Google Says AEO and GEO Are Still SEO — What the Official Guide Means (and What It Doesn't)

On May 15, 2026, Google published its first official guide for optimizing content for generative AI features in Search. The headline conclusion — repeated in the documentation, the announcement blog, and coverage from Search Engine Journal — is direct:

"From Google Search's perspective, optimizing for generative AI search is optimizing for the search experience, and thus still SEO."

Google named Answer Engine Optimization (AEO) and Generative Engine Optimization (GEO) by name, then placed both inside standard SEO. There is no separate AI index. No special file to upload. No AI-specific schema that unlocks citation slots. AI Overviews and AI Mode run on the same crawl, the same quality systems, and the same ranking signals as classic Google Search.

For Google surfaces, the GEO consulting gold rush just got an official rebuttal. But the guide's scope is Google-only — and what it leaves out matters as much as what it says.

What Google officially recommends

The guide, published at developers.google.com/search/docs/fundamentals/ai-optimization-guide, covers six areas:

1. Foundational SEO remains the foundation.

"AI Overviews and AI Mode are rooted in our core Search ranking and quality systems." Crawlability, site structure, E-E-A-T, and content quality that earns organic rankings are the same signals that earn AI citations on Google.

2. Create valuable, unique, non-commodity content.

Google emphasizes original expertise, not reformatted summaries of what already exists. Content that adds genuine value — original research, first-hand experience, unique data — is what generative features select.

3. Provide local, shopping, image, and video content.

Multimodal and structured content types (Business Profiles, product feeds, YouTube, images) feed the surfaces Google's AI features increasingly cite — especially after the self-citation shift toward Google-hosted cards. See Google AI Mode's Self-Citation Shift.

4. Monitor performance through Search Console.

Google Search Console now reports AI Overview and AI Mode appearance data. This is Google's measurement layer for its own surfaces.

5. Apply SEO best practices to generative AI search.

The guide explicitly states you do not need to accomplish everything in the document to succeed — "plenty of content thrives in Google Search (including generative AI experiences) without any overt SEO at all."

6. Initial guidance on AI agents.

A quickly evolving section on agentic search — acknowledging that the space is moving faster than documentation can follow.

What Google explicitly rejects

The guide's mythbusting section names tactics the AEO/GEO industry has promoted — and calls them unnecessary for Google Search:

This is the strongest official statement yet against the formatting-checklist branch of GEO advice. Google is telling site owners: do not add new things for AI. Keep making original, high-quality content under normal SEO best practices.

The July 2026 GEO evidence review by Olivier Martinez reached a compatible conclusion from the research side: generic heuristics transfer poorly, relevance and context position are the most reproducible levers, and body-only rewrites can reduce retrieval presence by 16% (SAGEO Arena). See GEO in 2026: What the Evidence Review Says.

What the guide gets right

There is no separate Google AI index. This confirms what practitioners suspected: AI Overviews, AI Mode, and Gemini chat all ground on the Google Search index. Your Search Console data, sitemap submissions, and organic ranking efforts feed all three surfaces.

Quality signals transfer. The same E-E-A-T, crawlability, and content depth that earn blue-link rankings earn AI citations — at least as a starting point. Cyrus Shepard's evidence-weighted analysis placed search rank (9.4) and URL accessibility (9.5) as the top predictors of AI citation.

Cosmetic GEO tactics are unnecessary for Google. If your AEO program consists of adding statistics to paragraphs, restructuring content into "AI-friendly chunks," or creating llms.txt files — Google says stop. That effort is better spent on original content, earned media distribution, and technical crawlability.

The citation pool extends beyond page one. While the guide frames optimization as SEO, the data shows AI citations increasingly come from outside the top 10. Ahrefs found only 37.9% of AI Overview citations from top-10 pages (down from 76%). Google's fan-out retrieval pulls from a wider pool than the primary SERP. Good SEO gets you indexed; depth and breadth across sub-queries gets you cited.

What the guide does not address

Google's guide is authoritative for Google. It is silent on everything else — and the everything else is where most AI discovery is growing.

Non-Google engines operate on different backends

ChatGPT retrieves through Bing and OAI-SearchBot. Claude uses Brave Search. Perplexity uses its own index. None of these read Google's index. Google's statement that "GEO is still SEO" applies to Google surfaces only.

A team that reads this guide and concludes "we only need SEO" will be invisible on ChatGPT (70% of citations from outside both Google and Bing top 10), partially visible on Claude (Brave backend), and dependent on Perplexity's freshness engine — while correctly optimized for Google.

PrimeAIcenter's analysis of the guide put it directly: "The guide deliberately doesn't address: everything outside Google." Optimizing for Google alone is a Google play. Multi-platform AEO is a different discipline.

Cross-engine overlap is structurally low

Only 11% of domains are cited by both ChatGPT and Perplexity (CiteMetrix, 680M citations). Only 2.7% appear across all five major engines (SurfacedBy, 127K citations). Google's SEO excellence does not transfer to ChatGPT's Bing-weighted retrieval or Claude's Brave-backed search.

Earned media drives 84% of citations

Google's guide focuses on what site owners control — their own content, structure, and technical setup. It does not address the finding that 84% of AI citations come from earned media: journalism, third-party editorial, review platforms, and community content (Muck Rack, 25M links). Your website is the minority citation path.

Citation ≠ mention ≠ traffic

Google's Search Console reports AI appearance data, but the guide does not address the ghost citation problem — 62% of AI citations link to a source without naming the brand (Semrush/Indig). Nor does it address the traffic evidence: AI Overviews cut outbound clicks by 39.8% in a randomized study, and AI-referred visitors convert differently than organic visitors.

Volatility requires repeated measurement

Google's guide implies stable optimization — do good SEO and appear in AI answers. The 2026 volatility data shows otherwise: 73.5% of cited URLs appear once and never return (Trakkr), median citation lifespan is 11–15 days (Writesonic), and only 13.2% of brands hold steady presence over 90 days. Single-metric reporting misleads.

The correct reading: Google SEO is necessary, not sufficient

The guide should be read as a floor, not a ceiling.

For Google surfaces (AI Overviews, AI Mode, Gemini): Yes, GEO is SEO. Invest in crawlability, content quality, structured data, earned media, and Search Console monitoring. Skip llms.txt, chunking, and AI-voice rewrites.

For the full AI discovery landscape: Google SEO is one input among four retrieval backends. You also need:

1. Bing indexing for ChatGPT (Bing Webmaster Tools, IndexNow, OAI-SearchBot access)
2. Brave discoverability for Claude (no publisher tool — earn ranking through content quality)
3. Perplexity freshness (PerplexityBot access, 60–90 day refresh cycles)
4. Parallel monitoring across all engines on the same buyer prompts

Google's guide killed GEO as a Google-specific discipline. It did not kill GEO as a multi-platform practice. The term remains useful — even essential — for everything outside Google's index.

What to do now

If you have been doing "GEO hacks" for Google

Stop. Google named llms.txt, chunking, AI-specific schema, and inauthentic mentions as unnecessary or harmful. Redirect that effort toward:

• Original research and unique expertise
• Earned media distribution (target 4+ third-party platforms)
• Technical crawlability and site structure
• YouTube, local, and product content for Google-hosted AI surfaces

If you have been ignoring non-Google engines

Start. Google's guide confirms your Google strategy should be rooted in SEO. But ChatGPT processes 72 billion messages per month (Search Engine Land). Users under 44 average five search platforms (EMARKETER). Your buyers are not only on Google.

Build parallel monitoring for ChatGPT, Claude, Perplexity, and Google AI on the same prompt set. Report engine-by-engine.

If you are an agency or consultant

Use Google's guide to reset client expectations on the Google side: no separate AI index, no magic files, no formatting tricks. Then expand the conversation to multi-platform visibility — because that is where the growth and the gaps are.

How Obsurfable fits

Obsurfable implements what Google's guide cannot: parallel monitoring across ChatGPT, Claude, Gemini, Perplexity, and Google AI surfaces. When Google's guide says "do SEO," Obsurfable shows whether that SEO is actually producing citations and mentions on Google — and whether the same content is visible on the three non-Google backends your buyers also use.

The Visibility Director flags gaps across engines and drafts content for the specific retrieval backend where visibility is missing.

For the infrastructure behind non-Google engines, see Why Each AI Engine Reads a Different Search Index. For Google's own AI citation shifts, see The Gemini 3 Citation Reset.

FAQ

Did Google kill the GEO industry?

Google killed GEO as a Google-specific discipline with separate tactics. Multi-platform AEO/GEO — optimizing for ChatGPT, Claude, Perplexity alongside Google — remains essential and is not addressed by the guide.

Should I delete my llms.txt file?

Google says it is unnecessary for Google Search. It does not hurt, but the effort is better spent elsewhere. No major non-Google engine officially supports llms.txt either.

Does "still SEO" mean I only need Google Search Console?

For Google AI surfaces, Search Console is your primary measurement tool. For ChatGPT, you need Bing Webmaster Tools (which now includes AI Performance reporting). For Claude, Perplexity, and cross-engine comparison, you need a parallel monitoring tool.

Is the guide wrong about anything?

The guide is accurate for Google. Its limitation is scope, not accuracy. It does not address non-Google engines, earned media distribution, citation-mention gaps, or measurement volatility — all of which are central to a complete AEO program.

Bottom line

Google's official position is clear: for Google, AEO and GEO are still SEO. Skip the hacks. Build quality content. Use Search Console. But do not read the guide as permission to ignore ChatGPT, Claude, and Perplexity — because those engines read different indexes, cite different sources, and serve the same buyers through different retrieval pipelines. Google SEO is the floor. Multi-platform monitoring is the program.