Google Says AEO and GEO Are "Still SEO." Here's What That Actually Means for B2B CMOs.

Google Says AEO and GEO Are 'Still SEO.' Here's What That Actually Means for B2B CMOs.

Last week, Google quietly published the document every marketing leader has been waiting for: an official guide to optimizing for generative AI search. It's the first time Google has openly addressed the discipline that the rest of the industry has spent two years calling AEO (Answer Engine Optimization) or GEO (Generative Engine Optimization).

Google's verdict, in one sentence: it's all just SEO.

If you run marketing at a B2B SaaS company, you've probably been getting pitched by three different agencies, two new tools, and a handful of "AI search experts" who insist this is a whole new discipline. Google's official position is the opposite. From their perspective, "optimizing for generative AI search is optimizing for the search experience, and thus still SEO."

That's a tidy line. It's also slightly misleading — not because Google is wrong, but because most of the SEO budget you're spending right now isn't actually doing the things Google's guide describes. The work is recognizably SEO. The execution is not what most agencies are shipping.

Here's what Google actually said, what veteran developer Trevor Lasn read between the lines, and what you should change this quarter.

What Google actually published

The guide is short — under 2,000 words — and refreshingly direct for an official Google document. The headlines:

AI features run on the same Search index. AI Overviews and AI Mode use retrieval-augmented generation (Google calls it RAG, or "grounding") to pull relevant pages from the regular Search index, then summarize them. They also use "query fan-out": the model generates a handful of related sub-queries behind the scenes to gather more context before answering. Both techniques sit on top of the standard ranking systems, which means foundational SEO still decides whether you're eligible.

The content advice is sharper than usual. Google singles out "non-commodity content" as the bar. Their own example: a generic listicle like "7 Tips for First-Time Homebuyers" is commodity. A post titled "Why We Waived the Inspection & Saved Money: A Look Inside the Sewer Line" is not. The first can be written by any model from training data. The second cannot.

Five things Google says you can stop worrying about. This is the most useful section in the guide, and the one that contradicts a lot of advice circulating on LinkedIn:

1. You don't need an llms.txt file. Not a ranking signal. Not used by Google's AI features.
2. You don't need to "chunk" your content into tiny snippets for AI. Google's systems read the whole page.
3. You don't need to rewrite copy "for AI." The model understands synonyms and intent.
4. You shouldn't chase inauthentic mentions. The same spam systems that catch link schemes catch this.
5. You don't need extra structured data for AI. Schema is still useful where it powers rich results, but it isn't a separate AI lever.

Agentic experiences are the next surface. Google flags AI agents — Claude with computer use, ChatGPT's Operator, Perplexity's assistant — as the thing to start preparing for. Their guidance overlaps almost completely with classic accessibility work.

That's the official document. Now here's where it gets interesting.

The expert reframe: same discipline, sharper edges

Trevor Lasn's breakdown of the guide agrees with Google's framing — AEO and GEO are SEO — but argues the work has shifted in ways most teams haven't caught up with. Three points stand out.

1. Eligibility is a gate most sites quietly fail

Before any content or strategy work matters, your page has to be allowed to appear in AI features at all. Google's guide is explicit: a page is only eligible for AI Overviews if it's eligible to appear as a regular search snippet. That means it has to be indexed, crawlable in robots.txt, snippet-allowed (no nosnippet, no max-snippet:0), and renderable without forcing the crawler to wait on JavaScript.

Lasn points out that "many sites accidentally block one of those and tank their visibility." If your developers added a defensive WAF rule that blocks bots, or your CMS auto-applied nosnippet to legal pages, or your Next.js setup ships content client-side only — you can be ranking well in regular search and completely invisible in AI Overviews.

This is the single most important pre-flight check for any CMO who's "doing SEO" but doesn't know what their team configured at the infrastructure layer.

2. Different AI engines pull from different indexes

Google's guide only speaks for Google. But your buyers aren't only using Google. They're using ChatGPT Search, Claude with web search, Perplexity, and Microsoft Copilot. Lasn maps it out clearly:

• Google Search index → Google Search, AI Overviews, Gemini grounding
• Bing index → Bing, Microsoft Copilot
• OpenAI's index (OAI-SearchBot) → ChatGPT Search
• Anthropic + Brave → Claude web search
• Perplexity + Bing → Perplexity answers

The practical implication: there are five separate crawlers you need to let in if you want to be cited across the major AI surfaces. There are also several training crawlers — GPTBot, ClaudeBot, Google-Extended — that you can choose to block without affecting search visibility. Most agencies haven't audited a client's robots.txt for this distinction. It takes ten minutes and can quietly recover meaningful visibility.

3. The content that gets cited isn't the content that ranks

Both Google and Lasn agree on this, and it's the point worth printing and pinning above your content team's desk: a model only cites a page when the page contains something the model can't write from training data alone.

Lasn's example is the cleanest version of this advice we've seen. Two paragraphs on the same topic, same length:

Commodity: "Next.js 16 introduces async params, making route parameters asynchronous. This is a breaking change you should plan for when upgrading from Next.js 15."

Distinctive: "We migrated a 240-route Next.js 15 app to 16 last week. The async params change broke 47 pages in CI on the first run... Took 3 hours, almost all of it search-replace."

Both rank for similar queries. Only the second gets cited. The model can synthesize the first from training data. It cannot invent the "47 pages" or "3 hours" without quoting the source.

For B2B SaaS, the translation is direct: the case study, the customer call recording, the benchmark you ran against your own product, the failure post-mortem — these are the formats AI engines have to quote. The "ultimate guide to [generic topic]" is not.

What B2B marketing leaders should actually do this quarter

If you read both documents and pulled out just the actions, the playbook is shorter than the discourse implies. Five things to check:

1. Run an AI eligibility audit on your top 20 pages. Open Search Console, run a URL inspection on each page that drives revenue or pipeline. Confirm it's indexed, the rendered HTML contains your content (not just a JavaScript shell), and there's no nosnippet directive. If your blog runs on a client-rendered framework, this is where most of your AI visibility leaks.

2. Audit your robots.txt for AI search crawlers. Make sure Googlebot, Bingbot, OAI-SearchBot, Claude-SearchBot, and PerplexityBot are all allowed. Decide separately whether you want to block training crawlers (GPTBot, ClaudeBot, Google-Extended). Note the catch: blocking Google-Extended opts you out of being grounded in Gemini Apps and Vertex AI Grounding, but does not affect Google Search or AI Overview eligibility.

3. Audit your last 20 blog posts for "commodity vs. distinctive." Read each one and ask: could ChatGPT write this from training data alone? If yes, it's never going to be cited. Either kill it, or rewrite it with the data, screenshots, benchmarks, or customer stories that only your team can produce.

4. Test where you currently show up. Open ChatGPT, Claude, Perplexity, and Google's AI Overview and ask the questions your buyers ask. Note which competitors are getting cited. Track this monthly — it's the closest thing to a real AI ranking signal that exists right now, and most teams aren't measuring it at all.

5. Treat agentic readiness as accessibility. Real <button> elements with aria-labels. Real form inputs with real type attributes. Real headings in real hierarchy. This is the same work you'd do to comply with WCAG, and Google's guide says it's the same work that prepares your site for AI agents. Two checkboxes, one effort.

The bottom line

Google is right that AEO and GEO are SEO. They're also right that most of the "AI optimization hacks" being marketed at you are noise — llms.txt files, content chunking, rewriting for AI, schema theatrics. The work that matters is the work Google has been recommending for a decade, executed with sharper attention to three things: technical eligibility for AI surfaces, content that can't be synthesized from training data, and visibility across the five major AI engines, not just one.

The uncomfortable corollary is that your current SEO spend may not be doing any of this. Most retainers were scoped before AI Overviews existed. The keyword research, the link building, the thin "topic cluster" pages — they were built for blue-link rankings, not for being cited as a source inside an answer your buyer never clicks out of.

The marketing leaders who win the next two years aren't going to hire a "GEO agency" alongside their SEO agency. They're going to retire the playbook that's been quietly underperforming and rebuild around what actually gets brands cited. Google just published the spec. The question is whether your team is shipping against it — or still shipping the 2020 playbook in 2026.

At OutAnswer, we track exactly this: where your brand shows up across ChatGPT, Claude, Perplexity, Google AI Overviews, and Gemini, which competitors are getting cited instead of you, and what to fix to close the gap. If you're already spending on SEO and want to know whether it's working in the place your buyers are actually deciding, start with a free AI visibility scan.

Sources cited:

• Google's Guide to Optimizing for Generative AI Features on Google Search — Google Search Central, last updated May 15, 2026
• AEO and GEO for AI Overviews, ChatGPT, Claude, Gemini, and Perplexity — Trevor I. Lasn, May 17, 2026