Most of what gets sold as GEO, Google says you don't need

Ask yourself a simple question: when you last needed a service you knew nothing about, a lawyer, an agency, a software tool, did you scroll through ten blue links, or did you ask an AI?

A lot of buyers have already switched. ChatGPT, Gemini, and Google’s AI Overviews now answer questions that used to end in a search results page. And here’s the part most businesses haven’t worked through yet: those answers name names. When someone asks who the best web design agency is for a growing brand, the answer often comes back with specific companies in it.

The discipline of earning a place in those answers is called Generative Engine Optimization, GEO.

It is also the most oversold phrase in marketing right now, so this piece does two jobs. It explains what GEO genuinely is, and it shows you, using Google’s own documentation, which parts of the standard GEO sales pitch are not true.

What Google actually says you need

Start here, because it saves you money.

Google publishes a page about its AI features in Search. On it, in plain language, Google states:

There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary.

And, on the question of markup and files:

You don’t need to create new machine readable files, AI text files, or markup to appear in these features.

And specifically on structured data:

There’s also no special schema.org structured data that you need to add.

What Google says you do need is ordinary eligibility:

To be eligible to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet.

Read those four statements together and a lot of the GEO market falls over. There is no secret file, no private schema, no back door. If an agency is selling you an llms.txt file as the key to ChatGPT, or a special markup package as the way into AI Overviews, they are charging you for a door that is already open.

We publish an llms.txt on this site. It costs nothing and it is a reasonable thing to have. We do not tell anyone it moves rankings, because Google has said it does not use it for Search, and no major AI company has published anything claiming otherwise for third-party sites.

One mechanism is worth knowing, because it is real and Google describes it. AI Mode and AI Overviews use what Google calls a query fan-out: rather than answering from one search, the system issues several related searches across subtopics and assembles the answer from a wider set of links. That is why an answer can cite pages that would never have ranked on page one for the original question. It also means the practical target is broader than one keyword. You want to be a credible result for the whole family of questions around your topic.

So what is left, once the snake oil is gone

Quite a lot, and it is more demanding than buying a file.

Be eligible in ordinary search. This is Google’s stated prerequisite and it is where most businesses actually fail. A page that is not indexed cannot be cited. A page ranked ninetieth is unlikely to be pulled into an answer. Boring technical health, indexation, and genuine search relevance are the entry ticket, which is why GEO and SEO are far less separable than the market pretends.

Write in the shapes an answer is assembled from. Answer engines lift fragments. That rewards a direct answer near the top of a page, headings shaped like the questions people ask, clean definitions, and specific numbers or claims worth quoting. It punishes brochure copy that circles a subject for four paragraphs before committing to anything.

Be described the same way everywhere. Your name, what you do, where you operate, and who runs it should match across your site, your profiles, and any directory you appear in. Contradictory descriptions make a model less confident about who you are, and confidence is what a recommendation is made of.

Get mentioned where these systems read. This is the uncomfortable one, because you do not control it. A large share of AI recommendations are assembled from other people’s shortlists: roundups, comparison posts, directories, forum threads. If your category has a “best X” article that everyone cites, being absent from it costs you more than any amount of on-page work will recover.

What the numbers actually show

The reason this matters is not that AI is fashionable. It is that the click is disappearing.

Pew Research followed the real browsing behavior of 900 US adults across 68,879 Google searches in March 2025. When an AI summary appeared, people clicked a traditional search result in 8% of visits. When no summary appeared, they clicked in 15%. Clicks on a link inside the summary itself happened in 1% of visits.

The same study found that AI summaries appear most on exactly the queries content marketing has been trained to chase: 60% of searches beginning with a question word produced one, and 53% of searches of ten words or more, against 8% for one and two-word searches.

Ahrefs looked at it from the ranking side, comparing 150,000 keywords with AI Overviews against 150,000 informational keywords without, using Search Console click data. Position one drew 34.5% fewer clicks when an AI Overview was present. Ahrefs has since updated that same study and now reports a 58% lower clickthrough rate for the top-ranking page, so both figures come from the one piece of research at two points in time.

Put those together and the shape of the problem is clear. The long-tail question post, the format that built a decade of content marketing, is now the format most likely to be answered without anyone visiting you. Ranking for it is worth less than it was. Being cited in the answer, and being the name a person remembers when they go looking, is worth more.

What GEO cannot do

Anyone selling certainty here is selling something they do not have.

Nobody can guarantee a recommendation. Ask the same model the same question twice and you can get different companies. The honest framing is odds across many askings over time, which is why measurement means tracking a trend rather than checking a box.

Nobody can buy organic citations. There is no paid inclusion in the cited sources of an AI answer. If that is being sold to you, it is either advertising in a different costume or it is untrue.

It does not work in days. Google’s own guidance for search generally is to expect four months to a year between making changes and seeing benefit, and the AI surfaces sit downstream of the same index. Anyone quoting you a date is guessing.

How to measure it without buying a tool

There is a growing market of AI visibility platforms. You do not need one to begin, and at the start they mostly tell you what a spreadsheet would.

Do this instead, once a month:

1. Write down ten questions a real buyer would type, in their words. Not your keywords. The actual sentence someone types at eleven at night.
2. Ask each one in ChatGPT, Perplexity, Gemini, and Google, in a fresh session with no history.
3. Record who gets named, in what order, and which pages get cited underneath.
4. Keep the log. The single result is noise. The trend across months is the signal, and the list of who keeps appearing is your real competitor set.

That log is also the most persuasive document you can put in front of a prospect, because it is about them and it is not a claim you made up.

Where the window actually is

The honest version of the opportunity is narrower than the hype and still worth having.

GEO is not an untouched field, and plenty of businesses are working on it. What is genuinely uneven is the quality of the work. Most of the effort in the market right now goes into the things Google has explicitly said are not required, which means a business doing the unglamorous version, real indexation, answer-shaped pages, consistent description, and earned mentions, is competing against a lot of motion in the wrong direction.

That advantage will not last forever. It lasts while the market is still buying files and markup packages.

Where to start

Start by measuring, using the ten-question method above. If your name never appears, you have a visibility gap and a starting position.

That measurement is what our AI Visibility Audit does, and you can have it whether or not we work together. When you want the whole discipline rather than the snapshot, our AI Search Visibility service covers the fixes, the citable content, and the monthly proof. For the hands-on version, we wrote a step-by-step guide to getting recommended by ChatGPT, and a closer look at how pages get cited by AI answer engines.

And if none of your pages are showing up in ordinary search yet, that is the thing to fix first, because it is the prerequisite Google names. We covered it in why your business doesn’t show up on Google.