimization checklist.

Recent, specific reviews

Not just more reviews: reviews that mention the thing people ask for, from the last ninety days, with replies. Review summaries are built from them. Watch review velocity, not the lifetime total.

A page that answers the question

A location page that says "Check-in is 24 hours" in a sentence is quotable. A PDF of policies is not. See location pages for SEO for a template.

Consistency everywhere

Hours that match on Google, Apple Maps, Bing and Tripadvisor. One wrong number on one directory gives an engine a reason to hedge, and a hedging engine names someone else. Listings management and local citations are GEO work.

A GEO strategy in seven steps

GEO is not a project with an end. Answers change as reviews land and competitors update. This is the routine, written so you can run it by hand for one location.

1. Write the questions customers actually ask

   Ten to twenty per location: local ("hotel near Covent Garden with parking") and brand ("is Harbor Hotels Covent Garden good for families"). Start from reviews, calls and the front desk. Our guide on how to choose prompts to track has a method.

2. Make sure the engines can read you

   Allow OAI-SearchBot, PerplexityBot, Googlebot and Bingbot in robots.txt. Check that location pages are indexed and not hidden behind scripts, logins or PDFs.

3. Ask every engine and record a baseline

   ChatGPT, Gemini, Perplexity, Claude, Copilot and Google AI Overviews. Record whether you are named, where, how you are described and who else is.

4. Record the sources

   Perplexity and ChatGPT search show sources. For the others, trace the facts and phrases in the answer back to a profile, a review site or a page.

5. Close the gap at the source

   Fill the attribute, correct the hours on the directory, answer the reviews, publish the page that answers the question. One source at a time, per location.

6. Earn real mentions

   Local press, partner pages, industry associations and lists that people in your area actually read. No paid placements dressed up as editorial.

7. Check again every week

   Profile and listing fixes are usually picked up first; new pages and reviews take longer. Keep the log so you can see which engine caught up and when.

Location: {location name, neighborhood}

Local questions (customer does not know you yet)
1. best {category} near {landmark or neighborhood}
2. {category} near {neighborhood} open {day or late}
3. {category} in {city} with {attribute customers ask for}
4. affordable {category} near {neighborhood}
5. {category} near {neighborhood} for {audience: families, seniors, groups}

Brand questions (customer knows your name)
6. is {brand} {location} good
7. {brand} {location} vs {main competitor}
8. does {brand} {location} offer {service}

Competitor named instead (fill after the baseline)
9. ...
10. ...

A worked example: one location, four weeks

Northgate Family Dental (an invented example) writes twelve questions, including "dentist near Northgate open Saturday" and "dentist near Northgate that takes walk-ins". In the baseline, ChatGPT and Perplexity name two other practices for both. Gemini names Northgate for Saturday but says it is "closed weekends" in the same answer.

The log shows why. The Google profile has Saturday hours, but the practice's own location page still lists Monday to Friday, and a large directory has the old hours from two years ago. Perplexity cites that directory. Nothing on any page says "walk-ins welcome", even though the practice takes them.

Week one: fix the location page hours and add two plain sentences ("Open Saturdays 8am to 1pm." "We see walk-in patients when a chair is free; call ahead to check."). Week two: correct the directory listing and add the walk-in attribute where the profile allows it. Week three: ask recent Saturday patients for a review. Week four: re-run all twelve questions and compare. Whatever the result, the log now shows which fix each engine picked up, which tells the next location what to do first.

GEO content that earns citations

Google's guide says the most influential long-term factor is valuable, non-commodity content: first-hand experience and a point of view, not another generic summary (Google AI optimization guide). For a local business that is easier than it sounds, because you hold facts nobody else has.

• Location pages with real detail. Hours, services, parking, access, the team, nearby landmarks and the questions that location gets. One page per location.
• Pricing pages. If you publish prices, publish them plainly with what is included. "How much" questions are common and hard to answer without you.
• Policies in text. Cancellation, deposits, insurance accepted, returns, pets. Engines answer "can I" questions from these.
• Comparison pages you can stand behind. "Repair or replace", "same-day or booked appointment". Explain the trade-off honestly, with your recommendation.
• Local guides from people who live there. Where to park for the stadium, which entrance has step-free access. Useful to readers and quotable.
• Your own numbers. Average wait on weekday mornings, number of service bays, years at the address. The GEO paper found that adding statistics raised visibility; real figures from your own records are the honest way to do it.

Each of these also helps classic search, which is the point: the same page serves the map pack, the results page and the AI answer.

GEO for multi-location brands

With 20 or 200 locations you cannot hand-check everything weekly. Prioritize. Plot each location by how much demand it has and how often the engines name it, then work the top-left box first.

Head office owns the method and the question list; each location owns its facts. That split is the same one that keeps listings accurate, and it is how Locus Intelligence Enterprise runs governance by region and brand.

How to measure GEO

Track three numbers per location per week: how many engines name you for your question set, your position when they do, and the share of questions where a competitor is named and you are not (with the competitor's name). That last number tells you what to fix. Our guide to AI search visibility metrics defines each one, and share of voice in AI explains the most common summary metric.

Two free reports help. Google counts traffic from AI Overviews and AI Mode inside the Search Console Performance report under the Web search type (Google Search Central). Bing Webmaster Tools shows how often your pages are cited in Copilot and Bing AI answers, which pages, and sample "grounding queries" (Bing Webmaster blog).

Week of: {date}    Location: {location}

Question | Engine | Named? (Y/N) | Position | How described | Named instead | Sources cited | Fix this week
{question 1} | ChatGPT | N | - | - | {competitor} | {directory, article} | {action}
{question 1} | Perplexity | Y | 2 | "{words used}" | {competitor} | {sources} | -
...

Summary: named in {x} of {y} checks. Top gap: {source}. Owner: {name}.

Generative engine optimization services: what they include and how to choose

GEO services are agencies, consultants or software that run some or all of the routine above for you. A good one does four things: tracks real questions per engine, shows the sources behind each answer, fixes those sources, and reports change over time. Prices are rarely published; ask for a written scope.

Visibility by Locus Intelligence is our own product and one option in the software column: it runs your category's questions per location in ChatGPT, Gemini, Perplexity, Claude, Copilot and Google AI Overviews, records the sources each answer used and who was named instead, and drafts fixes. See how Visibility works.

GEO mistakes to avoid

See who gets named instead of you, and why

Visibility by Locus Intelligence runs the real questions for your category, per location, in every major engine, shows the sources each answer used and drafts the fixes.

Book a demo

AI visibility checker

Business and city, or a websiteCheck

Free, no account. You get whether ChatGPT, Gemini, Perplexity and Claude name you, how they describe you, and who they pick instead.