Google has released a new documentation page to help websites optimize generative AI features in search, including AI Insights And AI Fashion.
The page, “Optimizing Your Website for Generative AI Features in Google Search“, extends Google’s old strategy AI Feature Documentation released in 2025. The previous page explains how the AI features work, how inclusion is monitored, and how performance is reported. The new guide focuses more directly on optimization tips and tactics that Google says site owners may overlook.
Two sections are particularly worth highlighting. Google directly names popular optimization tactics that it says are unnecessary, and redefines the AEO/GEO conversation as part of standard SEO.
Google says AEO and GEO are “always SEO”
Google opens by confirming that fundamental SEO best practices remain relevant for AI generative search. Its AI capabilities are “rooted in our core ranking and search quality systems” and rely on retrieval augmented generation (RAG) and query distribution to surface search index content.
On the terminology debate, Google is direct. It defines “AEO” as “response engine optimization” and “GEO” as “generative engine optimization”, then states:
“From a Google Search perspective, AI generative search optimization comes down to optimizing the search experience, and therefore always SEO.”
This echoes positions taken by Google employees at conferences. Gary Illyes and Cherry Prommawin told Search Central Live attendees that GEO and AEO do not require separate frameworks. The position now appears in Google’s published documentation, providing an official reference for citation.
What Google Says You Don’t Need to Do
The guide includes a section “AI Generative Search to Bust Myths” listing tactics it deems unhelpful for Google Search. The guide is more explicit than the previous page on Google’s AI features, including naming llms.txt, chunking, inauthentic mentions, and directly AEO/GEO.
The guide states that site owners can ignore the following for Google Search.
On llms.txt files and other “special” tags, Google says you don’t need to create machine-readable files, AI text files, tags, or Markdown to appear in AI generative search. Google can discover and index many file types other than HTML, but that doesn’t mean those files get special treatment.
On “fragmentation“, the guide states that there is no need to break content into small chunks for AI systems. Google’s systems “are able to understand the nuance of multiple topics on a page and show the relevant element to users.” Danny Sullivan made similar comments in January 2026, saying he had spoken with Google engineers who recommended against chunking.
On rewriting content for AI systems, Google claims that AI systems can understand synonyms and general meanings. Site owners don’t need to capture every variation of long-tail keywords or write in a specific way for AI generative search.
In search of the inauthentic »mentioned“, the guide acknowledges that AI features can surface what is being said about products and services on blogs, videos and forums. But it says that searching for inauthentic mentions “is not as useful as it seems” because basic ranking systems focus on quality while other systems block spam.
On structured datathe guide states that this is not required for AI generative search and there is no special schema.org markup to add. He recommends continuing to use structured data as part of an overall SEO strategy for rich results eligibility.
Several recommendations run counter to the advice that appears in some AI search optimization guides. Multiple GEO Resources have promoted clustering and structured data as priorities for AI research visibility.
What Google says to focus on
The optimization tips follow familiar SEO territory, although Google contextualizes them for AI features.
Google places particular emphasis on “non-commercial content.” He contrasts commodity content (“7 Tips for First-Time Home Buyers”) with a non-market alternative (“Why We Forwent the Inspection and Saved Money: A Look Inside the Sewer Line”). The distinction is whether the content provides unique insight beyond common knowledge.
On a technical level, pages must be indexed and eligible for snippets to appear in generative AI features. Google recommends following crawling best practices, using semantic HTML where possible, following JavaScript SEO best practices, providing a good page experience, and reducing duplicate content.
Local and e-commerce optimization has its own section. Google recommends Merchant Center feeds and Google Business Profiles for visibility of products and local businesses in AI responses. He also mentions Business Agent, a conversational experience that lets customers chat with brands on Google Search.
Agent experiences receive initial guidance
A new section on agentic experiences describes AI agents as “autonomous systems that can perform tasks on behalf of people, such as making a reservation or comparing product specifications.”
Google notes that browser agents can access websites by analyzing screenshots, inspecting the DOM, and interpreting the accessibility tree. The guide links to web.dev guide to best practices for agent-friendly websites and refers to Universal Trade Protocol (UCP) as an emerging protocol that will “enable research agents to do more.”
Google announced the UCP earlier this yearand Vidhya Srinivasan’s annual letter states that it was co-developed with Shopify and that more than 20 companies have endorsed it.
Why it matters
This guide gives Google’s most explicit advice on what you should and shouldn’t do for generative AI features in search. It consolidates positions that were previously scattered across conferences, podcast appearances, and blog posts into a single reference.
The myth busting section carries the most weight. Google now tells you in its own documentation to avoid tactics promoted by a growing industry of AEO/GEO services. This doesn’t settle the debate for non-Google AI platforms like ChatGPT or Perplexity, which may weight signals differently. But for Google’s AI features, tips are now saved.
The Agent Experiences section places navigation agents and UCP in Google’s official documentation for site owners. The guidance is early and Google considers it optional for businesses where agent access is relevant.
Looking to the future
Google’s final section states that you don’t need to accomplish everything in the document to be successful. He notes that “a lot of content thrives in Google Search (including generative AI experiments) without any obvious SEO.”
Advice on agent experiences is labeled as something to explore “if it’s something that’s relevant to your business and you have the extra time.” This suggests that Google views agent optimization as forward-looking rather than urgent.
Featured Image: Anatolyr/Shutterstock
