Google Just Told the Entire AEO Industry It's Selling You Tools You Don't Need

Google Just Told the Entire AEO Industry It's Selling You Tools You Don't Need

Google published an official guide to optimizing for AI-powered search — AI Overviews, AI Mode, the whole generative layer sitting on top of regular search results. Buried in the middle of it, under a section literally called "Mythbusting," Google said something that should worry every agency currently selling "AEO audits" and "llms.txt optimization" as a service: you don't need most of it.

No special markup. No llms.txt files. No chunking your content into bite-sized AI-friendly pieces. No obsessing over schema beyond what you'd already do for regular SEO. Google's own words: "many suggested 'hacks' aren't effective or supported by how Google Search actually works."

We read the whole guide, not the press summary. We also happen to track our own AI-visibility score for kiraco.org — a real number, not a claim — across ChatGPT, Perplexity, Gemini, and Claude, not just Google. So we're publishing this two ways: what Google actually said, and what we're doing about it on a site that starts from a real, unflattering baseline.

What "AI Search" Actually Runs On, According to Google

Strip away the AEO/GEO branding and Google's answer is almost boring: AI Overviews and AI Mode are built on the same ranking systems as regular Search, plus two additional mechanisms.

The first is retrieval-augmented generation, or RAG — what Google calls "grounding." It's the process of pulling relevant, up-to-date pages from Google's existing search index to ground an AI-generated answer in real content, rather than letting the model answer from memory alone. The second is query fan-out: when someone asks a broad question, the AI model quietly generates several related sub-queries behind the scenes, runs all of them, and synthesizes the results. A single question to an AI Overview might trigger five or six real searches you never see.

The practical implication: if your page isn't eligible to rank in ordinary Google Search — indexed, crawlable, meeting basic technical requirements — it cannot appear in an AI Overview either. There's no separate "AI index" to optimize for. It's the same index, with an AI-written summary layered on top for some queries.

The Six Things Google Says You Can Stop Worrying About

This is the part worth sitting with, because it directly contradicts a lot of AEO advice currently being sold:

Special AI markup or llms.txt files. Google's exact line: "Google Search itself doesn't use them." One real nuance worth preserving, because Google states it explicitly: llms.txt files aren't harmful, and they may still matter for other AI systems that do read them — Google is only saying it personally ignores the file. If your AI-visibility goal includes ChatGPT or Perplexity, that's a different battlefield with different rules.

Chunking content into tiny AI-digestible pieces. "Google systems are able to understand the nuance of multiple topics on a page and show the relevant piece to users." There's no ideal page length. A long, thorough page and a short, focused one can both work.

Rewriting your content specifically for AI systems. Google's models understand synonyms and intent, so you don't need to manually cram in every phrasing variant of a question. Writing for a human reader remains writing for the AI.

Chasing inauthentic brand mentions. Buying placements or seeding fake mentions across low-quality sites doesn't help — Google's ranking systems filter for quality first, and spam-detection systems run independently of the generative layer.

Treating structured data as mandatory. Schema markup isn't required for AI features specifically. It still earns you rich results in regular Search, so it's worth doing — just not because of some AI-specific rule.

Producing content at scale to cover every query variation. Google calls this a violation of its scaled-content-abuse policy, and separately, just not effective — more pages doesn't mean more relevance.

What Actually Moves the Needle

Google's real answer is almost anticlimactic: the same things that have always mattered for good SEO.

The clearest instruction in the whole guide is the distinction between commodity and non-commodity content. Google's own example is worth quoting exactly: "commodity content (for example, something like '7 Tips for First-Time Homebuyers') is often based on common knowledge... In contrast, non-commodity content (such as 'Why We Waived the Inspection & Saved Money: A Look Inside the Sewer Line') provides unique expert or experienced takes."

Translate that for most businesses and it means: a generic listicle a competitor could publish in an afternoon is commodity content, no matter how well-optimized. A specific result, a specific number, a specific decision you actually made and can defend — that's the non-commodity version, and Google says it influences AI visibility "more than any of the other suggestions in this guide."

Quick answer: what should I actually do to rank in AI Overviews and AI Mode?
- Make sure your pages are indexed and crawlable in regular Google Search first — nothing else matters if this fails.
- Write content built on real experience or original data, not a summary of what's already out there.
- Organize pages with clear headings and paragraph structure — for readers, not for an AI parser.
- Use schema markup as part of normal SEO hygiene, not as an AI-specific hack.
- Skip llms.txt, content chunking, and keyword-variant stuffing for Google specifically — they do nothing there.
- Verify your site in Google Search Console so you can actually see what's being indexed.

Where Google's Guide Stops and the Real Work Starts

Here's the honest gap: Google's guide is scoped to Google Search. It doesn't cover ChatGPT, Perplexity, Claude, or Gemini's standalone app — and for most businesses today, "does the AI recommend me" means all of those, not just Google's AI Overview.

We track our own visibility across all of them, and the number is humbling: as of our June 2026 baseline, kiraco.org scores 13.5 out of 100 on our own AI Visibility Index, with a 0% mention rate across tracked queries and a 0.9% share of voice against competitors in our category. We're publishing that number because Google's guide made a point we agree with: don't fake authority you don't have yet. A real, unflattering baseline is more useful than a polished claim with nothing behind it.

The gap between "what Google says matters for its own AI Overviews" and "what actually gets an AI assistant to recommend your brand" is where the real work lives:

Original data gets cited. Every AI system that grounds its answers in real content will cite a source with actual numbers over one with opinions. Publishing a real measurement — even an unflattering one like ours — is more citable than a polished claim.

Reddit and Quora carry disproportionate weight with Perplexity and Google AI Overviews specifically, because both platforms are heavily represented in the web index these models draw from. This has nothing to do with your own website's technical structure.

Entity clarity matters more than most companies realize. If your brand name collides with something else online, or you don't have a clean, consistent presence across Wikidata, LinkedIn, and Crunchbase, AI models struggle to disambiguate who you actually are — no amount of on-page optimization fixes that.

Comparison and "alternative to" pages get cited constantly, because they answer the exact kind of question people ask an AI assistant directly: "what's an alternative to X."

None of this contradicts Google's guide. It extends it, because Google Search is one surface among several a business now needs to show up on.

What to Actually Do This Week

Start with the boring part, because it's also the correct part: verify your site in Google Search Console and confirm your important pages are actually indexed. Nothing downstream matters if this step fails.

Then pick one piece of genuinely non-commodity content — a specific result, a specific number, a decision you can defend with a reason — and write it the way Google's guide describes: organized, readable, built on real experience. Skip the llms.txt file for Google's sake; keep one anyway if you want visibility with other AI assistants, since it costs nothing and doesn't hurt.

If you're a business trying to be recommended by name — "what's a good WhatsApp AI provider," "best AI marketing agency" — measure where you actually stand before optimizing anything. We didn't publish our 13.5 because it's good. We published it because guessing your AI visibility is worse than knowing it and being honest about the number.

The Uncomfortable Part Nobody Selling AEO Services Wants You to Read

Google just handed away the industry's whole premise for free, in a guide anyone can read in fifteen minutes. There's no proprietary AEO framework Google is hiding from you. There's the same SEO fundamentals that have mattered for two decades, plus a genuine, harder frontier — original data, entity clarity, off-site presence — that no amount of on-page markup solves.

We're building toward that second part in public, starting from a real number instead of a claim. If you want to see what an AI-native WhatsApp platform built around that same principle looks like, see what we've built or read more about KIRA. Talk to Us on WhatsApp

Source: Google's Guide to Optimizing for Generative AI Features on Google Search