Review Of Google’s Guide on AEO / GEO AI Optimization: What Actually Works (And What Doesn’t)

Google recently published an official guide on succeeding in generative AI search features – covering AI Overviews, AI Mode, and the broader evolution of Search. As someone who’s spent years in the SEO trenches, I’ve read through the guide carefully, cross-referenced it with independent research, and I want to give you a clear-eyed review of what it actually says, what it glosses over, and where the real opportunities lie for website owners right now.

The short version: Google’s foundational advice is solid. But there’s one area – structured data and schema markup – where the industry narrative has gotten dramatically out of hand, and the guide’s mild pushback doesn’t go nearly far enough. Let me explain why.

What Is AEO and GEO? Google’s Official Position

According to Google, AEO (Answer Engine Optimization) and GEO (Generative Engine Optimization) are both terms used to describe work focused on improving visibility in AI-powered search experiences. Google’s official stance is that these are not separate disciplines – optimizing for generative AI search is still fundamentally SEO. The same core ranking and quality systems that power traditional Search also power AI Overviews and AI Mode.

I largely agree with this. The panic-driven rebranding of SEO as “AEO” or “GEO” has created a cottage industry of consultants selling snake oil dressed up in AI buzzwords. The reality is messier and more interesting than any rebrand suggests.

Google explains that its generative AI features rely on two specific technical mechanisms:

• Retrieval-Augmented Generation (RAG): Google’s AI doesn’t make things up from scratch – it retrieves relevant, crawlable web pages from its Search index using core ranking systems, then synthesizes information from those pages into a response. This is also called “grounding.”
• Query Fan-Out: When someone searches for something broad, Google’s AI generates multiple related sub-queries simultaneously to gather more comprehensive information. A query like “how to fix a weedy lawn” spawns concurrent searches like “best herbicides for lawns” and “how to prevent weeds without chemicals.”

Understanding these two mechanisms is actually more useful for your AI optimization strategy than any AEO tactic I’ve seen promoted online. If your content doesn’t rank in regular Search, it won’t be retrieved for RAG. If your content doesn’t cover a topic with enough depth and breadth, it won’t satisfy fan-out queries. Those two facts alone should guide your entire content strategy.

The Foundational SEO Advice Google Gets Right

The guide’s core content recommendations are genuinely good, and they align closely with what I’ve observed working across dozens of client sites. Let me break down the most important points.

Non-Commodity Content Is the Single Biggest Lever

Google explicitly distinguishes between commodity content (generic, widely available knowledge like “7 Tips for First-Time Homebuyers”) and non-commodity content (unique, experience-driven insights like a first-person account of waiving a home inspection and what actually happened). AI systems are far more likely to cite non-commodity content because it provides information that can’t be sourced from hundreds of other identical pages.

This is something I’ve been saying to clients for years, and it’s satisfying to see Google state it plainly. The problem is that most websites are still producing commodity content at scale and wondering why AI Overviews never cite them. Your content needs a genuine point of view, real-world experience, or proprietary data to stand out in an AI-mediated search environment.

A few practical examples of what non-commodity content looks like in the wild:

• A contractor writing about a specific job gone wrong and what they learned – not “5 Tips to Hire a Contractor.”
• A financial advisor sharing what actually happened when a client followed conventional wisdom – not “How to Save for Retirement.”
• An SEO professional publishing original test results from their own experiments – not a rehash of a Moz article from three years ago

Technical SEO Remains Non-Negotiable

Google’s guide reaffirms that pages must be indexed and eligible to appear in Search with a snippet to even qualify for AI features. This means all your standard technical fundamentals still matter:

• Pages must be crawlable – AI models use publicly accessible content grounded in the Search index
• Page experience matters – mobile usability, load speed, and clear content hierarchy all contribute
• Duplicate content should be reduced – it wastes crawl budget and dilutes your authority signals
• JavaScript SEO best practices apply – if Google can’t render your content, it can’t cite it
• Search Console verification helps diagnose technical issues early

None of this is new, but it’s worth reiterating: you cannot shortcut your way into AI citations by adding a text file or tweaking markup. If your technical foundation is broken, nothing else matters.

Schema Markup and AI Citations: The Truth the Industry Doesn’t Want to Hear

This is the section I most want to focus on, because the structured data conversation in the SEO and GEO community has become genuinely misleading.

What Google Actually Says About Schema

Google’s official guide explicitly states that structured data is not required for generative AI search, and there is no special schema.org markup you need to add to appear in AI Overviews or AI Mode. The guide categorizes schema overemphasis as one of several myths to ignore. The only legitimate benefit Google acknowledges for structured data is eligibility for rich results in traditional Search.

That’s a significant statement from Google, and it runs directly counter to a wave of content circulating online that positions JSON-LD schema as some kind of AI citation unlock.

What the Ahrefs Study Actually Found

Ahrefs recently published one of the most methodologically rigorous studies on this question I’ve seen to date. Here’s what the data showed:

In a large-scale analysis of 6 million URLs, AI-cited pages were nearly three times more likely to have JSON-LD schema than non-cited pages. That sounds like a strong signal – until you look at the causal mechanism. The correlation exists because pages with schema tend to be on better-maintained, more authoritative websites with strong content and backlink profiles. The schema itself wasn’t doing the work. The site quality was.

To test causation directly, Ahrefs tracked 1,885 web pages that added JSON-LD schema and matched them against 4,000 control pages, then measured citation changes across Google AI Overviews, AI Mode, and ChatGPT. The results were clear:

• No meaningful uplift in citations was observed from adding schema
• AI Overview citations on treated pages actually fell by 4.6% relative to control
• The decline was statistically significant – roughly a 1 in 2,500 probability of occurring by chance
• Translated to real numbers, that’s approximately 12 fewer daily citations per page
• The results across ChatGPT and AI Mode showed either a small decline or statistically insignificant difference

To be fair, the Ahrefs team was careful to note that other factors – Google algorithm updates, content staleness, and recrawling frequency – may have contributed to the decline. The schema addition itself may not have caused harm. But the data makes one thing unambiguous: adding schema did not help AI citation rates at all.

My Take: Schema Is a Rich Snippet Tool, Not an AI Citation Tool

After running my own observations across client sites over time, I’m confident in saying this clearly: schema markup serves one primary purpose in modern Search, and that is eligibility for rich results – star ratings, how-to steps, product pricing in SERPs, and similar visual enhancements.

Rich results are valuable. I’m not dismissing schema entirely. But the use cases are actually quite narrow:

If your e-commerce site isn’t using Product schema, you’re leaving rich result eligibility on the table. If you’re a local business without LocalBusiness markup, fix that. But if someone is telling you that adding JSON-LD to your blog posts will get you cited in ChatGPT or Google AI Overviews, they’re selling you something that the data does not support.

What Actually Gets You Cited in AI Models

This is where I want to be direct, because the real answer is less exciting than a technical shortcut, but it’s what actually works.

Strong Core SEO Is the Foundation

AI citation in Google’s systems begins with the same process as traditional ranking. If your page isn’t indexed, isn’t trusted, and isn’t considered relevant by Google’s core systems, the AI won’t retrieve it via RAG. Citation probability is therefore a downstream function of your organic search authority, not a separate optimization track.

The pages that get cited in AI Overviews share common characteristics that have nothing to do with schema:

• They rank well organically for relevant queries
• They’re on domains with strong topical authority in their niche
• They contain direct, clear answers to specific questions
• They’re written for humans, not crawlers – well-structured, genuinely useful
• They cover topics with depth and nuance rather than surface-level summaries

Third-Party Signals and Brand Mentions Are More Powerful Than Any Markup

One thing I’ve consistently observed – and what the Ahrefs correlation data reinforces – is that sites with high citation rates aren’t distinguished by their technical markup. They’re distinguished by the quality and quantity of signals pointing to them from across the web.

This includes:

• Backlinks from authoritative domains: Links remain one of the strongest authority signals, and that authority directly influences which pages get retrieved in RAG-based systems
• Brand mentions and co-citations: When your brand is referenced across forums, review sites, industry publications, and social platforms, Google’s systems develop a stronger entity understanding of who you are and what you’re authoritative on
• User engagement signals: Pages that users actually read, share, and return to perform better across both traditional and AI-mediated search
• Content freshness and recrawl frequency: Pages that are actively maintained and updated are more likely to be fresh in the index when a relevant query triggers RAG retrieval
• E-E-A-T signals: Demonstrated expertise, experience, authoritativeness, and trustworthiness – including author bios, citations, and credentials – contribute to the quality signals that influence AI retrieval

The uncomfortable truth for people hoping for an easy AI optimization hack is that there isn’t one. The sites winning in AI Overviews are the same sites that have been winning in organic Search: authoritative, well-maintained, topically deep, and genuinely helpful.

The Myths Google Calls Out (And Why They Matter)

Google’s guide includes a mythbusting section that I found genuinely valuable. Let me add my own commentary to each point.

LLMS.txt Files: Not a Citation Signal

You don’t need to create llms.txt files or any special AI-readable markup to appear in generative AI search. Google is explicit about this. The file may be crawled, but it receives no special treatment. I’ve seen this promoted aggressively in some circles, and it’s a distraction from work that actually moves the needle.

Content “Chunking” for AI: Not Required

Breaking your content into tiny fragments to aid AI comprehension is unnecessary. Google’s systems are sophisticated enough to identify the relevant portion of a longer page and surface it in response to a specific query. Write for your readers. Use clear headings and structure. That’s it.

Rewriting Content in “AI-Friendly” Language: Counterproductive

This one genuinely frustrates me because I’ve seen it cause real damage. When writers twist their natural voice to match some imagined AI preference, the content becomes hollow and robotic – which is exactly the kind of content AI systems are designed to deprioritize. Write clearly and specifically for humans. The AI can handle synonyms and semantic meaning far better than most SEOs give it credit for.

Over-Focusing on Structured Data: A Red Herring for AI Goals

Google says it directly. The Ahrefs study confirms it empirically. My own observations support it. Schema is not an AI citation lever. Use it where it earns rich results. Don’t obsess over it for anything else.

What Google’s Guide Is Missing (In My Opinion)

The guide is well-written and refreshingly honest about what doesn’t work. But there are a few areas where it’s either vague or deliberately non-committal.

It Doesn’t Address the E-E-A-T Signal Gap for Smaller Sites

The advice to “create unique, valuable, non-commodity content” is correct but incomplete. A small website with excellent original content and no backlinks will consistently lose AI citations to a mediocre page on a high-authority domain. The guide doesn’t address this power imbalance honestly, and it should. Building the third-party signal profile that supports AI citation takes time and deliberate effort beyond content quality alone.

It’s Vague on How Fan-Out Queries Should Influence Content Strategy

The concept of query fan-out is introduced but not fully developed into actionable guidance. My practical interpretation: if you understand the cluster of related sub-queries that your core topic generates, you should be covering those sub-topics with depth on your site – not as thin, standalone pages, but as genuinely useful content that satisfies each sub-query independently. This is topical authority building, and it directly improves your surface area for RAG retrieval.

The Merchant Center and Local Business Section are Underemphasized

For e-commerce and local businesses, the guide’s recommendation to use Merchant Center and Google Business Profiles is buried at the end and treated almost as an afterthought. In my experience, these are among the highest-leverage actions for businesses in those categories – both for traditional Search and for generative AI responses that include product listings and local business information.

A Practical AI Optimization Framework That Actually Works

Based on Google’s guide, the Ahrefs research, and my own applied experience, here’s the framework I’d recommend for any website owner focused on improving AI citation rates:

1. Nail the technical foundation first. If you’re not indexed, nothing else matters. Use Search Console to identify and fix crawling and indexing issues before anything else.
2. Build genuine topical authority. Cover your niche with depth. Answer the questions your audience is actually asking. Go deeper than any competitor page. This expands your surface area for fan-out query retrieval.
3. Create content with a unique point of view. First-hand experience, original data, proprietary case studies, expert opinions – anything that can’t be replicated by a competitor running the same prompt through an AI content tool.
4. Build real third-party signals. Earn links from relevant, authoritative domains. Get your brand mentioned in publications your audience trusts. Build an entity profile that Google can understand and trust.
5. Use schema where it earns rich results. Product, LocalBusiness, Review, HowTo – apply structured data where it generates visible SERP enhancements. Don’t treat it as an AI optimization tactic, because it isn’t one.
6. Maintain your content actively. Stale content loses citation frequency. Pages that are updated, accurate, and recrawled regularly are better candidates for AI retrieval.
7. Optimize for human readability. Clear structure, useful headings, well-organized paragraphs – these aren’t just UX niceties. They directly influence how easily AI systems can identify and extract citable passages from your content.

The Bottom Line on Google’s AEO / GEO Guide

Google’s guide is one of the more honest and practically useful documents they’ve published on Search optimization in a while. The core message – that AEO and GEO are fundamentally SEO, that good content and technical fundamentals matter more than any AI-specific tactic, and that schema is not an AI citation shortcut – is correct and deserves to be amplified.

What I’d add from my own work and from the research: the sites that win in AI Overviews and AI Mode are not winning because of clever markup or workarounds. They’re winning because they’ve invested in genuine authority – topical depth, real-world expertise, legitimate backlinks, and brand presence that extends beyond their own domain.

Schema has its place. Use it for rich results. But if you’re spending energy on JSON-LD in hopes of unlocking AI citations, redirect that energy toward content that’s genuinely better than anything else in your niche and toward the third-party signals that tell Google you’re the authoritative source on the topics you cover. That’s the work that compounds.

Work With Someone Who Gets This Right

If you’re trying to navigate AI Overviews, traditional Search, and the intersection of both without wasting budget on tactics that don’t move the needle, I help businesses get this right at Affordable SEO Expert. My approach is grounded in what the data actually shows, not what’s being promoted in the latest AI optimization trend cycle. If you want a straight assessment of where your site stands and what will actually improve your visibility in AI search, reach out and let’s talk.

Frequently Asked Questions

Does schema markup improve your chances of appearing in Google AI Overviews?

No. Google’s official guide explicitly 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. An independent Ahrefs study tracking 1,885 pages that added JSON-LD schema found no meaningful uplift in AI citations – and actually observed a small decline in AI Overview citations relative to control pages. The correlation between schema and cited pages exists because authoritative sites tend to use both schema and strong SEO practices, not because schema itself drives citations.

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 targeting AI-powered search experiences. Google’s official position is that these are not distinct disciplines – they are extensions of SEO. Because Google’s AI features like AI Overviews use the same core ranking and indexing systems as traditional Search (via retrieval-augmented generation), the practices that improve organic Search visibility also improve AI citation rates.

What actually gets a page cited in AI Overviews and ChatGPT?

Pages cited in AI systems share several common characteristics: strong organic search rankings, high domain authority, genuine topical depth, clear and direct writing, and robust third-party signals including backlinks and brand mentions. AI citation is a downstream function of overall search authority. Pages that AI systems retrieve via RAG are pages that Google’s core ranking systems already trust and consider relevant – which means improving AI citation rates requires the same work as improving organic search performance.

Is it worth using schema markup at all in an AI-first search environment?

Yes, but with a realistic understanding of its purpose. Schema markup remains valuable, specifically for earning rich results in traditional Search – star ratings, product pricing, FAQ accordions, how-to steps, and similar SERP enhancements. These visual elements can improve click-through rates and user engagement. However, schema should not be prioritized as an AI citation strategy, as current research shows no evidence it influences AI retrieval or citation frequency.

How does query fan-out in Google’s AI affect content strategy?

Query fan-out means Google’s AI generates multiple related sub-queries to gather comprehensive information for a single user query. This has a direct implication for content strategy: websites with broad topical coverage across a niche have a larger surface area for AI retrieval. Rather than creating thin standalone pages targeting every keyword variation, the more effective approach is building genuine topical authority through in-depth content that satisfies the cluster of sub-queries naturally associated with your core topics. This improves both organic ranking and AI citation probability simultaneously.