Guides
What is generative engine optimization, in plain words
No acronyms first: it is the work of getting a page read and repeated by a tool that writes an answer instead of listing links.
What is generative engine optimization? It is the work of making pages easy for an AI assistant to find, read and repeat. The assistant does not hand you a list of links to choose from. It writes one answer, and your brand is either in that answer or it is not.
The term is younger than the practice. A 2023 research paper titled GEO: Generative Engine Optimization gave the idea a name and a test: change a page, then see how often its content shows up in a generated answer.
Two kinds of result
A search engine returns pages. You scan them and pick one. A generative engine, such as ChatGPT with search on, Perplexity, Gemini or Google's AI Overviews, does the reading for you: it turns your question into several smaller searches, opens the pages that come back and composes a reply with a few sources attached.
That means a page can be useful without ranking first, and it can rank first without ever being quoted. What counts is whether the assistant found the passage, trusted it and repeated it.
What it asks of a page
Four things cover most of it, and none of them is exotic.
- Let the crawlers in. Check robots.txt and any CDN bot rule. If an AI search crawler is refused, no amount of good writing will put the page in an answer.
- Keep the words in the HTML. Most AI crawlers read what the server sends and do not run JavaScript. A price or an FAQ that appears only after the app loads can be missing entirely.
- Answer first, then support it. Open a section with the direct answer, name the product instead of writing "it", and add a number or a named source.
- Be one brand. One name, one logo and consistent profile links, so the assistant does not treat your company as two.
What success looks like
Not a position. One of three outcomes: the answer names your brand, it links your page as a source, or it repeats your fact without credit.
That is harder to count than a ranking, which is why sampling exists. The AI Answer Visibility tool at seogeoaeo.ai puts one buyer question to ChatGPT, Perplexity and Gemini with web search on, then reports whether each reply names you, links your site and which competitors it names instead. A run costs 120 credits.
On the page side, cheaper checks cover the setup: the Robots.txt and AI Crawler Checker is 10 credits, the Brand Entity Checker 8 and the Citability Analyzer 5. Answers shift between runs, so repeat the same question and compare, rather than trusting one screenshot.
What it is not
It is not a replacement for SEO. A page that cannot be crawled, indexed or understood has nothing to quote. It is not paid placement, since you cannot buy a line in a model's answer. It is not a single file you upload: seogeoaeo.ai's own reference says llms.txt is optional and is not a search or visibility requirement. And it is not a guarantee, because no tool can see why an engine chose one source over another.
Where to start
Pick one page that sells something and one question a customer would ask. Check crawler access and the served HTML, rewrite the opening of each section as an answer, then sample the question and write down the date. A website assessment at seogeoaeo.ai covers the page side in one run: 67 credits for the home page, 47 for each extra page, with scores and a prioritized fix list you can work through.