Google has finally published its official stance on AEO, GEO, and what website owners should actually be doing to show up in generative AI search experiences like AI Overviews and AI Mode. For anyone who has been wading through the noise of AI optimization theories, vendor pitches, and half-baked “strategies” floating around the SEO community, the guide is a breath of fresh air – and, honestly, a bit of a gut punch to some very popular advice.

We’ve read the guide in full, cross-referenced it against our own testing, and looked closely at the recent Ahrefs study analyzing schema markup’s actual effect on AI citations across 6 million URLs. What we found reinforces what we’ve suspected for a while: most of what’s being sold as “AI optimization” is either redundant with good SEO, completely ineffective, or actively misleading.

Here’s our full breakdown of Google’s guidance, what it actually means in practice, and where we agree, disagree, or have additional data to bring to the conversation.

What Google Is Actually Saying About AEO and GEO

Google’s position is that AEO (answer engine optimization) and GEO (generative engine optimization) are not separate disciplines from SEO – they are SEO, reframed for a generative AI context. The systems powering AI Overviews, AI Mode, and other generative features are rooted in Google’s core Search ranking and quality systems. Optimizing for them means doing foundational SEO well, not chasing new AI-specific tactics.

This is important because there is an entire cottage industry built around convincing brands that AI search requires a whole new playbook. Google is directly saying: it doesn’t. The same signals that make your content rank well in traditional Search – authority, relevance, quality, crawlability – are the same signals that determine whether your content is cited in AI responses.

The two core mechanisms Google describes are worth understanding:

• Retrieval-Augmented Generation (RAG): AI responses are grounded in real web pages retrieved from Google’s index. The model doesn’t just make things up – it fetches relevant pages and uses them to generate responses. If your page isn’t indexed and ranking, it can’t be retrieved.
• Query Fan-Out: When someone asks a broad question, Google’s AI generates multiple related sub-queries to gather more complete information. This means a single well-written, in-depth page can be cited across multiple related queries – not just the exact one it was “optimized for.”

Both mechanisms run through Google’s existing Search infrastructure. There is no separate AI index. There is no separate AI ranking system. What gets you cited is what gets you ranked.

Schema Markup and AI Citations: What Google Says vs. What the Data Shows

Google’s guide explicitly states that structured data is not required for generative AI search, and there is no special schema.org markup that will get you cited in AI Overviews or AI Mode. The primary value of schema remains what it has always been: eligibility for rich results in traditional Search. The Ahrefs study confirms this – adding JSON-LD schema to pages produced no meaningful uplift in AI citations and may have correlated with a slight decline.

This is one of the most practically important points in Google’s guide, and it cuts against a lot of advice being circulated right now. Let’s be specific about what the evidence actually shows.

The Ahrefs Study: Schema Doesn’t Move the Needle on AI Citations

Ahrefs conducted a large-scale study analyzing 6 million URLs to understand the relationship between schema markup and AI citation rates. At first glance, the data looked promising for schema advocates: pages cited by AI were nearly three times more likely to have JSON-LD markup than non-cited pages. That correlation, however, is almost certainly a proxy for site quality, not a causal relationship.

Sites that implement structured data tend to be better maintained, more technically sound, and more authoritative overall. Of course they get cited more – but schema isn’t why.

To test causation, Ahrefs tracked 1,885 pages that added JSON-LD schema and matched them against 4,000 control pages that didn’t. They then measured citation changes across Google AI Overviews, AI Mode, and ChatGPT. The result? No meaningful uplift. AI Overview citations on treated pages actually fell by 4.6% relative to control – a statistically significant decline, not a gain.

Now, Ahrefs is careful to note that this could reflect other factors: Google algorithm updates, content staleness, infrequent recrawling. But the data is clear that adding schema did not produce more AI citations. We’ve seen similar patterns in our own testing. Schema simply isn’t the lever people think it is for AI visibility.

So What Is Schema Actually Good For?

Schema markup has a legitimate and valuable role in SEO – just a narrower one than many AI optimization vendors would have you believe. Its primary benefit is rich result eligibility in traditional Google Search: review stars, FAQ dropdowns, product pricing, recipe cards, event listings, and similar enhanced SERP features.

These rich results can meaningfully improve click-through rates for certain content types. That’s real value. But it’s a different value proposition from AI citation, and conflating the two leads to misallocated effort.

“Schema is a tool for earning real estate in traditional Search results. It is not a signal that tells AI systems to trust or cite your content. Treating it as an AI optimization tactic is a category error.”

If you have schema implemented correctly for your content type – product, article, local business, FAQ – keep it. It still serves a purpose. Just don’t expect it to change your AI citation rates.

What Actually Gets You Cited in AI Responses

AI citations are primarily driven by the same factors that drive traditional Search rankings: content quality, topical authority, backlink profile, E-E-A-T signals, and technical crawlability. Third-party signals – links, mentions, reviews, discussions across the web – are strong predictors of AI citation likelihood. Pages that rank well tend to get cited. Pages with genuine authority in their niche tend to get cited. There is no shortcut that bypasses this.

Here’s what we consistently observe as the actual drivers of AI citation:

• Strong organic rankings: If your page is already ranking in positions 1–5 for a query, it is significantly more likely to appear in AI Overviews and AI Mode responses for related prompts. RAG retrieves from the index – ranking matters.
• Backlink authority: Pages with strong link profiles from authoritative, relevant domains are more likely to be retrieved and surfaced. This hasn’t changed.
• Content depth and specificity: Shallow content that summarizes what everyone else has already said gets passed over. AI systems are increasingly good at identifying genuine informational value versus regurgitated content.
• Third-party mentions and brand signals: Discussions in forums, independent reviews, citations in other articles, and mentions across diverse sources all contribute to how AI systems perceive entity authority. Google’s guide even notes that its generative AI features “can show what’s being said about products and services across the web.”
• First-hand experience and unique perspective: Google specifically calls out first-hand reviews and personal experience as examples of the kind of unique viewpoint that AI systems look for. Aggregated opinions and generic advice don’t stand out.

The Role of E-E-A-T in Generative AI Search

Google’s framework of Experience, Expertise, Authoritativeness, and Trustworthiness (E-E-A-T) runs directly through its generative AI features. This isn’t a separate consideration – it’s baked into the quality signals that determine what gets retrieved and cited.

What this means practically: content attributed to credible authors with demonstrated expertise, published on sites with strong domain authority, and corroborated by external sources performs better in AI responses. Anonymous content on low-authority domains does not, regardless of how well it’s structured or how much schema markup it carries.

Google’s Official Mythbusting: What You Can Stop Worrying About

One of the most useful parts of Google’s guide is its direct debunking of popular AI optimization myths. We appreciate the directness here, because a lot of the practices being sold to site owners right now are somewhere between ineffective and counterproductive.

LLMS.txt Files

Google is explicit: you do not need to create an LLMS.txt file or any other machine-readable file to appear in generative AI search. Google may discover and crawl such files, but they receive no special treatment. The LLMS.txt concept has been aggressively marketed as a way to “tell AI what to do with your site” – this is not how Google’s systems work.

“Chunking” Content for AI

Google’s systems can understand nuance across an entire page and surface the relevant portion in response to a specific query. There is no requirement to break content into smaller pieces for AI comprehension. Page length should be determined by audience needs and content requirements – not by assumptions about AI parsing behavior.

Rewriting Content “Just for AI”

AI systems understand synonyms, related concepts, and contextual meaning. You don’t need to obsessively target long-tail keyword variations or rewrite content in a specific AI-friendly style. Write clearly for humans. The AI can handle the rest.

Overfocusing on Structured Data

Google says it plainly: “Structured data isn’t required for generative AI search, and there’s no special schema.org markup you need to add.” The Ahrefs data backs this up. Schema has its place, but AI citation is not it.

What Google’s Guide Actually Recommends: The Real AEO / GEO Strategy

Strip away the mythbusting and the technical explanations, and Google’s actual recommendations are admirably straightforward. Here’s what the guide emphasizes:

1. Create Content That Is Genuinely Non-Commodity

Google’s language here is worth quoting directly from the guide: contrast “7 Tips for First-Time Homebuyers” (commodity) with “Why We Waived the Inspection & Saved Money: A Look Inside the Sewer Line” (non-commodity). The distinction is real. AI systems are trained to surface information that goes beyond what’s already everywhere. If your content could have been written by anyone with 20 minutes and a search engine, it probably won’t be cited by AI.

Non-commodity content has:

• A specific, defensible point of view
• First-hand experience or expertise that can’t be easily replicated
• Information that adds to the conversation rather than restating it
• Depth that rewards the reader for spending time with it

2. Maintain Technical SEO Fundamentals

Crawlability, indexability, page experience, and snippet eligibility are prerequisites for AI citation. A page that isn’t indexed cannot be retrieved by RAG. A page that is blocked from crawling cannot be used to ground AI responses. Technical SEO isn’t glamorous, but it’s the foundation everything else sits on.

Google specifically calls out:

• Meeting Search technical requirements for indexing and snippet eligibility
• Following crawl budget optimization practices for large sites
• Proper JavaScript SEO implementation
• Good page experience across devices
• Reducing duplicate content

3. Build Real Authority Through Real Signals

The guide acknowledges that quality links, genuine third-party mentions, and brand presence across the web all contribute to how AI systems evaluate and surface content. This aligns with everything we know about how Google’s core ranking systems work. The path to AI citation runs through the same authority-building work that serious SEO has always required.

4. Use Merchant Center and Google Business Profile Where Applicable

For e-commerce and local businesses, Google’s AI responses can include product listings and business information pulled from structured data sources like Merchant Center feeds and Google Business Profiles. This is a concrete, actionable area where structured inputs genuinely do affect AI visibility – though it’s distinct from schema markup on web pages.

Our Take: AEO and GEO Are Legitimate Framings of an Old Problem

We don’t think the terms AEO and GEO are useless – they’re useful for framing conversations with clients and stakeholders about where search is headed and what content strategy needs to prioritize. But as distinct technical disciplines requiring new tactics, they’re largely overstated.

The websites getting cited most frequently in AI Overviews and ChatGPT responses aren’t the ones that added LLMS.txt files or implemented special AI schema. They’re the ones with strong content, real topical authority, solid technical SEO, and meaningful backlink profiles. The same sites that were winning in traditional Search are largely winning in generative AI Search – because the underlying retrieval systems are the same.

“The uncomfortable truth about AEO and GEO is that the best strategy for AI visibility is also the best strategy for traditional SEO. There is no separate game to win. There is only one game, and it rewards the same things it always has: genuine authority, original content, and technical excellence.”

Where we think the AI framing does add value is in content strategy: specifically, the emphasis on directness, depth, and unique perspective. AI systems are particularly good at identifying and surfacing content that directly answers specific questions with specificity and authority. Writing with that intent – clear, direct, well-organized, genuinely useful – is good for AI retrieval and good for human readers. It’s good content strategy, full stop.

Practical Recommendations: What to Actually Do

Based on Google’s guide, the Ahrefs schema study, and our own ongoing analysis, here is what we recommend for any site owner or SEO practitioner looking to improve visibility in generative AI search:

Do These Things

• Invest in content that reflects genuine first-hand experience, expertise, or original research
• Build backlinks from authoritative, relevant domains – this remains the strongest external signal
• Ensure every important page is properly indexed and eligible for snippets
• Optimize for page experience, especially on mobile
• Use schema markup where it earns rich results – for product pages, local business listings, articles, FAQs – because those rich results still matter for traditional Search CTR
• Build brand presence through legitimate PR, editorial mentions, forum contributions, and community engagement
• Create content that directly answers specific questions with depth and authority
• Verify your site in Search Console and fix technical issues proactively

Stop Wasting Time On These

• LLMS.txt files and similar “AI signal” files
• Adding schema markup specifically to improve AI citation rates
• Creating thin pages targeting every possible long-tail variation of a query
• Rewriting content in a specific “AI-friendly” format or style
• Chunking content into short segments based on assumptions about AI comprehension

Comparison: Schema Markup vs. Real AI Optimization Factors

Frequently Asked Questions

Does schema markup help you appear in Google AI Overviews or AI Mode?

No. Google’s own guide states that structured data is not required for generative AI search and that there is no special schema.org markup needed to appear in AI Overviews or AI Mode. The Ahrefs study, which tracked 1,885 pages that added JSON-LD schema against 4,000 control pages, found no meaningful uplift in AI citations and observed a statistically significant 4.6% relative decline in AI Overview citations on schema-treated pages. Schema’s primary value remains its contribution to rich result eligibility in traditional Search.

What is the difference between AEO, GEO, and SEO?

AEO (answer engine optimization) and GEO (generative engine optimization) are terms used to describe optimization work focused on visibility in AI search experiences. Google’s official position is that both are subsets of SEO, not separate disciplines. Because Google’s generative AI features are built on its core Search ranking and quality systems – including RAG and query fan-out – the same foundational SEO practices that drive traditional Search rankings are what drive AI citation likelihood. There is no separate technical playbook for AEO or GEO when it comes to Google Search.

What actually determines whether a page gets cited in Google AI Overviews?

Pages that get cited in AI Overviews tend to share the same characteristics as pages that rank well in traditional Search: strong backlink authority, high E-E-A-T signals, genuine content depth, and proper technical indexation. Google’s generative AI features use Retrieval-Augmented Generation (RAG), which pulls from the existing Search index. If a page isn’t indexed, it can’t be retrieved. If it lacks authority, it won’t be prioritized. Schema markup, LLMS.txt files, and content reformatting strategies do not meaningfully affect citation rates.

Do LLMS.txt files help Google’s AI find and cite your content?

No. Google has explicitly stated that LLMS.txt files and similar machine-readable files receive no special treatment in Google Search or its generative AI features. Google may discover and crawl such files, but they are not processed differently from any other file type and do not influence AI citation rates or Search visibility. Creating LLMS.txt files is not a recommended AI optimization strategy for Google Search.

Is it worth creating separate content pages targeting AI fan-out queries?

Google directly addresses this and advises against it. While query fan-out is a real mechanism – where the AI generates related sub-queries to build a more complete response – creating separate thin pages targeting every possible fan-out query variation is considered scaled content abuse under Google’s spam policies. More importantly, it’s ineffective. A high quantity of low-quality pages doesn’t increase site authority or citation likelihood. A single authoritative, in-depth page can be cited across multiple related queries without the need for separate targeting pages.

The Bottom Line on Google’s AEO / GEO Guide

Google’s guide on AI optimization is one of the more useful pieces of official Search documentation published recently – not because it reveals anything dramatically new, but because it cuts through an enormous amount of noise and gives practitioners a clear, authoritative baseline to work from.

The core message is consistent with what serious SEOs have observed: the foundations haven’t changed. Content quality, topical authority, technical excellence, and genuine third-party signals are what drive visibility in both traditional Search and generative AI experiences. Schema markup has a legitimate role for rich result eligibility, but it isn’t an AI citation strategy. LLMS.txt files, AI-specific content rewrites, and chunking tactics are not supported by how Google’s systems actually work.

What has changed – and this is worth taking seriously – is the bar for content quality. AI systems are increasingly good at identifying genuine informational value versus content that recycles what’s already everywhere. Non-commodity content, first-hand expertise, and unique perspective matter more now than they did two years ago. That’s the real adaptation that AEO and GEO demand.

At Marketing 1on1, our approach to AI search optimization is grounded in exactly this: building real authority through strong content, legitimate link acquisition, and technical SEO execution that ensures content is indexed, crawlable, and eligible for citation. If you want to work with a team that operates from evidence rather than speculation, we’d be glad to talk.