What is GEO? Generative engine optimization guide

Generative engine optimization (GEO) means shaping what AI tools write about your market so that your brand is named, recommended and linked. A "generative engine" is any product that writes an answer instead of listing links: ChatGPT, Gemini, Claude, Perplexity, Microsoft Copilot, and Google's AI Overviews and AI Mode.

The term comes from a 2023 research paper by Princeton, Georgia Tech, the Allen Institute and IIT Delhi, which tested how page changes affect whether AI answers cite a source. Today "GEO" is the umbrella term most marketers use. AEO (answer engine optimization) and LLM SEO describe the same work, and GEO vs SEO vs AEO shows where GEO and classic SEO part ways.

GEO, AI search optimization and LLM SEO

You will see the same work sold under several names. They overlap almost completely:

Term	Who tends to use it	What it usually means
Generative engine optimization (GEO)	Researchers, most marketers	Getting named and cited in AI-written answers
AI search optimization	Agencies, SEO tool vendors	The same, with emphasis on Google AI Overviews, AI Mode and ChatGPT search
LLM SEO	Developers, SaaS marketers	The same, sometimes also covering what models learned in training
Answer engine optimization (AEO)	Older term from voice and featured snippets	Writing pages that answer a question directly

Pick whichever word your team searches for. The checklist below is the same for all of them.

Why GEO matters now

Buyers ask AI tools questions they used to type into Google, especially comparisons and recommendations: "best CRM for a 5-person agency", "is X worth it", "X vs Y". The answer names three to six products. If yours is not one of them, the buyer may never visit your site.

Google itself now answers many searches with an AI Overview above the results. Your page can rank on page one and still lose the click if the overview names a competitor.

How generative engines build an answer

The engine turns the question into one or more web searches.
It reads the top results, often from Bing or Google, plus its own index.
It writes an answer from what it read and its training data, and links some of the sources.

So GEO has three levers:

Be findable. Pages that rank for the searches the engine runs get read. Classic SEO still feeds GEO.
Be quotable. Pages with a plain answer, specific facts, prices and comparison tables give the engine something to lift.
Be mentioned elsewhere. Lists, reviews, forums and comparison pages that name you get read too. For recommendation questions they often matter more than your own site.

A GEO plan you can run in a week

Day 1: choose 15 to 30 questions

Write the questions a buyer asks on the way to choosing a product like yours. Mix:

Category questions: "best [category] for [audience]".
Comparisons: "[you] vs [competitor]", "[competitor] alternatives".
Problem questions: "how do I [job your product does]".

Skip vanity questions about your own brand name until later; buyers who already know your name are not the gap.

Day 2: record today's answers

Ask every question in ChatGPT and Gemini, and check whether Google shows an AI Overview for the matching search. For each answer note:

Is your brand named?
Is your site linked?
Which competitors are named?
Which pages are cited?

Day 3: rank the cited pages

Put every cited URL in one list and count how often each appears. A short list of pages usually accounts for most citations. Split it into your pages, competitors' pages and third-party pages.

Days 4 and 5: change the pages that feed the answers

Third-party pages that cite competitors but not you: contact the author with a short note on what is missing, a comparison, or a trial account. Correct outdated facts about you.
Your own pages: put a direct answer in the first paragraph, add a comparison table, show current prices, and state facts plainly (numbers, limits, supported platforms).
Access: confirm that robots.txt lets AI crawlers in (OAI-SearchBot and GPTBot for OpenAI, PerplexityBot, ClaudeBot; Google's AI Overviews use the normal Googlebot) and that key content is in the HTML, not only rendered by JavaScript.

After that: measure on a schedule

Answers vary between runs and between engines, so one check proves little. Ask the same questions weekly and track two numbers per engine: the share of answers that name you (mention rate) and the share that link you (citation rate).

AI search optimization checklist

Use this for each page you want AI answers to cite:

The page ranks, or can rank, in Google and Bing for the matching search. Engines mostly read pages that already rank.
The first paragraph answers the question in two or three plain sentences.
Prices, limits and supported platforms are written as text with a date, not hidden in images or tabs.
Comparisons are in a table with named products.
Claims have a source or your own data behind them.
The content is in the HTML the server returns, not only rendered by JavaScript.
robots.txt allows OAI-SearchBot, GPTBot, PerplexityBot, ClaudeBot and Googlebot.
Third-party lists and reviews for your category mention you, or you have contacted their authors.
You track the same questions weekly and know your mention and citation rate per engine.

Structured data is not on the list as a requirement. The schema guide in the list below explains what it does and does not do.

GEO myths to skip

"Add an llms.txt file and AI will cite you." No major answer engine has said it uses llms.txt to choose sources. It helps agents read your docs; it is not a ranking signal.
"Stuff pages with statistics and quotes." The research found that adding sources and statistics helped in tests, but invented numbers hurt trust and can get you corrected in public. Add real data or none.
"There is an AI search volume for each prompt." Nobody outside the AI companies can measure that. Choose questions by buying intent.

Measuring GEO with zaDojo

zaDojo tracks AI answers alongside normal SEO data, from the web app or from your AI agent over MCP:

Prompt Research: enter a keyword and get related questions, ChatGPT's answer to each and the cited sites. About €0.25 per keyword, US English, with questions drawn mostly from Google's "People also ask".
Prompt Tracking: your saved questions, asked daily, weekly or monthly in ChatGPT and Gemini, with Google AI Overviews, Claude and Perplexity as options. It records mentions, competitors and cited pages. A scheduled check costs about €0.002 per question per engine, so 30 questions on two engines weekly is roughly €0.50 a month.
Citations view: every cited page, split into yours and third-party, so day 3 above is a filter instead of a spreadsheet.
AI Visibility Audit skill: your agent reads the answers and cited pages and proposes the few changes most likely to get you named.

The plan is €10 a month with €10 of usage included; new accounts get €0.10 of trial credits with no card. See the AI visibility docs for every option and cost.

Everything on AI search visibility

Each guide below covers one part of GEO in more depth:

What is AEO: the answer engine view of the same work, with a one-week start and how to measure it.
GEO vs SEO vs AEO: what changes from classic SEO, what does not, and how to split your time.
AI Overviews and AI Mode: how Google picks the pages its AI features cite.
Perplexity SEO: how Perplexity chooses sources and how to get onto them.
How to track your brand in ChatGPT: mention rate, citation rate and AI share of voice, week by week.
ChatGPT for SEO: using ChatGPT as a tool for your own SEO work, with real data connected.
llms.txt: what the file does for AI agents, and why it is not a citation signal.
Schema markup for AI: which structured data is worth adding and what it cannot do.
Best AI visibility tools: trackers compared on price and coverage.

Free tools and features for the same job:

AI Overview checker: see whether a Google search shows an AI Overview and which sites it cites.
llms.txt generator: build a draft llms.txt for any site from its sitemap.
AI visibility checker: a quick free look at how AI answers treat a brand.
AI brand visibility tracking: scheduled tracking of mentions, citations and competitors across engines.

Questions people ask

Is GEO the same as SEO? It builds on SEO, because engines read pages that rank. The difference is the goal (being named in an answer, not ranking a link) and the extra weight on third-party mentions.

Can I pay to appear in ChatGPT answers? Not in the organic answer. Ads in some AI products are separate and labeled.

How often do AI answers change? Often. The same question can name different brands on different days and in different engines, which is why tracking over time beats a single screenshot.