GEO: how a website gets cited in AI answers

For years, being found online meant one thing: ranking on Google. Today a growing share of questions goes to AI assistants instead, and Google itself answers many searches with an AI summary at the top of the page. The answer often comes first, and the links come second, if at all.

Generative Engine Optimization (GEO) is about making sure your business shows up in those answers: mentioned, quoted and linked as a source.

1. GEO: what it is and how it differs from SEO

SEO aims for a place in a list of results. GEO aims for a place in a written answer. The difference matters:

A list shows ten links. An answer often names two or three sources, or none.
A list sends clicks. An answer may settle the question right there, so the mention itself becomes the result: your name in front of the person asking.
A list ranks pages. An answer stitches together statements, so single facts and clear passages count, not just whole pages.

GEO doesn’t replace SEO. Most assistants look things up in a search index before they answer, so a site that search engines can’t find or don’t trust rarely appears in AI answers either. GEO builds on SEO and adds to it.

2. Sources: how assistants choose them

When an assistant answers a question about current or specific facts, it typically runs a web search in the background, reads a handful of pages and writes an answer from them. Two things decide whether a page makes it in:

It gets found: it ranks in the search the assistant runs, and the assistant’s crawler is allowed to read it.
It’s useful: it answers the question clearly, in a passage that can be quoted without the rest of the page.

On top of that, assistants learn from their training data what a business is and what it’s known for. Consistent information across the web (your site, directory listings, profiles, articles that mention you) shapes that picture.

3. Content: what makes it quotable

Content that ends up in AI answers tends to share a few traits:

The answer comes first. A question as a heading, followed by a direct answer in the first sentence or two, then the details.
Facts are specific. Numbers, dates, places, prices where possible, names of services. “We build websites” is hard to quote; “We build accessible websites for craft businesses in Berlin” is not.
The structure is clear. Headings, short paragraphs, lists and tables make it easy to pull out one piece.
There’s an FAQ. Real questions customers ask, answered in a few sentences each.
Someone stands behind it. An author, a company, contact details, references. Assistants, like search engines, prefer sources they can attribute.
It’s current. A visible date and content that gets updated signal that the information still holds.

4. Technical: making it readable for machines

Crawlers allowed: the AI crawlers of the big providers are explicitly allowed in robots.txt. Blocking them by accident is easy, for example with a blanket rule or a default setting in a security tool.
Content in the HTML: pages whose text only appears after JavaScript runs are harder for many crawlers to read. Text that’s in the HTML from the start is safe.
Structured data: Schema.org markup describes the business, its services, its FAQ and its articles in a form machines read directly.
llms.txt: a plain-text summary of the site for language models. It’s a proposal, not a standard, and not every provider reads it, but a clean overview does no harm.
Speed and stability: a page that times out or errors when the crawler arrives isn’t in the answer.

5. Measuring: what to look at

GEO is harder to measure than SEO, because there’s no single ranking to check. What helps:

Ask the assistants. Pose the questions your customers would ask and see who gets mentioned, and how your business is described.
Check referral traffic. Visits from AI tools show up in analytics as their own sources.
Watch mentions. Being named on other sites, in directories and in articles feeds into what assistants know.

A GEO audit puts all of this together: what assistants currently say about a business, where the gaps are, and what content and technical changes would close them.

The short version

GEO aims for a place in AI answers, not just in a list of links.
It builds on SEO: assistants search before they answer.
Quotable content answers first, names specific facts, has a clear structure and a real author.
Technical basics: crawlers allowed, text in the HTML, structured data, llms.txt.
Measuring means asking the assistants, checking referral traffic and watching mentions.

If you’d like to know how AI assistants see your business, get in touch.