Learn how to rank in AI overviews using Google's own May 2026 guide. What to do, what to skip, and the myths the guide officially debunks.
If you've been wondering how to rank in AI overviews, you've likely waded through two years of conflicting advice — new file formats, AI-friendly content structures, special schema, and prose rewritten to sound "machine-readable." Most of it was invented from thin air. Then, on May 15, 2026, Google published its first official guidance on the subject, and it turns out the answer is far closer to traditional SEO than the consulting industry wanted you to believe.
Then, on May 15, 2026, Google published its first official guidance on how to rank in AI overviews. The document lives at developers.google.com/search/docs/fundamentals/ai-optimization-guide — announced by John Mueller via the Google Search Central Blog and placed under the search fundamentals section, which means it went through Google's legal and policy review. This is the canonical position of Google Search, not a blog post or a conference comment.
What it actually says will surprise a lot of people who spent the last two years buying into AEO and GEO consulting packages. Here is a close reading of the guide: the things it tells you to do, the tactics it explicitly buries, and what it pointedly leaves out.
The guide was last updated July 10, 2026, and covers Google's two main AI features: AI Overviews and AI Mode. Both, Google confirms, are powered by retrieval-augmented generation (RAG), a mechanism that pulls relevant pages from Google's existing Search index to ground the AI response. A "query fan-out" layer breaks a search into related sub-queries to gather a fuller answer, pulling supporting links from the same index.
The practical implication: if your page is not in Google's index, it cannot be cited in an AI Overview, full stop. The AI layer sits on top of Search, not beside it.
Google AI Overviews now reach more than 2.5 billion monthly users, according to Google's own figures. AI Mode passed 1 billion monthly users as of May 2026. AI Overviews appear on roughly 21% of all searches overall, but trigger on 57.9% of question-based queries and 59.8% of "why" queries, according to Ahrefs data from 1.9 million citations analyzed. For ecommerce content operators, informational and question-format posts are where the citation opportunity is densest. Transactional queries like "buy" or "shop" rarely trigger AI Overviews because Google won't replace a purchase journey with a summary.
Google says this matters "more than any of the other suggestions in this guide." The distinction it draws is sharp. Commodity content is its own example: a generic "7 tips for first-time homebuyers" post anyone could write. Non-commodity content is a first-hand account of a specific decision, with the reasoning and trade-offs included. It is the kind of piece a generative AI could not produce from common knowledge alone.
The test Google gives: would a generative AI model produce this from existing information on the web? If yes, the content won't differentiate your page. First-hand experience, original data, proprietary insight, and strong examples are what move the needle. For a Shopify store, this means your product blog posts need to reflect your actual inventory experience, your customers' specific outcomes, or knowledge that only someone operating in your niche would have.
If you want a practical framework for turning AI-drafted posts into something genuinely citable, the ten-minute edit process we outlined here addresses exactly that gap.
Google is explicit: a page must be indexed and eligible to appear in Google Search with a snippet to be eligible as a supporting link in AI Overviews or AI Mode. This sounds obvious, but one area the guide emphasizes more than usual is JavaScript rendering. Most AI crawlers struggle with content that only appears after client-side rendering. If your product names, pricing, or feature specifications live behind JavaScript, both traditional search and AI systems may not see them.
Use Google Search Console to check indexing status before worrying about anything else on this list. An unindexed page is invisible regardless of content quality.
Google confirms that images and video bring more opportunities for pages to appear beyond text links in AI responses. If your pages include relevant visuals, such as screenshots, charts, diagrams, product imagery, or short video explainers, they may have more ways to contribute to search experiences than a text-only article. For ecommerce sellers especially, product imagery with accurate alt text and structured product data can support discoverability across shopping-related AI features.
The guide's framing shifts the brief for content teams: a strong article should be treated as an information asset, not just copy. Ask what the reader would benefit from seeing, not just reading.
Authorship, credentials, first-hand experience, and reputation signals are explicitly named in the guide as important across all Google systems, not just traditional search. Verified author profiles and demonstrated expertise are confirmed signals for AI features. Consistent business details across directories, brand mentions across third-party sites, forums, reviews, and industry publications all strengthen your entity signals and improve your chances of AI Overview citation.
For local businesses and product pages, the guide specifically recommends a verified Google Business Profile and a Merchant Center product feed as the primary mechanisms for AI-powered local and shopping features. Schema alone is not sufficient for these use cases.
Google's exact framing: "optimizing for generative AI search is optimizing for the search experience, and thus still SEO." There is no alternative path to AI citation that bypasses traditional SEO. Strong organic rankings are, statistically, the clearest path to AI Overview inclusion. According to Ahrefs data from 1.9 million citations, 76% of AI Overview citations come from pages already ranking in Google's top 10, with the median cited position at #2.
Content density beats content length. Ahrefs data shows near-zero correlation (Spearman ~0.04) between word count and AI citation. Direct answers and factual specificity matter far more than article length. A page ranking #4 with a single excellent, self-contained paragraph can win a citation while a sprawling #1 page with an unfocused introduction does not.
This connects directly to the broader topic cluster approach. Topical depth across a set of related posts signals authority more reliably than any single long article. Our piece on pillar pages and hub-and-spoke content covers how to build that kind of topical depth for a product category.
This is the part of the guide that the industry needed to hear, and it is unusually direct. Google does not just fail to endorse these tactics. It explicitly says they are not needed or not effective.
Google's guide explicitly states that site owners do not need to create any new machine-readable files or "AI text files" to appear in AI Overviews or other AI search features. If you have seen consultants selling llms.txt setup as a Google AI visibility tactic, that service rests on a false premise for Google's surfaces specifically. (llms.txt may still have a niche use case for non-Google AI platforms — more on that in our full breakdown here.)
The guide states that Google's AI systems can read nuance and handle multiple topics on a single full-length page. You do not need to fragment your articles into tiny, self-contained pieces for AI extraction. Artificial content chunking, definition-heavy rewriting, and LLM-specific formatting are unnecessary. Write in natural prose that serves your human readers and Google's AI can handle the rest.
Google says there is no special Schema.org structured data required for AI Overviews or AI Mode. Standard structured data — Article, Product, FAQ, Offer — is still worth implementing for rich results in regular Search. But adding experimental or AI-specific schema types on top of your existing markup won't trigger citation. Schema validity matters more than schema volume. It is also worth noting that Google retired FAQ rich results for most page types back in 2023, so that particular markup obsession is doubly moot now.
Writing in a stilted, definition-heavy style intended to sound "machine-friendly" does not help and likely hurts readability. Google's systems parse natural language effectively. The guide's implied rule: write for humans first. If it reads well to a person, it reads well to the model. The effort you would spend rewriting readable content into AI-speak is better spent adding original data or a first-hand case study.
One important caveat to flag: Google's guide is scoped to "generative AI features on Google Search." It says nothing about ChatGPT, Claude, Perplexity, or Copilot. Those platforms operate on different models, different crawlers, and different citation logic. The tactics that earn Google AI Overview citations may not translate identically to those platforms.
If your customers are increasingly starting product research in ChatGPT or Perplexity, and for many categories they are, optimizing for Google alone is not a complete strategy. The overlap between what ChatGPT and Perplexity cite is surprisingly small (around 11%), which means being cited on one platform does not automatically earn you visibility on the other. Our piece on ChatGPT vs Perplexity citations goes deeper on this gap.
The guide also does not address AI agents directly with final answers. It notes agents as "quickly emerging and evolving" and points to a new Universal Commerce Protocol for agent interactions. That space is still in flux, but the accessibility and machine-readability of your site structure will matter as agents begin visiting pages to book, compare, and transact on behalf of users.
Pull up your last ten published posts and run each one through these questions. Every "no" is a prioritized action item.
If you want to track which queries are triggering AI Overviews and whether your pages are showing up in them, the tools and methods for that are covered in this guide on tracking AI visibility in 2026.
The core message of Google's guide is good news for content operators who have been doing the fundamentals well: there is no new game. The same things that earn you organic rankings, specifically original, experience-led content, clean technical health, and real topical authority, earn you AI Overview citations. The tactics that were always shortcuts (thin content, manufactured mentions, schema stuffing) don't work here either.
The one real shift is the bar for originality. AI systems can synthesize common information quickly, which raises the standard for what deserves a citation. Commodity content is easier than ever to produce. Content that no competitor could plausibly publish, because it reflects your specific experience, your specific data, or your specific customers, is what stands out precisely because of that gap.
That is where the effort should go. Not into llms.txt files, not into AI-friendly chunk formatting, and definitely not into rewriting readable posts into stiff, machine-targeted prose. Produce fewer generic posts and more posts built on something real. The index will reward it, and the AI layer that reads from the index will too.
No. Google's May 2026 AI optimization guide explicitly states that llms.txt files are not needed to appear in AI Overviews or AI Mode. You do not need to create any new machine-readable files for Google's AI features. That said, llms.txt may still have a niche use for non-Google AI platforms like ChatGPT or Perplexity.
No. Google confirmed in its official guide that AI Overviews and AI Mode run on the same core ranking and quality systems as standard Search. There is no separate 'AEO' or 'GEO' algorithm underneath. If your page ranks well in organic search, it is already eligible to be cited.
No. Google's guide states that content does not need to be broken into small chunks for AI systems to understand it. Google's AI can parse nuance and cover multiple topics on a single full-length page. Writing clearly for human readers is enough.
According to Ahrefs analysis of 1.9 million citations, 76% of AI Overview citations come from pages that already rank in Google's top 10, with the median cited position sitting at #2. Strong organic rankings are the single most impactful input for AI Overview inclusion.
Not directly. Google's guide states no special Schema.org structured data is required for AI Overviews or AI Mode. Standard schema types such as Article, Product, and FAQ still earn rich results in regular Search, which improves visibility and click-through rates, but adding AI-specific schema types won't trigger citation.
Non-commodity content is original material that a generative AI model could not produce from common knowledge alone. Google's own example contrasts a generic '7 tips' post with a first-hand account that includes specific reasoning, data, and trade-offs. The test: if an AI could write it without unique inputs, it won't stand out.
The playbook
One email when we publish something worth reading. No drip sequence, no spam, unsubscribe any time.