In May 2026, Google published its first official guide to optimizing for AI Overviews and AI Mode, and the headline was blunt: “From Google Search’s perspective, optimizing for generative AI search is optimizing for the search experience, and thus still SEO.” In plain terms, there’s no secret AI ranking system to game on Google. The same things that earn rankings, crawlable pages and genuinely useful, original content, earn a place in AI answers.
That doesn’t mean nothing changed. It means the work changed shape, a lot of “AEO hacks” can be dropped, and the budget belongs on fundamentals done better. Here’s what the guide says, what to stop doing, and where the picture differs for ChatGPT and Perplexity.
Key takeaways
– Google’s AI optimization guide (published May 15, 2026) says AI Overviews and AI Mode are “rooted in our core Search ranking and quality systems.”
– The guide names tactics you can ignore for Google: llms.txt files, “chunking” content, rewriting for AI, chasing inauthentic mentions, and overfocusing on structured data.
– To appear in AI features, a page must be indexed and eligible for a snippet, and the site must be included in Search generative AI features in Search Console.
– The guidance covers Google only. ChatGPT, Perplexity and other assistants run their own retrieval, so they need their own testing.
– The practical shift: fewer, better, more original pages; clean technical foundations; and real brand reputation.
What exactly did Google say about AEO and GEO?
Google said they aren’t separate disciplines for its search engine. The guide defines AEO as “answer engine optimization” and GEO as “generative engine optimization,” then says that optimizing for generative AI search on Google is still SEO.
It also explains the mechanics. AI Overviews and AI Mode use two techniques to pull from Google’s index:
- Retrieval-augmented generation (RAG), or grounding: the AI finds relevant pages in the index and builds its answer from them.
- Query fan-out: the system runs several related searches at once to gather more sources than the original query alone would return.
Both start from the same index and quality systems that power regular search. That’s why Google says SEO best practices “continue to be relevant.” As Search Engine Journal noted, Google staff had said the same at Search Central Live events. The difference now is that it’s in writing, in official documentation you can send to a client or your boss.
What does Google say you can stop doing?
The guide includes a mythbusting section. For Google Search, it says you don’t need:
| Tactic | What Google says |
|---|---|
| llms.txt and other “AI files” | Google Search doesn’t use them. They won’t help or hurt your Google visibility. |
| “Chunking” content into tiny sections | Not needed. Write for people. |
| Rewriting content specifically for AI | Not needed. |
| Seeking inauthentic mentions | Not as helpful as it might seem. Spam systems still apply. |
| Special schema for AI | Structured data isn’t required, and there’s no special AI markup. Keep using it for rich results. |
If an agency proposal is built around these, it’s worth asking why. And if you’re an agency that sold them last year, the guide is a good moment to reset expectations with clients honestly.
One caveat: Google says it’s fine to keep an llms.txt file for other services that use one. It simply does nothing for Google.
What should you focus on instead?
The guide points to three things, and all of them are familiar. The difference is the standard you hold them to.
1. Non-commodity content
Google asks for “valuable, unique, non-commodity content.” Commodity content is anything a reader could get from a dozen other pages. Non-commodity content brings something only you can: first-hand experience, original data, a clear expert opinion, real examples, local knowledge.
A quick test for any page: if you removed your logo, could a reader tell who wrote it? If not, it’s commodity.
2. A clean technical foundation
Google’s AI features page lists the basics that still matter: allowing crawling in robots.txt and at the CDN level, internal links that make content findable, good page experience, important content in text form, structured data that matches visible text, and up-to-date Merchant Center and Business Profile information.
There’s one new item. The guide says a site must be included in Search generative AI features in Search Console to be eligible. Google added a control in 2026 that lets site owners opt out of these features. Check that nobody switched it off by accident.
3. Real reputation
The guide notes that AI features can show what’s being said about products and services across blogs, videos and forums. That’s an argument for earned coverage, reviews and expert visibility, not fake mentions.
Does “still SEO” mean nothing has changed?
No. The ranking foundation is the same, but the outcome you measure is different.
Clicks are harder to win when an AI answer sits above the results. In Pew Research Center’s browsing study, people ended their browsing session on 26% of searches that showed an AI summary, compared with 16% of searches without one. So “still SEO” doesn’t mean “same reporting.” It means:
- Measuring visibility in AI answers as well as rankings. Search Console’s Generative AI performance report now shows AI Overview and AI Mode impressions by page.
- Judging content by whether it earns citations and qualified visits, not just sessions.
- Covering the related questions that fan-out searches look for, not just one head keyword.
Does Google’s guidance apply to ChatGPT and Perplexity?
No. The guide is explicit that it covers Google Search features like AI Overviews and AI Mode.
ChatGPT, Perplexity, Gemini (the app), Claude and Copilot each use their own retrieval, and some lean on Bing or their own crawlers. A few practical differences:
- Crawler access: check that your robots.txt and firewall aren’t blocking the AI crawlers you want, such as OAI-SearchBot for ChatGPT search or PerplexityBot. Each company publishes its own crawler documentation.
- Bing matters more: several assistants use Bing results, so Bing Webmaster Tools is worth setting up if it isn’t already.
- Testing is manual: there’s no Search Console for most assistants. Run a fixed set of buyer prompts monthly and record whether you’re mentioned or linked.
The good news is that the foundations overlap heavily. Crawlable, clear, original, well-cited content with a real reputation behind it tends to do well everywhere.
What should agencies change this quarter?
Here’s a practical checklist:
- Audit proposals and SOWs for line items built on tactics Google now calls unnecessary. Replace them with work that improves content quality or reputation.
- Check eligibility. Confirm each client site is indexed, snippet-eligible and included in Search generative AI features in Search Console.
- Run a commodity audit on the top 20 traffic pages. Mark each as “unique,” “partly unique” or “commodity,” and plan upgrades for the last group.
- Add AI visibility to reporting. Include the Generative AI performance report and a monthly prompt test.
- Set up Bing Webmaster Tools and review AI crawler access for non-Google assistants.
- Brief clients in plain language. Share the guide and explain what you’re doing more of and less of.
What should business owners ask their agency?
- “Which parts of our plan are based on Google’s published guidance, and which are guesses?”
- “Are we paying for llms.txt, chunking or ‘AI rewrites’? Why?”
- “Which of our top pages would a reader call original?”
- “How are you measuring our visibility in AI Overviews and in ChatGPT?”
- “Are we included in Search generative AI features in Search Console?”
A good agency will welcome these questions. They’re the same ones it should be asking itself.
How do you run a commodity audit?
A commodity audit is the fastest way to act on Google’s “non-commodity content” advice. Take your 20 highest-traffic or highest-value pages and ask four questions of each:
- Experience: Does the page show we’ve actually done this work, with real examples, photos, steps or lessons?
- Evidence: Does it include anything only we know, such as our own data, pricing ranges, timelines or results (with a method note)?
- Point of view: Does it take a clear position, or just list what everyone else lists?
- Answer quality: Does it answer the main question in the first two or three sentences?
Score each page “unique,” “partly unique” or “commodity.” Then fix the commodity pages first, starting with the ones closest to revenue, such as service, pricing and comparison pages.
Illustrative example (placeholders, not client data): A [roofing company]’s “How much does a roof replacement cost?” page lists national averages copied from other sites. The upgrade adds the company’s own typical price range for [city] ([X]–[Y]), the three factors that most often change a quote in its experience, and a short note from the owner on when repair beats replacement. Same URL, same topic, but now it says something no other page can.
This kind of upgrade is “still SEO,” and it’s exactly what Google’s guide is asking for.
Common misreadings of the guide
- “So AI search doesn’t matter.” It does. Google is saying the path to visibility runs through SEO, not around it.
- “So we can keep doing what we did in 2019.” Thin, keyword-led pages are exactly what non-commodity guidance rules out.
- “Structured data is useless now.” No. It still drives rich results. It’s just not an AI shortcut.
- “This covers ChatGPT too.” It doesn’t. Test other assistants separately.
Frequently asked questions
Is AEO different from SEO?
For Google Search, no. Google’s May 2026 guide says optimizing for its generative AI features “is optimizing for the search experience, and thus still SEO.” For other AI assistants, the foundations overlap but each has its own retrieval, so testing is still needed.
Do I need an llms.txt file?
Not for Google. Google says its search doesn’t use llms.txt, so it won’t help or hurt your Google visibility. You can keep one if another service you care about uses it.
Does structured data help with AI Overviews?
Google says structured data isn’t required for its AI features and there’s no special AI markup. It still helps with rich results, so keep it accurate and matched to visible content.
How do I make sure my site is eligible for AI Overviews?
Your pages must be indexed and eligible to show with a snippet, and your site must be included in Search generative AI features in Search Console. Then the usual quality and relevance factors decide whether you’re shown.
What is non-commodity content?
Content that offers something readers can’t get from many other pages, such as first-hand experience, original data, expert opinion or specific local knowledge. Google’s guide names it as a priority for AI search visibility.
Does Google’s guide apply to ChatGPT?
No. It covers Google Search’s AI features only. ChatGPT, Perplexity and other assistants run their own systems, so check crawler access and test them directly.
Want a plan built on what actually works?
We help businesses and agency partners focus their SEO budget on the work Google’s own guidance supports, and on measuring visibility in AI answers across Google and beyond.
Prefer to talk? Call (813) 397-3665 or email strategy@seooutsourcing.com.
