There is an industry selling Answer Engine Optimization right now.

Courses, certifications, agencies, tools. All built on the idea that AI search is a new discipline you need new tactics for.

Google published a guide for this in May 2026. Five months later it has a heading for both acronyms and one paragraph that answers them.

"From Google Search's perspective, optimizing for generative AI search is optimizing for the search experience, and thus still SEO."

That is the whole of AEO and GEO, from the company that runs the search engine they are optimising for.

And the same page carries a list of things you can stop doing. Not theories from a forum. Google's own list, headed:

"Mythbusting generative AI search: what you don't need to do"

Five bullets follow. Here they are, each with what it means if you have already paid for it.

AEO, GEO, and SEO, in one line each

SEO is making a page that Google can find, understand, and show to somebody.

AEO is "Answer Engine Optimization". Being the answer when an AI writes a reply.

GEO is "Generative Engine Optimization". Same goal, newer acronym.

Google's position is that the third one is the first one, and the other two are names for it.

"If you're considering third-party "AEO" or "GEO" advice or services, review our guidance on evaluating third-party SEO advice."

That is a search engine telling you to be sceptical of the people selling you search advice. It links to a page about third-party SEO advice that predates all of this.

Myth 1: you need an llms.txt file for AI search

An llms.txt is a text file some people publish so AI systems can read a short version of the site. There are generators for it now.

Google's answer:

"You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search (including its generative AI capabilities), as Google Search itself doesn't use them."

And then the part that settles whether the file hurts:

"Doing so will neither harm nor help your site's visibility or rankings in Google Search, as Google Search ignores them."

Ignoring is the operative word. Not penalised. Not blocked. Ignored.

I went further into this contradiction in Google Search calls llms.txt unnecessary, where the same company publishes one for itself.

Myth 2: you have to chunk your content for AI

"Chunking" means cutting a page into small pieces so a model can digest them. It is on a lot of checklists.

"There's no requirement to break your content into tiny pieces for AI to better understand it. Google systems are able to understand the nuance of multiple topics on a page and show the relevant piece to users."

And the sentence that should end the length argument everywhere:

"There's no ideal page length, and in the end, make pages for your audience, not just for generative AI search."

This is the same answer Google has given about word count for years. I counted how often Google uses the phrase "word count" in the article about thin content. It is zero.

Myth 3: you have to rewrite your content for AI systems

The advice here is usually "write in short sentences", "answer the question first", "cover every phrasing".

"You don't need to write in a specific way just for generative AI search. AI systems can understand synonyms and general meanings of what someone is seeking, in order to connect them with content that might not use the same precise words."

Then the direct answer to the keyword-stuffing instinct:

"This means you don't have to worry that you don't have enough "long-tail" keywords or haven't captured every variation of how someone might seek content like yours."

You do not need a page for every phrasing. Google is telling you it already connects near-misses. That is the same answer Google gives about AI-written content, which I checked in Does Google penalise AI content?.

Myth 4: you should chase the follow-up questions AI asks itself

This is the most expensive myth and the least known.

Google's AI features generate extra queries to research a question. Google named it:

"Query fan-out: A set of concurrent, related queries generated by the model to request more information and fetch additional relevant search results to address the user's query."

The guide's own example. A user types "how to fix a lawn that's full of weeds", and the extra queries are "best herbicides for lawns", "remove weeds without chemicals", "how to prevent weeds in lawn".

A whole cottage industry appeared around this: find the fan-out queries, write a page for each one, own the answer.

Google's response:

"While it might be tempting to create separate content for every possible variation of how people might search (for example, by focusing on other queries that people have asked, or fan-out queries), doing so primarily to manipulate rankings or generative AI responses in Google Search violates Google's scaled content abuse spam policy."

Google named the tactic and then named the policy it breaks. Writing pages for fan-out queries, to catch them, is scaled content abuse. That is the same policy that got a 1,248-page site removed from AdSense, and I went through what it actually says rather than what people repeat about it.

Myth 5: you need special structured data for AI search

Structured data is the machine-readable summary in a page's code. Every AI-search checklist has a section on adding new types of it.

"Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add. However, it's a good idea to continue using it as part of your overall SEO strategy, as it helps with being eligible for rich results on Google Search."

Required: no. Useful for rich results: yes. The reason to have it is the reason that already existed.

And the one line about the tools

After the five bullets, the guide adds something I have not seen quoted anywhere:

"No third-party tool has access to our internal ranking or AI systems."

Any tool claiming to read Google's internal metrics is describing something it cannot do. Google also says to "be wary of third-party tools that promise ranking success".

What actually decides it: two mechanisms

Strip the acronyms and Google describes two things.

Retrieval-augmented generation. Also called grounding. Google's AI answers are built from pages in the ordinary Search index, retrieved by the ordinary ranking systems, then summarised with links to the sources. The index it reads is the one you are already trying to be in.

Query fan-out. The model writes extra queries, reads more results, then answers.

That is also why which AI crawlers you allow is a different decision from which ones you block, which I went through in Blocking GPTBot does not remove you from ChatGPT.

Both start with a page that is indexed and eligible to be shown with a snippet, which is what Google says makes a page eligible for generative AI features at all.

"a site must be included in Search generative AI features in Search Console to be eligible for display in generative AI features on Google Search"

So the entry requirement is not a new technique. It is being in the index, and being included in the features.

What Google says to do instead

The guide spends more room on this than on the myths, and it names the thing it wants:

"a first-hand review provides a unique perspective based on personal experience, whereas a summary of existing content simply restates information already available elsewhere."

Its example of content that adds nothing:

"Commodity content (for example, something like "7 Tips for First-Time Homebuyers") is often based on common knowledge, which could originate from anyone"

And its example of the opposite:

"non-commodity content (such as "Why We Waived the Inspection & Saved Money: A Look Inside the Sewer Line") provides unique expert or experienced takes that go beyond common knowledge and the ordinary."

Five tips about buying a house is commodity. The sewer line you actually looked at is not.

That is the entire strategy, and it is the same thing the helpful content documentation has said for three years. Nothing about it changes because the reader arrived through an AI answer.

How you measure any of it

Two things in Search Console, both worth knowing before you read anything else about this.

There is a Generative AI performance report, which shows how your content performs in generative AI features. It shipped in June 2026 and most people have not opened it.

And there is the inclusion requirement above, which you check in Search Console rather than assume.

What this site is doing about it

I have eight articles and every one of them is SEO. No llms.txt, because I checked the file Google publishes for itself and did not ship one for mine.

Checked against the five myths:

llms.txt                 none, deliberately
chunking                 nothing to undo
AI-specific rewriting    nothing to undo
fan-out chasing          no pages built for it
extra structured data    none added for AI
third-party tool         none used

Nothing on this site changed because of the guide, and that is the finding. I had already done the things that matter and none of the things that do not, because the things that do not are the ones being sold.

The one thing I will act on is the measurement. The report exists, this site is three days old, and it will say zero for a while. Reporting that number is more useful than a screenshot of a peak.

The short version

Google named AEO and GEO in a heading and answered both in one paragraph: it is still SEO.

The same guide lists five things you can ignore. llms.txt files, chunking your content, rewriting for AI systems, building pages for the follow-up queries AI generates, and adding structured data for AI specifically. Then it warns that no third-party tool can see Google's internal metrics.

One of those five is not just useless. Building pages to catch the extra queries is named as scaled content abuse.

What decides your visibility is the same thing that always did: a page in the index, eligible for a snippet, that says something nobody else has said.

Google's example of the difference is a list of homebuying tips against a story about the sewer line somebody actually inspected.

Sources

  • Google Search documentation, Optimizing your website for generative AI features on Google Search, May 2026
  • Google Search Central blog, the May 2026 announcement of that guide
  • Search Console help, the Generative AI performance report
  • Google Search documentation, creating helpful, reliable, people-first content
  • This site's own checks against all five items, 13 October 2026