Share this post:
On May 15, 2026, Google published something the search industry has been begging for and dreading in equal measure: an official guide on how to optimise your website for generative AI features in Google Search. The document covers AI Overviews, AI Mode, and the rest of Google’s expanding generative surface. And buried inside it is a section called “Mythbusting generative AI search” that has set SEO Twitter (sorry, X) on fire.
Google took direct aim at several tactics that an entire cottage industry has been selling under the banners of GEO (Generative Engine Optimisation) and AEO (Answer Engine Optimisation). No, you don’t need llms.txt files. No, you don’t need to “chunk” your content into tiny blocks. No, you don’t need special schema markup for AI search. And no, you don’t need to rewrite your pages in some AI-friendly language. Google’s exact words: “From Google Search’s perspective, optimising for generative AI search is optimising for the search experience, and thus still SEO.”
That sentence alone caused a minor earthquake. Some celebrated it. Lily Ray called it vindication. Pedro Dias posted “I guess the GEO bros will not be happy.” Mike King of iPullRank fired back within 72 hours, calling the entire guide “naive and self-serving.” The industry, as usual, split into camps. And somewhere in the middle of all of it, CMOs and marketing directors are trying to figure out what this means for their 2026 budgets.
Below: what Google said, where the guide gets it right, where it might be misleading, and what you should change (or keep doing) as a result.
What Google’s Guide Says (And Why It Matters)
The guide lives on Google’s Search Central documentation under a new “Generative AI fundamentals” section. It consolidates advice Google employees have scattered across conference talks, podcasts, and blog posts over the past 18 months into a single reference document. For anyone who has been tracking the breadcrumb trail of Gary Illyes’ comments, Danny Sullivan’s tweets, and Cherry Prommawin’s conference slides, most of the content will feel familiar. The difference is that it’s now official documentation you can point a client to when they ask why you aren’t buying an llms.txt generator tool.
Google’s core argument is straightforward. Their generative AI features are built on top of the same ranking and quality systems that power traditional search results. They use retrieval-augmented generation (or RAG, which Google also calls “grounding”) to pull content from the existing search index and then generate a summarised answer with citation links. A second mechanism called “query fan-out” fires multiple related sub-queries at the same time to assemble more complete answers. If someone searches for “how to fix a lawn full of weeds,” the system might run parallel retrievals for herbicide recommendations, chemical-free removal methods, and prevention strategies.
The practical implication: there is no separate AI index. Your content either ranks in Google’s search index and can be retrieved for AI answers, or it doesn’t exist to the generative system at all. If your pages are already performing in traditional search, you have a shot at appearing in AI Overviews and AI Mode. If they aren’t, no amount of GEO tricks will get you there on Google’s platform.
The guide then breaks into three broad areas of advice. First, create content that is unique, compelling, and non-commodity. Google draws a clear distinction between generic “7 Tips for First-Time Homebuyers” content and original perspective pieces like “Why We Waived the Inspection and Saved Money.” If your content could have been written by anyone with a search bar and fifteen minutes, AI systems have no reason to prefer it. Second, follow established technical SEO best practices (crawlability, semantic HTML, good page experience, reduced duplicate content). Third, consider local search, e-commerce, image, and video optimisation where relevant, since AI features surface these differently.
None of that is controversial. The controversy lives in the mythbusting section.
The Mythbusting Section: Five Tactics Google Says You Can Ignore
Google’s mythbusting section names five specific practices the company says you don’t need for their generative AI features. Each one targets something that GEO and AEO service providers have been actively selling.
LLMs.txt files and special markup. Google says you don’t need machine-readable files, AI text files, or Markdown versions of your pages to appear in generative AI search. Google can crawl and index many file types beyond HTML, but that doesn’t mean any of them receive special treatment inside the generative pipeline. Some SEO platforms have been charging monthly fees for “GEO file generation” tools. For Google’s ecosystem, that investment buys you nothing.
“Chunking” content. The idea here was that you should break long-form content into tiny, self-contained blocks so AI systems can extract answers more cleanly. Google says their systems can already understand the nuance of multiple topics on a single page and pull out the relevant piece for users. There is no ideal page length. Danny Sullivan reinforced this in January 2026, saying engineers recommended against chunking.
Rewriting content for AI systems. You don’t need to write in a particular way for generative AI search. Google’s AI systems understand synonyms and general meanings, so you don’t need to obsess over capturing every possible long-tail keyword variation. Write for your readers, not for a language model’s expected input format.
Seeking inauthentic “mentions.” Google acknowledges that their AI features can surface what’s being said about products and services across blogs, videos, and forum discussions. But chasing fake mentions across the web to inflate your brand’s AI visibility won’t work the way some agencies promise. Google’s core ranking systems focus on quality content, and their spam systems catch manufactured signals. If someone is selling you a “get mentioned in 50 blogs” package specifically for AI search visibility, take a hard look at what you’re paying for.
Overfocusing on structured data. Structured data isn’t required for generative AI search, and there’s no special schema.org markup that unlocks AI visibility. Google recommends keeping structured data as part of your broader SEO strategy because it helps with rich results in traditional search, but it’s not the AI search lever that some guides have been positioning it as.
The cumulative message is blunt: if you’ve been paying for GEO-specific tooling and tactics that revolve around these five areas, Google is telling you those investments don’t move the needle on their platform.
Where Google Gets It Right
For Google’s own ecosystem, the guide lands several punches that the industry needed to hear. The obsession with llms.txt files was always questionable for Google specifically. The spec was designed with platforms like Anthropic’s Claude in mind, and while it may have value for those systems, treating it as a Google AI Overviews lever was a stretch from day one. The same applies to aggressive content chunking. If you were restructuring perfectly good long-form articles into tiny fragments purely because a GEO course told you to, you were probably hurting readability for humans while gaining nothing from Google’s AI.
The warning about inauthentic mentions also deserves attention. A segment of the GEO industry has been selling “AI mention building” packages that look suspiciously like the old-school link schemes from 2012, rebranded with AI terminology. Google linking this practice to their spam detection systems is a signal worth taking seriously. Getting your brand mentioned in genuine, high-quality editorial contexts still has value. Paying for bulk placement across low-quality blogs to game AI search is a different game entirely, and one that Google is watching.
The technical SEO fundamentals section is also solid ground. Pages need to be indexed and eligible for snippets to appear in generative features. Crawlability still matters because generative AI models use the search index as their source material. None of this should surprise anyone who has been paying attention, but it’s worth repeating because the GEO hype cycle created an illusion that some new, separate set of rules had replaced the basics. It hadn’t. Not on Google.
Where the Guide Falls Short (and Where the Industry Pushback Gets Interesting)
Mike King published his response three days after the guide dropped, and his critique raises points that marketers shouldn’t dismiss even if they agree with Google’s broader direction.
King’s central argument is that “it’s still SEO” is a strategically convenient position for Google, not a technically accurate one. SEO as a discipline has spent years struggling to expand its scope inside organisations, fighting for influence over content strategy, product decisions, and technical architecture. By collapsing all AI search optimisation back into “it’s just SEO,” Google keeps practitioners focused on their own platform rather than developing cross-platform expertise that would spread attention and budget toward competitors like ChatGPT, Perplexity, and Bing Copilot.
He also challenges the chunking dismissal with technical depth. King points to Google’s own MUVERA research, their work on passage indexing, and their patents on pairwise passage selection. If Google’s retrieval systems operate at the passage level (and their own research suggests they do), then how content is structured at the passage level does matter, even if Google doesn’t want you thinking about it in “chunking” terms. Bing’s documentation is more candid here. Microsoft has published that their systems require “chunking/transformations [that] preserve meaning and claims used in the answer.” That’s a different message from Google’s “just write normally” guidance.
Then there’s the multi-platform problem. Google’s guide is explicitly about Google’s generative AI features. It says nothing about ChatGPT, Perplexity, Claude, or any other AI search surface. And those platforms don’t all work the same way. ChatGPT’s browse feature, Perplexity’s citation engine, and Bing’s Copilot each have their own retrieval methods and quality signals. Optimising for Google’s AI features alone may be enough if Google search drives the bulk of your pipeline. But for many B2B SaaS companies, Perplexity and ChatGPT have become meaningful referral sources, and ignoring those surfaces because “Google says GEO is just SEO” could leave visibility on the table.
Microsoft’s Bing team, for their part, has taken a noticeably different approach. In February 2026, they launched “AI Performance” inside Bing Webmaster Tools, giving publishers the first real dashboard for tracking citation frequency, page-level activity, and grounding query phrases across Copilot and Bing’s AI features. While Google has been telling publishers “don’t worry about it, just do good SEO,” Microsoft has been building the measurement infrastructure that treats AI search as a distinct channel worth tracking. The contrast is hard to miss.
What This Means for Your Strategy in 2026
So where does this leave you if you’re a marketing leader trying to allocate budget and attention?
Google’s AI Features: Trust the Fundamentals
For Google’s ecosystem specifically, the guide is a credible reference point. Your investment should go toward content quality, technical SEO health, and building genuine brand authority. If your pages rank well in traditional Google search, you are already in the running for AI Overviews and AI Mode. Focus your energy on creating content that offers a genuine perspective rather than regurgitating the same commodity information that fifteen competitors have already published. Google’s own distinction between commodity and non-commodity content is the most important takeaway from the entire guide. First-hand experience, original data, expert analysis, and a distinctive point of view are the signals that make AI systems want to cite your page over someone else’s.
If you’ve been paying for llms.txt generation tools, chunking services, or AI-specific schema packages purely for Google visibility, you can redirect that spend toward stronger content and technical foundations with confidence.
Beyond Google: Multi-Engine Visibility Still Matters
The guide’s biggest blind spot is everything outside Google. ChatGPT, Perplexity, Claude, and Bing Copilot each represent growing channels for discovery, and they don’t all follow Google’s playbook. The BrainZ Digital team has seen this across client portfolios: sites gaining meaningful referral traffic from ChatGPT browse, Perplexity generating citations that convert, and Bing’s share in B2B SaaS search journeys creeping upward quarter over quarter. None of that is captured by Google’s advice to “just do SEO.”
For multi-engine visibility, the work looks different. Your brand’s entity footprint across the web (Wikipedia, Reddit, third-party publications, review platforms) influences how LLMs outside Google understand and recommend you. Your content’s clarity and structure at the passage level affects how retrieval systems extract and cite your information. Your presence in authoritative “best of” lists and comparison content shapes which brands AI systems surface when users ask for recommendations. This work overlaps with traditional SEO, yes. But it also extends beyond what any single search engine’s optimisation guide will tell you to do.
The Measurement Gap Isn’t Going Away
Google doesn’t offer anything like Bing’s AI Performance dashboard. You can’t log into Google Search Console and see how often your pages were cited in AI Overviews or AI Mode. That measurement gap means you need to build your own monitoring approach, whether that’s through regular prompt testing across AI platforms, tracking referral traffic from AI sources in your analytics, or working with tools that are starting to offer AI citation tracking.
The fact that Google doesn’t provide this data doesn’t mean AI search visibility is unmeasurable. It means the measurement responsibility falls on you and your team, or the agency you work with.
Don’t Chase Tactics. Build an AI-Proof Brand.
The real lesson underneath all the industry drama is one that predates Google’s guide: brands that are genuinely authoritative, consistently helpful, and structurally clear in how they present information will perform across every search surface, generative or otherwise. Whether Google calls it SEO or GEO or something else entirely doesn’t change the mechanics of earning trust from systems designed to find and surface trustworthy answers.
If your content is strong enough that a human expert would cite it in their own work, AI systems will follow. If your brand has enough genuine presence across the web that a human researcher would encounter it while doing a thorough review, language models will have learned from those signals too. The tactical details of how each platform’s retrieval pipeline works matter less than whether your brand has earned the right to be recommended in the first place.
Conclusion: Google Said the Quiet Part Out Loud
Google’s new guide is the most direct, consolidated piece of AI search guidance the company has ever published. It confirms what experienced SEOs have been saying for over a year: there is no magic GEO formula that replaces doing the fundamentals well. For anyone who was buying into the hype around llms.txt files, content chunking hacks, and bulk mention-building as the keys to AI search domination, this guide is a cold bucket of reality. Google’s AI features pull from the same index, use the same quality signals, and reward the same content principles as traditional search. That part of the message is clear, and it’s worth internalising.
But Google’s perspective is Google’s perspective. The company has an obvious interest in keeping SEO practitioners focused on its platform, and the guide’s silence on multi-engine optimisation is telling. If your strategy for 2026 stops at “do good SEO for Google and the AI features will follow,” you might be leaving significant visibility gaps across ChatGPT, Perplexity, and Bing’s AI surfaces, especially in B2B categories where those platforms are gaining traction fast.
The BrainZ Digital approach has always been to optimise for understanding, not for algorithms. You build your brand’s entity clarity so that AI systems across every platform can correctly interpret who you are, what you offer, and why you’re credible. You create content designed for humans that also happens to be structured in a way that retrieval systems can parse and cite. You maintain a consistent brand footprint across the channels and platforms where AI models learn about the world. And you measure your visibility across the full landscape, not just the slice that one search engine reports on.
Google’s mythbusting section killed a few bad tactics that deserved to die. The paid llms.txt tools, the aggressive chunking gimmicks, the spray-and-pray mention building. Good riddance. But the guide also tried to collapse an expanding, multi-surface search landscape back into a single-platform frame. That’s the part you should push back on. The brands that win over the next 12 months will be the ones that take Google’s quality advice seriously while recognising that optimising for one platform is no longer enough. Google gave you the foundation. Building the full structure across every AI search surface is your job, and it starts with understanding that GEO isn’t a separate discipline from SEO. It’s what SEO grows into when search stops being a list of links and becomes a conversation that happens everywhere.
If you want assistance with your organic and SEO strategy, we are here for you! You can read more about our AI SEO services here, or contact us directly to learn how we can best support you in reaching your business goals.
