The big picture: SEO is not dead, it just grew up
There has been no shortage of breathless predictions about the death of traditional SEO in the age of AI search. Vendors promising to “optimise for AI Overviews” have proliferated. New acronyms, AEO (Answer Engine Optimisation), GEO (Generative Engine Optimisation), have appeared in every marketing newsletter.
Google’s newly published guide cuts through the noise. Its position is unambiguous: generative AI features on Google Search are built on top of the same core ranking and quality systems that have always powered Search. The foundational advice has not changed. What has changed is how Google surfaces and presents that content to users.
From Google's perspective, optimising for generative AI search is optimising for the search experience — and thus still SEO.
The guide explains that AI Overviews and AI Mode rely on two underlying techniques: retrieval-augmented generation (RAG), which grounds AI responses in real, indexed web pages, and query fan-out, where the model generates a cluster of related sub-queries to build a fuller answer. Both processes draw directly from Google’s established search index and ranking signals.
What Google actually wants: non-commodity content
The most important strategic guidance in the document concerns content quality and it is more specific than the usual “write helpful content” advice. Google draws a sharp distinction between commodity content and non-commodity content.
Commodity content is defined as anything based on common knowledge that could have been written by anyone or, pointedly, by a generative AI model. The guide uses the example of “7 Tips for First-Time Homebuyers” to illustrate the type: well-formatted, technically accurate, and entirely forgettable.
Non-commodity content, by contrast, provides a genuinely unique perspective rooted in first-hand experience or deep expertise. Google’s own example is striking in its specificity: “Why We Waived the Inspection & Saved Money: A Look Inside the Sewer Line.” That is a real, lived experience rendered into content no AI could fabricate.
The practical implication is clear: if your content strategy relies on AI tools to generate articles that summarise what already exists, you are creating exactly the kind of commodity content Google’s systems are now better than ever at recognising and deprioritising.
Technical SEO: no reinvention required
On the technical side, the guide’s message is reassuring. Existing best practices remain fully relevant. Content must be crawlable and indexable, pages must offer a good user experience across all devices, and duplicate content should be reduced where possible.
The guide does add a nuanced note on semantic HTML: it is not required to be perfect, but well-structured markup helps users with assistive technologies navigate content and, importantly, it helps the emerging generation of AI browser agents parse and interact with your pages. This points forward to a world where AI agents book appointments, compare products, and gather information on users’ behalf, visiting websites as a kind of automated proxy for the human.
The myth-busting section is the most valuable part
Perhaps the most practically useful section of the guide is its explicit list of things you do not need to do for AI search. Several popular tactics are dismissed outright.
LLMS.txt files
Google says you do not need to create AI-specific text files or special machine-readable markup. Google may crawl them, but they receive no special treatment.
"Chunking" content
There is no requirement to break pages into small, discrete fragments for AI comprehension. Google’s systems understand nuance and multi-topic pages just fine.
Rewriting for AI systems
You do not need to adopt a special writing style, stuff in long-tail keyword variations, or capture every possible query permutation. AI models understand synonyms and intent.
Seeking inauthentic "mentions
Paying for or manufacturing brand mentions across blogs and forums is ineffective. Google’s ranking systems focus on genuinely high-quality content and actively filter spam.
Over-investing in structured data for AI
Schema.org markup is not required for AI Overviews or AI Mode. Continue using it for rich results eligibility, but do not treat it as an AI search lever.
The horizon: agentic experiences
The guide concludes with a forward-looking section on AI agents – autonomous systems that can browse the web, compare options, and complete tasks on users’ behalf. Google notes that browser agents analyse visual renderings, inspect DOM structure, and read accessibility trees. Websites that are semantically well-structured and accessible are, therefore, better positioned for this emerging traffic type.
Protocols such as the Universal Commerce Protocol (UCP) are specifically flagged as worth watching – early signals of infrastructure designed to allow Search agents to take actions on commerce sites directly.
This is not something most publishers need to act on immediately. But it is a useful reminder that the definition of a “website visitor” is in the process of expanding beyond the human reader.
What to do (and not do) right now
- Continue applying foundational SEO best practices. Nothing has been invalidated – it has only become more important.
- Audit your content for commodity vs. non-commodity split. Where can you bring genuine first-hand expertise or a unique point of view?
- Do not invest in LLMS.txt, content chunking, or AI-specific rewriting. Google says these are unnecessary.
- If you sell products or run a local business, ensure your Merchant Center and Google Business Profile data is current and complete.
- Keep semantic HTML and accessibility in good shape — not just for users, but for the AI agents beginning to browse on their behalf.
Is your content ready for AI search?
Keeping up with Google’s evolving search landscape is a full-time job. If you’re unsure how your site measures up against these new guidelines, or where to start, we can help. Our team audits content, technical structure, and visibility to make sure you’re not leaving traffic on the table.
