TL;DR

Mike King of iPullRank published a detailed takedown of Google's new AI search optimization guide on May 18, 2026, arguing it protects Google's platform interests rather than helping publishers navigate multi-platform AI search. His core thesis: Google dismisses GEO, content chunking, and llms.txt not because they don't work, but because they help publishers build visibility on platforms Google doesn't control. The real question isn't who's right. It's what you can measure today across ChatGPT, Perplexity, Claude, and Gemini instead of waiting for Google to tell you what matters.

Key Takeaways

Google's May 15 guide calls AEO and GEO "still SEO" and dismisses chunking, llms.txt, and passage-level optimization
King argues this follows a 15-year pattern where Google absorbs new surfaces into "just SEO" without acknowledging new skill requirements
Microsoft's Bing team has taken the opposite approach, openly naming GEO and shipping an AI Performance dashboard with citation metrics
The practical takeaway: don't wait for platform guidance to catch up. Measure your actual AI brand citations across all major engines now.

What King Said and Why It Matters

On May 18, 2026, Mike King, founder and CEO of iPullRank, published "Google's Guidance on AI Search is Naive and Self-Serving" on ipullrank.com. The piece arrived three days after Google released its updated guide through Google Search Central titled "Optimizing your website for generative AI features on Google Search."

King's argument cuts deeper than a standard industry disagreement. He frames the guide not as flawed technical advice, but as strategic communication designed to keep practitioners focused on Google's ecosystem. The pattern, he writes, has repeated for more than fifteen years: mobile was "just SEO," voice was "just SEO," AMP was "just SEO." Each time, publishers absorbed more work without proportional recognition that the skill requirements had expanded.

With AI search, King argues the gap is wider than ever. Traditional SEO covers keyword research, technical auditing, link building, and rank tracking. AI search adds vector distance measurement, RAG pipeline analysis, passage-level content engineering, brand citation tracking across LLM platforms, and agent protocol design. Pretending these skill sets are identical leads organizations to underhire for the problem.

Three Objections That Should Change How You Think

1. Chunking matters whether Google admits it or not

Google's guide states there is no need to break content into small pieces for AI systems to understand it. King counters with a technical argument: chunking is what RAG systems do to your content regardless of whether you optimize for it. A passage focused on one clear idea retrieves better than a passage covering three topics, because vector similarity scoring rewards semantic coherence. This is verifiable on public APIs.

King also points to Google's own MUVERA research and passage indexing patents, neither of which align with the public guidance that chunking is irrelevant. Meanwhile, Bing's documentation states directly that transformations must preserve the meaning and claims used in answers.

2. llms.txt is useful outside Google's walls

Google says publishers don't need machine-readable files like llms.txt. King's response: that's true for Google and misses the entire point. Anthropic has published documentation supporting llms.txt in Claude. Other systems read it. The honest guidance would be: Google doesn't process llms.txt, other systems may, decide accordingly. Instead, the guide conflates "we don't use it" with "you don't need it."

3. Bing is doing the opposite, openly

This might be the most telling contrast. In February 2026, Microsoft launched an AI Performance dashboard in Bing Webmaster Tools that gives publishers visibility into citation frequency, page-level activity, grounding query phrases, and temporal trends. Jordi Ribas, Corporate Vice President of Search and AI at Microsoft, openly named Generative Engine Optimization without dismissive framing. As King notes, read the three Bing posts and then read Google's mythbusting section, and they don't appear to describe the same technology.

The Missing Piece: Measurement Over Guidance

Here's where we see the debate differently from both sides.

Google says everything is fine and nothing has changed. King says Google is protecting its platform. Both positions contain truth, but neither answers the question a brand manager or SEO lead actually needs answered on Monday morning: are AI search engines citing my brand, and what drives those citations?

The uncomfortable reality is that Google organic rankings and AI citations are increasingly divergent. Research from the 5W Public Relations study showed that overlap between traditional search rankings and AI recommendations dropped from roughly 70% to around 20%. You can rank first on Google and be invisible in ChatGPT. You can have zero organic traffic and get cited by Perplexity.

This gap means that guidance from any single platform is structurally incomplete. What matters is cross-platform measurement.

Five Things to Do This Week

1. Audit your AI citations across engines. Check what ChatGPT, Perplexity, Claude, and Gemini say about your brand right now. Not what you hope they say. What they actually say. Tools like BeRecommended automate this across all major AI engines in one scan.

2. Compare AI citations with your Google Search Console data. Look for the gap. Which queries bring organic traffic but zero AI mentions? Which topics get AI citations but no Google clicks? That gap is your priority map.

3. Write passage-first content. Whether or not you call it chunking, make each section of your content a self-contained answer. One idea per passage. Clear entity references. Specific numbers over vague qualifiers. This improves retrieval regardless of which system is doing the retrieving.

4. Add llms.txt if you haven't. The cost is near zero. The upside is real for systems that read it. Google ignoring it is Google's choice, not your constraint.

5. Track changes monthly. AI search visibility is not static. Models update, retrieval pipelines shift, and your competitors are publishing too. Monthly measurement catches regressions before they compound.

FAQ

Is GEO actually different from SEO?

Yes, though they share a foundation. Traditional SEO optimizes for one ranking system and human click behavior. GEO optimizes for multiple retrieval systems, synthesis pipelines, and AI-generated answers that may never link to your site. The skill overlap is real but partial. Treating them as identical leads to blind spots in AI visibility.

Should I follow Google's AI search guide?

The sections on creating valuable, non-commodity content with genuine first-hand experience are solid advice. The sections dismissing chunking, llms.txt, and GEO as unnecessary reflect Google's platform perspective. For a complete strategy, pair Google's guidance with cross-platform measurement.

Does content chunking actually improve AI citations?

Evidence from RAG pipeline research and public API testing suggests that focused, single-topic passages retrieve better than multi-topic blocks. This is consistent with Bing's published documentation and Google's own passage indexing research, even though Google's public guide downplays it.

How do I know if AI engines are citing my brand?

Query the major AI search engines (ChatGPT, Perplexity, Claude, Gemini) with questions your target audience asks. Compare responses against your brand and competitors. Alternatively, use an automated AI brand visibility tool like BeRecommended that tracks citations across all major platforms.

The Bottom Line

Mike King's critique of Google's AI search guide is the most technically grounded challenge to come from the SEO community this year. His core argument, that the guide serves platform interests rather than publisher interests, is supported by documented contradictions between Google's public guidance and its own research.

But the real takeaway isn't about who's right in the Google vs. GEO debate. It's that relying on any single platform's guidance is a losing strategy when your audience is fragmenting across six or more AI answer engines. Measure your actual visibility. Optimize for what the data shows. Let the platform debates sort themselves out.

Sources

Mike King: Google's Guidance on AI Search is Naive and Self-Serving (iPullRank, May 18, 2026)
Mike King: Google's AI search guide serves platform over the open web (PPC Land, May 18, 2026)
Google's new AI search guide challenges everything SEOs thought they knew (PPC Land, May 18, 2026)
Google: Optimizing your website for generative AI features on Google Search (Google Search Central, May 15, 2026)
Bing gives publishers first look at how AI systems cite their content (PPC Land, February 10, 2026)

Mike King Just Exposed Google's AI Search Guide as Platform Strategy. Here's What to Actually Do.