That Marketing Buddy

What Is GEO (Generative Engine Optimization)? And What Is AI Visibility?

GEO (generative engine optimization) is the work of getting your business named and cited when an AI assistant writes an answer. AI visibility is the result: how often, and how well, you show up in those answers. Here is how it works, what to do first and how to report it.

Joonas Rotko
Joonas Rotko

SEO & GEO professional · Updated November 17, 2026 · 11 min read

Affiliate links on this page may earn me a commission, at no extra cost to you. They don't change what I recommend.

In short

GEO is the work of getting your business named or cited in answers written by AI assistants and Google's AI features. AI visibility is the result you measure: mentions, citations and your share of answers across a fixed set of questions. Google says its AI features rest on its normal search systems, and OpenAI says a site must allow OAI-SearchBot to appear in ChatGPT search. Do the SEO basics and the GEO work together: in my view, nearly every business whose services people ask about gains from starting now. In my two definition searches, the top 10, Google's AI Overview and ChatGPT named mostly different sources, so check each one.

Contents (8)
  1. 1How Do You Express AI Visibility?
  2. 2How Does an AI Assistant Choose Which Businesses and Sources to Name?
  3. 3What Did Two Definition Searches Name?
  4. 4What Does GEO Work Look Like, and in What Order?
  5. 5Who Benefits From GEO?
  6. 6What Do Marketing Teams and Agencies Need on Top of That?
  7. 7What Are the First Steps for a Small Business?
  8. 8Frequently Asked Questions

GEO, short for generative engine optimization, is the work of getting your business named and cited when an AI assistant writes an answer. The assistant can be ChatGPT, Gemini or Perplexity, or a Google feature such as AI Overviews. AI visibility is the result: how often, and how well, your business shows up in those answers. GEO is what you do, AI visibility is what you measure.

Tip

Key points

GEO is mostly good SEO and good marketing, pointed at the places assistants read. Do both together.

AI visibility has three measures: mentions, citations and share of answers.

Access comes first: a page that is not indexed, or a crawler that is blocked, cannot be named.

Other sites matter a lot for recommendation questions, so part of the work happens off your own site.

Measure with the free platform data first, then a tracker. Show the sample size next to every percentage.

GEO is also called AI SEO or LLM SEO, and some people use AEO (answer engine optimization) for part of it. The terms are used loosely. I use GEO for the whole job of being named in AI-written answers. The side-by-side is in my SEO vs GEO vs AEO comparison.

How Do You Express AI Visibility?

Three numbers cover almost everything: mentions, citations and share of answers. They start from the same setup: a fixed list of questions a customer might ask, run again and again in the assistants you care about.

The three ways to express AI visibility

Mention

What it counts
The assistant names your business in the text of the answer
Example
"Taskwell is a project tool for small agencies" appears in the answer

Citation

What it counts
The assistant links one of your pages as a source
Example
Your pricing page is listed under the answer as a source

Share of answers

What it counts
The percentage of runs of a question where you are mentioned or cited
Example
You are named in 6 of 20 runs, so 30%

Taskwell is a made-up business. Share of answers needs many runs because an assistant can answer differently each time.

A business can be mentioned without being cited, and cited without being recommended, for example as the source of a price. Track both, and write down the wording too. How to run the measurement, and what to report, is in my guide to tracking your brand in AI answers. The tactics differ by assistant, so getting cited in Google AI Overviews and showing up in ChatGPT answers each have their own page.

How Does an AI Assistant Choose Which Businesses and Sources to Name?

It draws on three things: what the model learned in training, what it fetches from the web while answering, and what other sites say about you. The mix differs between assistants and the vendors publish little detail, so I rely on their own documentation where it exists.

Does the assistant search the web while it answers?

Google describes its AI features as search-based. Its documentation says AI Overviews and AI Mode may use a "query fan-out" technique, "issuing multiple related searches across subtopics and data sources", and then identify supporting pages to link (Google Search Central: AI features). In its guide to generative AI features, Google adds that these features ground answers by "relying on our core Search ranking systems to retrieve relevant, up-to-date web pages from our Search index" (Google Search Central: optimizing for generative AI features).

In plain words: one question becomes several searches, and the answer is written from what comes back. This is Google's description of Google's features. Other assistants publish less.

What decides whether you can be named at all?

Access is the first gate. For Google, a page must be "indexed and eligible to be shown in Google Search with a snippet", with no additional technical requirements. For ChatGPT, OpenAI documents that "sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers", though they can still appear as navigational links. OpenAI also lists GPTBot, which it uses for training its models, and ChatGPT-User, which it says is not used to decide whether content appears in search.

OpenAI says a robots.txt change can take about 24 hours to take effect (OpenAI: overview of crawlers). That is about when a crawler can reach you, not about when you will be named.

Check robots.txt once and decide on purpose which crawlers you allow. Then check the layer teams often forget: whether your CDN or firewall blocks the same crawlers, because a block there can stop a crawler whatever robots.txt says.

Does ranking in Google decide who gets named?

It helps, but it does not settle it. Ahrefs analyzed 863,000 search results and 4 million AI Overview URLs and found that 37.9% of the cited URLs were also in the first 10 results of the same search, while 31.0% were beyond the top 100 (Ahrefs, March 2026). Ahrefs sells SEO software, and the data comes from its own Brand Radar tool, so its matching method shapes the numbers. One possible reason for the cited pages outside the top 10 is the fan-out method above: a page may rank for one of the smaller searches.

Ahrefs does not test that.

Do other sites matter more than my own?

For recommendation questions, often yes. When someone asks "best X for Y", an assistant reads lists, reviews, directories and discussions, not only company sites. That is why GEO has work outside your own website: accurate listings, reviews and mentions in the places assistants read.

What Did Two Definition Searches Name?

In these two searches, the top 10, Google's AI Overview and ChatGPT named mostly different sources. On November 16, 2026, I ran "what is generative engine optimization" and "what is ai visibility" as US mobile searches, put the same question to ChatGPT with web search on, and sorted who was named.

Who each answer type named, two definition searches

what is generative engine optimization

Top 10 results
7 agency or software company pages, plus Quora, Reddit and LinkedIn
Google AI Overview
9 links: Coursera, Wikipedia, Reply, Evergreen Media, a Storyblok blog post, Quora, Instagram and 2 YouTube videos
ChatGPT with web search
7 domains: Search Engine Land, HubSpot, PMG, Semrush, Ahrefs, Profound and an academic preprint on arXiv

what is ai visibility

Top 10 results
7 software or data company pages, 1 agency, YouTube and one Google-hosted result
Google AI Overview
6 links: Frase, Visiblie, The Write Direction, YouTube and 2 more via Google
ChatGPT with web search
8 entries: Semrush, Similarweb (2), Yext, a Google guide, a SEO and GEO glossary, openseo.gg and Ahrefs

Data: DataForSEO SERP API and ChatGPT scraper with web search forced on, US, mobile, November 16, 2026. Some Google links are redirects, so the site is read from the title.

For the GEO search, the only domain in two lists was Quora (top 10 and AI Overview). ChatGPT's seven domains did not overlap with either. For the AI visibility search, the AI Overview and the top 10 shared three sources (Frase, The Write Direction and YouTube), and ChatGPT shared two with the top 10. With two searches these are examples, not rates.

What I take from it: for these searches the lists were mostly company, video and Q&A sites, so read any definition, including mine, with that in mind. And the three places picked different sources, so ranking in Google does not tell you what ChatGPT names. Check each one.

Note

Limits of this test

Two searches, one day (November 16, 2026), one country (US), mobile results. ChatGPT answers change between runs, so a repeat could name other sources.

The categories (agency, software company) are my reading of each site's own pages.

This shows who was named for two definition searches. It says nothing about recommendation searches such as "best X", where the pattern may differ. A larger test, with its own limits, is in [my SEO vs GEO vs AEO comparison](https://thatmarketingbuddy.com/blog/seo-vs-geo-vs-aeo).

What Does GEO Work Look Like, and in What Order?

GEO work is mostly ordinary good marketing pointed at the places assistants read. Google says so for its own features: it states that "optimizing for generative AI search is optimizing for the search experience, and thus still SEO", and that you do not need new AI text files such as llms.txt, special markup, or writing made only for AI (Google Search Central). The GEO part is what sits around that base.

This is the order I would work in. It is my suggestion, not a tested method.

  1. Access. Pages indexed, crawlers allowed in robots.txt and in your CDN or firewall. Nothing else counts without this.
  2. Facts. Name, address, hours, prices and what you do should match on your site, Google Business Profile and every listing. An assistant that finds two versions has to guess.
  3. Pages that answer real questions. For questions customers actually ask, put the direct answer in the first sentence under a question heading. Do not create near-copy pages to catch more searches.
  4. Presence where assistants read. Review sites, directories, articles in publications of your field and community discussions that name businesses like yours. This is slower and mostly outside your own site, so give it an owner and a monthly slot.
  5. Measure. Free platform data first, then a fixed prompt set in a tracker.

For the full step list, see my guide on how to rank in AI search. If you sell through a local profile, the ChatGPT Business Hub guide covers claiming your business inside ChatGPT.

Who Benefits From GEO?

Almost every business whose services or products people ask questions about. If a buyer can type "who is good at this", "what does this cost" or "which one should I pick" about your category, an assistant is already answering, with or without you in the answer. GEO is already here. In my view, businesses that invest in it actively now can grow their visibility, and their business with it, significantly.

Do the SEO basics well at the same time, because assistants search the web to write their answers and find pages that are indexed, clear and trusted.

Where GEO work starts, by type of business

Software and B2B services

What buyers ask assistants
Shortlists, comparisons, pricing, alternatives
Where to start
Clear product and pricing pages, and accurate entries on the comparison pages assistants read

Agencies and consultants

What buyers ask assistants
"Who does X for companies like mine"
Where to start
Service pages that say who you serve and what it costs, plus mentions on sites in your field

Local service business

What buyers ask assistants
"Best X near me", opening hours, prices
Where to start
Listings and reviews that match everywhere; this also helps normal search

Online shop

What buyers ask assistants
Product questions and "which one for my use"
Where to start
Complete, accurate product pages and answers to the questions buyers ask

Brand-new site

What buyers ask assistants
Category questions where you are not yet known
Where to start
Get pages indexed and build mentions on other sites from the start, because assistants read both and a new brand has little written about it yet

A rule of thumb from my own work, not a measurement.

On volume: Ahrefs' October 2026 benchmark of 118,671 sites found AI assistants sending 0.46% of visits, against 24.74% from Google (Ahrefs). Ahrefs sells SEO software. AI visits are a small share today and arrive with high intent, and being named shapes the shortlist even when nobody clicks. My SEO vs GEO vs AEO comparison has the conversion studies.

What Do Marketing Teams and Agencies Need on Top of That?

A fixed prompt set, an attribution setup and a report that shows sample sizes. The basics above are the same for every business. These parts are what make the work reportable.

  • Prompt set. Choose prompts from Search Console queries, sales calls and the questions customers ask. Split prompts that name your brand from prompts that do not, and report them separately. Brand prompts flatter the result.
  • Attribution. Visits from chatgpt.com, perplexity.ai and similar referrers show in analytics, but some AI visits can arrive as direct traffic, so treat the count as a floor. For Google, the Generative AI performance report in Search Console shows how often your pages appear in AI Overviews and AI Mode (Search Console Help: Generative AI performance report), but the clicks are counted inside the Web search type and cannot be filtered out.
  • Free platform data. ChatGPT Business Hub Insights shows mentions and website clicks. The AI Performance report in Bing Webmaster Tools shows citations in Microsoft's AI answers. Neither shows competitors.
  • Sample size. Answers change between runs. Report counts next to every percentage, judge trends over months, and keep the prompt set frozen for a quarter.
  • Ownership. SEO tasks sit in the site editor. Off-site tasks (reviews, listings, mentions) need a named owner and a monthly slot, or they do not happen.

The details, with a report template, are in how to track your brand in AI answers. If you want a tool, my comparison of AI visibility tools shows what they cost.

What Are the First Steps for a Small Business?

Start with five questions and a notebook, before you buy anything.

  1. Write down the five questions a new customer would ask an assistant to find you, such as "best bookkeeper for freelancers in Austin".
  2. Ask each one in ChatGPT, Gemini and Perplexity, and in Google, and write down every business and source named.
  3. Mark where you appear, and which competitors do. Open two of the cited pages and see why they were picked: a clear answer, a list, a review.
  4. Fix the cheap things: matching details on your site and listings, a direct answer on the page for each question, crawler access.
  5. Claim your free profiles and repeat the same questions in a month. If the list grows past about twenty questions, a tracker saves time.

Frequently Asked Questions

Is GEO the same as SEO?

No, but they overlap heavily. SEO aims at rankings and clicks from a list of results. GEO aims at being named in an AI-written answer, and for Google's AI features it builds on the same ranking systems. The comparison has the table.

Do I need llms.txt for GEO?

Google says you do not need new AI text files for its generative features. Other assistants publish no requirement that I have found, so I do not treat it as a first step.

How long does GEO take to show results?

I have no measured timeline. A crawler access change can apply within about a day, per OpenAI, but that only means the crawler can reach you. Mentions that depend on other sites, such as reviews and lists, take as long as those sites take to add you. Plan in months, and judge by a trend over several months, not one check.

Sources

Once you know your first five questions, a tracker can run them for you: compare AI visibility trackers by real price.

Compare AI Visibility Trackers →

Read next

Joonas Rotko

Joonas Rotko

SEO & GEO professional, founder of That Marketing Buddy

I'm a Finnish SEO and GEO professional with 11+ years in digital marketing, including years as a co-owner of a digital marketing agency. On That Marketing Buddy I test and compare SEO and AI-visibility software for entrepreneurs, marketing professionals and agencies.

Hands-on tests

Get my newest hands-on tests of SEO and GEO tools

I test SEO and AI visibility tools on my own sites and send you what I found: real prices, how accurate the data was and what didn't work.

At most one email a week. Unsubscribe anytime. I use your email only for this. Privacy policy