What Google actually says about AI visibility.
In May 2026, Google published its official guide to optimising for generative AI features in Search. It's direct, it busts several popular myths, and it has clear implications for any business trying to get found in AI-driven results. Here's what it says — and what it means in practice.
The headline: SEO is still the foundation
The most important thing Google's guide establishes — and it does so immediately — is that optimising for generative AI search is still SEO. Not a replacement for it. Not something separate that sits alongside it. The same foundation.
Google's AI Overviews and AI Mode are built on top of Google's existing Search index and ranking systems. The AI doesn't go off and find content independently — it retrieves content that Google has already crawled, indexed, and evaluated for quality. If your site isn't indexed, crawlable, and meeting Google's quality standards, it won't appear in AI-generated responses.
This matters because a significant amount of advice circulating about AI visibility — particularly around AEO and GEO — frames these disciplines as alternatives to SEO, or as separate things requiring entirely new strategies. Google's position is unambiguous: they are still SEO.
Google's words: "From Google Search's perspective, optimising for generative AI search is optimising for the search experience, and thus still SEO."
How Google's AI actually retrieves content
Google's guide explains two specific techniques its AI uses — and understanding them has practical implications for how you structure your website.
Retrieval-Augmented Generation (RAG)
When Google's AI generates a response, it doesn't just use its training data. It runs a retrieval step first — pulling relevant, up-to-date pages from the live Search index, then using those pages to inform and ground the response. This is why content quality and freshness still matter: the AI is reading your actual pages, not a static snapshot of the web from its training period.
The implication is direct: if your content is clear, accurate, and well-structured, the AI has reliable material to work with. If it's vague or out of date, the AI will favour a competitor whose content gives it more confidence.
Query fan-out
When a user asks a question, Google's AI doesn't just run that single query. It generates multiple related searches simultaneously to gather a broader picture before producing its response. A question about "how to improve my website's AI visibility" might generate fan-out queries around structured data, content clarity, entity signals, and search performance.
The practical implication here is significant: you don't need to target every possible search variation with a separate page. Content that covers a topic thoroughly — addressing the related questions and subtopics naturally — is more valuable than a collection of thin pages each targeting a single keyword variant.
"Google's AI doesn't retrieve content from a parallel universe. It retrieves content from the same Search index your SEO efforts are already targeting."
What Google says actually improves AI visibility
Google's guide is specific about what makes content perform well in generative AI features. These aren't vague aspirations — they're practical criteria with direct implications for how you build and write your website.
Non-commodity content with a unique point of view
Google draws a clear distinction between commodity content and non-commodity content. Commodity content is anything generic — content that could have been written by anyone, that simply restates what already exists elsewhere, or that an AI model could easily generate itself. Google gives a direct example: "7 Tips for First-Time Homebuyers" is commodity content. A first-hand account of waiving an inspection and the financial reasoning behind it is not.
For service businesses, this is a pointed challenge. Generic service descriptions, vague about pages, and blog posts written around broad topics without specific insight are exactly the kind of content Google's AI systems are least likely to select. Content that reflects genuine expertise, first-hand experience, and a specific point of view is what gets cited.
Content organised for human readers first
Google's guide is explicit: write for people, not for AI. Clear paragraphs, logical sections, descriptive headings, and a structure that helps a human reader navigate the page. This isn't in tension with AI visibility — it is AI visibility. AI systems extract information from pages more reliably when that information is clearly organised for a human reader.
Images and video where they add genuine value
Google's AI Overviews can surface images and video, not just text links. Pages that include relevant, well-labelled images and video have additional opportunities to appear. This isn't about adding stock photography for the sake of it — it's about supporting your content with visuals that genuinely add to it.
Technical foundations: crawlable, indexed, fast
Pages must be indexed and eligible to appear in Google Search to have any chance of appearing in AI-generated responses. This means clean crawlability, no accidental noindex tags, proper internal linking, and good page experience — particularly on mobile. Google is direct: meeting best practices doesn't guarantee indexing, but not meeting them guarantees exclusion.
What Google says you don't need to do
This is the most practically useful section of Google's guide. It names specific tactics that are widely promoted as AI visibility strategies — and states directly that they don't work for Google Search, or aren't required.
Google explicitly states you do not need to create llms.txt files, AI-specific text files, Markdown files, or new markup formats to appear in generative AI search. These are not factors in how Google's AI retrieves or evaluates content.
There is no requirement to break content into short, isolated fragments for AI to process. Google's systems can understand nuance across longer pages. Page length should be determined by what serves your audience — not by a theory about how AI prefers content to be formatted.
You don't need to write in a particular way aimed at AI. Google's AI understands synonyms, general meaning, and topic relevance. You don't need to capture every long-tail keyword variation or force specific phrasing into your content to satisfy AI requirements.
Paid mentions, low-quality link schemes, and manufactured citations are not effective strategies for AI visibility. Google's spam systems apply to generative AI features just as they do to standard Search. Authentic credibility signals — real reviews, genuine coverage, legitimate references — are what matter.
Google states directly that structured data is not required for generative AI search, and there is no special schema.org markup needed for AI features. Structured data remains useful as part of a broader SEO strategy — particularly for rich results eligibility — but it is not a direct lever for AI visibility in isolation.
What this means if you're a service business in the UK
Reading Google's guide through the lens of a UK service business, several practical implications stand out.
Generic website content is increasingly a liability
The distinction Google draws between commodity and non-commodity content applies directly to the homepage copy, service descriptions, and blog content that most service businesses have published. If your content sounds like it could have been written for any business in your sector, it's commodity content. It won't be selected by Google's AI over a competitor whose content is specific, experienced, and genuinely useful.
The technical basics still matter — and are often missing
Google's guide reaffirms that crawlability, indexability, and page experience are the prerequisite for any AI visibility at all. Many small business websites have technical issues that prevent pages from being properly indexed — and therefore prevent them from appearing in AI-generated responses regardless of how good the content is. An audit that starts with the technical layer is not excessive caution — it's necessary.
Local business signals remain important
Google specifically mentions that AI responses can include local business information, and that Google Business Profile plays a role in this. For service businesses operating in a defined area, ensuring your Business Profile is complete, accurate, and consistent with your website is part of the AI visibility picture — not separate from it.
Agentic search is coming — and good structure prepares you for it
Google's guide closes with a section on AI agents — autonomous systems that interact with websites directly, inspecting page structure, DOM elements, and accessibility trees to complete tasks on behalf of users. This isn't the main priority for most service businesses right now, but Google's advice for preparing for it is the same as good general web practice: clear navigation, accessible HTML, descriptive links, and clean contact flows. Building a well-structured site now means you're ready for this development when it becomes more relevant.
The bottom line from Google: Plenty of content thrives in AI search without any specific AI optimisation at all — because it's genuinely useful, clearly written, and properly indexed. That is the standard to aim for. Everything else is secondary.
How Web for AI's approach aligns with Google's guidance
Web for AI's approach to AI visibility has always been grounded in the same principles Google's guide now makes official. We don't sell AI-specific hacks, we don't recommend llms.txt as a meaningful strategy for Google Search, and we don't treat schema as a shortcut that bypasses the need for good content.
What we do is audit the real factors that determine whether a service business website is understood and recommended by AI-driven search: content clarity, entity signals, technical health, schema that confirms and supports visible content, and a structure that makes pages genuinely useful for human readers and AI systems alike.
Google's guide confirms that this is the right approach. The practical question for any business reading it is: does your website actually meet this standard? That's exactly what a free AI visibility audit is designed to establish.
Unique, non-commodity content
Content with a genuine point of view, first-hand experience, and specific insight — not recycled common knowledge that any AI could generate.
Clear technical structure
Crawlable, indexed pages with good page experience, clean URL structure, and no technical blockers preventing Google from accessing your content.
Human-first organisation
Content written and structured for people — clear paragraphs, descriptive headings, logical flow. AI readability follows directly from human readability.
Images and video where relevant
AI Overviews can surface visual content, creating additional visibility opportunities beyond text links for pages that support content with genuine media.
Local business signals
Google Business Profile, consistent NAP data, and local business schema — particularly important for service businesses operating in defined geographic areas.
Foundational SEO
Everything that has always mattered in SEO still matters. AI visibility on Google is built on top of the same index and quality systems — not separate from them.
Google's recommended priority order — applied to your website
Based directly on Google's guide, these are the areas worth addressing — in the order that makes the most practical difference.
Get the technical basics right
Confirm your pages are indexed, crawlable, and not accidentally blocked. Check for noindex tags, robots.txt issues, and page experience problems. Without these in order, nothing else matters.
Make your content genuinely useful and specific
Replace commodity content — generic service descriptions, vague about pages, thin blog posts — with content that reflects real expertise, specific experience, and genuine value for your audience.
Structure pages for human readers
Clear headings, logical sections, and a content hierarchy that makes it easy for a real person to find what they're looking for. This is also what makes pages easy for AI to extract from.
Keep local signals consistent
Ensure your Google Business Profile is complete and matches your website. Name, address, phone number, services, and service area should be consistent across every digital presence.
Use structured data to support, not substitute
Schema confirms what's already on your pages. Implement it where it adds clarity — Organisation, LocalBusiness, Service, FAQPage — but only when backed by visible, accurate page content.
Ignore the tactics Google says don't work
Don't spend time on llms.txt, content chunking, AI-specific rewrites, or inauthentic mentions. Google has explicitly said these are not effective for its AI features.