Google just published its first official AI Optimization Guide. If you lead marketing at a company watching AI search start to chip away at your traffic, you should read it. Then read it a second time, with one question in mind: what isn't in here?
The guide is useful. It's also incomplete in a lot of ways that matter, and the gap is where the actual strategy work happens.
What Google confirmed
The headline finding is that SEO still runs the show. Google's AI Overviews and AI Mode pull from the same Search index, using retrieval-augmented generation (RAG) and query fan-out to assemble answers. Clean technical structure, semantic HTML, fast pages, and unique non-commodity content remain the foundation. Structured data is "not required" but still useful. None of that is surprising. All of it is correct.
For teams that have been building sites the right way, this validates the approach. Clean builds, solid information architecture, schema where it earns its place, and content written for people first. That's been WAIO Methodology's posture from day one, and Google just published the receipt.
A second piece worth noting: Google explicitly called out "AEO" and "GEO" as terms it considers a subset of SEO. From their seat at the table, that's a defensible position. From the rest of the market's seat, it's a narrower view than it looks.
Where the guide gets thin
The whole document is about Google. Specifically, it's about Google AI Overviews and AI Mode. That's one surface in a market that now includes ChatGPT, Claude, Perplexity, Gemini, and a growing list of agentic browsers and assistants. They don't all rank the way Google ranks. They don't all cite the way Google cites. And they don't all read your site the way Googlebot reads it.
Google can confidently tell you what works for Google. It cannot tell you what works for the half-dozen other AI systems your customers are already using to find answers. The minute you treat this guide as the complete map, you're optimizing for one surface in a world that has many.
The inauthentic mentions question
One of the mythbusting bullets in the guide is that "inauthentic mentions" don't work. Google's argument: ranking systems block spam, generative AI features rely on the same quality signals, so manufacturing brand mentions across the web won't change what gets cited.
Ethan Smith at Graphite pushed back on this on LinkedIn, and I think he's right. His read: LLMs would prefer that this doesn't work, but in practice it appears to. Whether or not Google wants you doing it, what marketers can observe on their own sites suggests influence-the-corpus tactics still produce results. Google's stated policy and the observable behavior of these systems aren't the same thing.
That's a useful pattern to apply across the entire guide. Treat official guidance as a strong signal of intent. Then test the systems themselves to see actual behavior.
Agent trust is the new domain authority
A pattern keeps showing up across independent datasets. BrightEdge's 16-month study found that Google AI Overview citations overlap with traditional organic rankings 54% of the time, growing from 32% at rollout. B2B Tech sits at 71%, healthcare at 75%, education at 73%. The trust-dependent verticals lead. Ramp ran their own experiment tracking over 1,300 AI bot visits from Claude, Perplexity, and ChatGPT, and found AI models repeatedly cite the pages they already trust. Cleanly structured, markdown-friendly content beat over-engineered schema designed to "hack" AI.
Translation: the foundation of getting cited by AI is the same foundation that's been getting you ranked for years. Domain authority, content depth, technical hygiene, and the slow compounding of being a useful site. New language, same job.
If your site has organic authority today, the work is keeping the technical surface clean and the content unique enough to deserve citation. If it doesn't, you're building both at once, and AI search doesn't shortcut that build.
What we're watching in practice
What we're seeing across sites moving through WAIO Methodology lines up with these patterns. AI citations follow the same fundamentals: clean structure, content depth, and technical hygiene that holds across all three audiences a modern site has to serve, people, search engines, and AI systems.
This is the part Google's guide can't really tell you, because it would read like Google saying "do the boring fundamentals well." The boring fundamentals are the strategy. The AEO and GEO conversation has been so noisy about new tactics, new file formats, new schemas, that the quiet truth keeps getting buried: the sites winning AI real estate today are the ones that built well in the first place.
What to do this week
- Read Google's AI Optimization Guide end to end. It's the new baseline, and your team should be working from the same baseline.
- Audit your top 10 organic pages against semantic HTML, structured data, and content uniqueness. That's where AI citations will compound first.
- Don't rebuild your site for AI. Build it for people, then make sure search engines and AI systems can read what you built.
If you want a faster diagnostic, run a free audit on your homepage with WAIO Engine. It scores your site against the three audiences this post is built on. Takes about a minute. No email gate before the result.
For the full thinking behind the three audiences and why the boring fundamentals compound, read WAIO Methodology.
