Last updated 15 May 2026, Google quietly dropped something worth paying attention to: an official guide for optimizing your site for generative AI features in Search, AI Overviews, AI Mode, basically an overview on the whole thing.
If you’ve been doing SEO for any length of time, you’ll read parts of it and think “…that’s it?” And honestly, that’s the point.
So what are AEO and GEO, exactly?
You’ve probably seen these terms bouncing around LinkedIn and Instagram. AEO (Answer Engine Optimization) and GEO (Generative Engine Optimization) are labels that have sprung up to describe work specifically aimed at getting visibility in AI-generated search results rather than traditional blue links.
Google’s take? From Google Search’s perspective, optimizing for generative AI search is just optimizing for the search experience and therefore still SEO.
They’re not wrong. The entire guide essentially reframes existing SEO practice for an AI context rather than introducing a separate discipline. Whether that’s reassuring or anticlimactic probably depends on how much money you’ve spent on AEO consultants lately.
How Google’s AI features actually work
Before getting into recommendations, it’s worth understanding what’s actually happening under the hood.
Google’s generative AI features rely on two core techniques. The first is retrieval-augmented generation (RAG), a method for improving the accuracy and freshness of AI responses by pulling relevant, up-to-date pages from the Search index, then generating a response grounded in that content. The second is query fan-out, where the model generates a set of concurrent related queries to gather more information and address the user’s original question more fully. For example, a query like “how to fix a lawn full of weeds” might fan out into “best herbicides for lawns,” “remove weeds without chemicals,” and “how to prevent weeds in lawn.”
This matters because it tells you what the AI is actually doing: it’s not a separate system with different rules. It’s retrieving from the same index, using the same quality signals, and then synthesizing. Your SEO fundamentals affect your eligibility before the AI even enters the picture.
What Google actually wants you to do
Write content that a real person couldn’t easily Google elsewhere
Google describes the distinction between commodity and non-commodity content with a concrete example. “7 Tips for First-Time Homebuyers” is commodity content. It offers common knowledge, no unique insight and could have come from anyone. “Why We Waived the Inspection & Saved Money: A Look Inside the Sewer Line” is non-commodity content. This is a specific, experienced take that goes beyond what a quick search would surface.
The bar they’re setting is: does your content offer something a generative AI model couldn’t easily produce itself? First-hand experience, expert perspective, a specific angle this what they want. If you’re summarising what others have already written, you’re producing exactly the kind of content the AI is trying to replace, not surface.
Google also flags a common mistake: creating pages targeting every variation of how someone might search for a topic (essentially trying to chase fan-out queries) primarily to manipulate rankings. This violates their scaled content abuse policy, and they’re explicit that a high volume of pages doesn’t make a site more relevant or higher quality.
Worth filing that away the next time someone pitches you a scaled content. Note pSEO (programmatic SEO) when done right can get results but when it’s poorly implemented it can lead to boom-and-bust traffic.
Keep your technical house in order
Nothing surprising here. A page needs to be indexed and eligible to appear in Google Search with a snippet before it can appear in generative AI features at all. So crawlability, indexation, page experience, these all still matters, for the same reasons it always has.
On semantic HTML, Google’s position is reasonable: it’s not required to have perfectly valid HTML, but using semantic markup where possible helps other users like screen readers parse and navigate your page more easily. Write for humans, and the technical side largely takes care of itself.
The mythbusting section (this is the useful bit)
This is the part of the guide that should save you both time and money. Google lists specific tactics that are being circulated online from “gurus” and people who give SEO a bad name, and tells you plainly that they don’t work. Here’s a summary:
llms.txt files
You don’t need to create new machine-readable files, AI text files, special markup, or Markdown files to appear in generative AI search. Google may discover and crawl many kinds of files on a site, but that doesn’t mean they’re treated in any special way.
This one was making the rounds for a while, especially on LinkedIn. Add an llms.txt to your site so AI systems know how to read you. This is false and this myth needs to be nipped in the bud.
“Chunking” content
Another popular theory that has been disproven. There’s no requirement to break your content into small pieces for AI to understand it. Google’s systems can understand nuance across multiple topics on a single page and show the relevant section to users. There’s no ideal page length (again another debate amongst SEO experts). Just write for your audience, not for generative AI search.
Rewriting content for AI systems
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 looking for, and can connect users with content that doesn’t use the exact same words. You don’t need to worry about capturing every long-tail keyword variation. This is the same as how Google knows to group terms like “best” and “top-rated”.
Paid or artificial “mentions”
Google’s generative AI features can surface what’s being said about products and services across blogs, videos, and forum discussions, but seeking inauthentic mentions isn’t as helpful as it might seem. Their core ranking systems focus on high-quality content while other systems block spam, and the generative AI features depend on both.
Over-investing in structured data
Structured data isn’t required for generative AI search, and there’s no special schema.org markup you need to add. It’s still worth using as part of your overall SEO strategy because it helps with eligibility for rich results but it’s not a lever for AI visibility specifically. Interestingly, Google has announced they have deprecated FAQSchema indicating this could be the trend moving forward.
The agentic piece
This section is early-stage but worth noting. AI agents are autonomous systems that can perform tasks on behalf of people, such as booking a reservation, comparing product specs. You are used to chatting in conversations with the likes of ChatGPT, Claude and Gemini but browser agents are becoming more common, and may access your website to gather data by analyzing visual renderings like screenshots, inspecting the DOM structure, and interpreting the accessibility tree.
Google points to an agent-friendly website best practices guide and mentions emerging protocols like Universal Commerce Protocol (UCP) that will allow Search agents to do more.
There’s nothing actionable for most sites right now, but this is where things are heading. Accessibility-first development and clean DOM structure will matter more as agents become a bigger part of how people interact with search results.
What to take away from all of this
We have been waiting to hear Google’s official take on this. And it could change again in the next couple of year but. But with this document, that Google published, they want to stop people wasting time on tactics that don’t work. The “mythbusting” section reads less like guidance and more like a cease and desist on bad advice that you see on social media and shady groups trying to make a quick buck.
For most sites, the actual to-do list is short:
- Make sure your pages are crawlable and indexed first
- Write content that draws on genuine experience or expertise, not just research
- Stop chasing AI-specific hacks that Google has explicitly said don’t move anything
Your competitors who have sites doing well in AI Overviews are generally the same sites that were doing well before AI Overviews. The fundamentals didn’t change, the presentation of results did. Instead of the list of results on Google, there is now a range of options.
If your content is thin, generic, or exists primarily to rank rather than to help someone, that was already a problem. The AI era just makes it more obvious.
Not sure where your site stands?
If you’re unsure whether your content is pulling its weight in Search, whether traditional or AI, that’s exactly what we look at. PrimaryRush is a digital agency focused on SEO and content for small and mid-sized businesses. We do the kind of work this guide is actually describing: auditing what’s there, fixing what’s broken, and building content that’s worth ranking.
If you want a straight answer on where your site is at, get in touch.