Google just published its official guidance on how to show up in AI search — the AI Overviews and AI Mode results that now sit at the top of so many queries. It's the closest thing we have to a rulebook from the company that owns the game.
Most marketers haven't read it. Understandably — it's written for site owners and developers. So I read the whole thing and translated it into plain English for the people who actually run marketing budgets, not the ones writing meta tags.
Here's the short version: there is no separate AI playbook. In Google's own words, optimising for AI search is optimising for Search. It's still SEO. Everything else below is a consequence of that one idea.
First, how AI search actually works
You can't optimise for something you don't understand, so two terms are worth knowing — both straight from Google's guide.
Retrieval-augmented generation (RAG), which Google also calls “grounding.” Instead of inventing an answer, the AI pulls relevant, up-to-date pages from Google's existing Search index, reads them, and generates a response with clickable links back to those sources. The pages it pulls are the ones your normal ranking already surfaced.
Query fan-out. When you ask a question, the model quietly fires off a set of related searches around it — ask “how to fix a lawn full of weeds” and it also searches “best herbicides for lawns,” “remove weeds without chemicals,” and so on — then assembles an answer from all of them.
The practical implication of both: AI search is not a new front door to your website. It's the same Search index, repackaged. If you rank, you're in the running. If you don't, no amount of “AI optimisation” gets you in.
The 10 takeaways
1. There's no separate AI playbook
This is the headline, and Google states it directly: from Search's perspective, optimising for generative AI search is optimising for the search experience — and that's still SEO. “AEO” (answer engine optimisation) and “GEO” (generative engine optimisation) describe the same job under new names. If a vendor is selling GEO as a brand-new discipline, be sceptical.
2. AI Overviews pull from the same index
Because AI features are rooted in Google's core ranking and quality systems, the work that earns rankings is the work that earns AI citations. You're not optimising two different systems. You're optimising one.
3. Unique, non-commodity content wins — more than anything else
Google says this plainly: creating content people find unique, compelling and useful will influence your presence in AI search “more than any of the other suggestions in this guide.” Commodity content — “7 Tips for First-Time Homebuyers,” the stuff anyone could write and a model could generate — adds little. A first-hand review, a real point of view, expertise that goes beyond common knowledge: that's what gets cited.
4. Write for humans, not models
Clear structure, headings, paragraphs, good writing. Google is explicit that you don't need to rewrite content in some special way “for AI” — its systems understand synonyms and meaning, so you don't have to stuff every long-tail variation of a phrase. Write the page your reader would enjoy and find satisfying. That's the whole test.
5. Stop paying for llms.txt and “special” AI files
One of the clearest myth-busts in the guide: you do not need to create new machine-readable files, “AI text files” (like llms.txt), markup, or Markdown to appear in AI search. Google may crawl many file types, but that doesn't mean they're treated specially. If part of your budget is going here, you can stop today.
6. Stop “chunking” your content
There's no requirement to break your pages into tiny pieces so AI can “understand” them better. Google's systems already handle multiple topics on a page and surface the relevant part. There is no ideal page length. Build the page for your audience, not for the model.
7. Structured data isn't an AI hack
Schema markup is not required for AI search, and there's no magic schema.org type that gets you into AI Overviews. Keep using structured data — it genuinely helps you qualify for rich results in regular Search — but use it for that reason, not because someone promised it's your way into the AI answer.
8. Chasing “mentions” doesn't work
There's a cottage industry around getting your brand “mentioned” across the web to influence AI. Google addresses it head-on: seeking inauthentic mentions isn't as useful as it seems. Spam systems block the manufactured stuff; quality systems reward the real thing. Earn coverage by being worth covering — don't game it.
9. The boring technical basics still decide eligibility
To appear in AI Overviews or AI Mode, a page has to be indexed and eligible to show in Search with a snippet. That means the fundamentals still gate everything: crawlable, meets Search's technical requirements, fast, good page experience, not buried behind broken JavaScript, minimal duplicate content. Unglamorous, and still the price of entry.
10. Get ready for agentic search
The one genuinely new frontier in the guide. AI agents — autonomous systems that book, compare and buy on a user's behalf — are starting to visit websites directly. They read your DOM, your accessibility tree, sometimes screenshots. Accessible, semantic, well-structured pages are how you prepare, and the good news is it's the same hygiene that helps human users and screen readers. Emerging standards like the Universal Commerce Protocol (UCP) point to where this is heading.
What this means for your budget
Zoom out and the guide is quietly telling marketing leaders where to stop spending: llms.txt files, content-chunking exercises, schema-as-a-silver-bullet, and mention-farming services. None of it moves the needle in Google's AI features, and some of it edges into spam-policy territory.
Where to spend instead is equally clear: unique, experience-led content that a model couldn't generate, solid technical foundations, and content organised for the humans you actually want as customers. Less exciting than a new acronym. It also happens to be what works — in AI search and in Search overall.
At Saigon Digital this is exactly how we approach AI search and SEO: no separate “GEO package,” just the fundamentals done properly, aimed at the people with the credit card. If your current agency's AI strategy is a list of the hacks above, it might be time for a second opinion.
The question worth asking your team
When did your team last read Google's actual guidance — instead of a hot take about it? The guide is free, it's short, and it quietly contradicts a lot of what's being sold as cutting-edge right now.
Sources & further reading
- Google Search Central — Optimising your website for generative AI features on Google Search (the official guide): developers.google.com/search/…/ai-optimization-guide
- Google Search Central — AI Features and Your Website: developers.google.com/search/docs/appearance/ai-features
- Google Search Central — Creating helpful, reliable, people-first content: developers.google.com/search/…/creating-helpful-content
