Google AI Overviews now appear on roughly half of US searches, so being included in them is one of the most valuable things a page can do, and one of the most misunderstood. The good news is that Google has actually published how to do it, and the honest version is less exotic than most advice suggests. You do not need special markup, content chunking, or an llms.txt file. You need to be indexed, genuinely useful, and structured so a machine can lift a clean answer. This is the practical how-to, grounded in Google's own guidance rather than the hacks being sold around it.
How inclusion in an AI Overview actually works
An AI Overview is the AI-generated summary Google shows above the traditional results. Getting into one is a chain, and each link is a requirement.
First, your page has to be in Google's index, because AI Overviews are built on the same index as normal search. Second, it has to be eligible for a snippet, since Google draws AI Overview content from snippet-eligible pages. Third, it has to be retrieved and then selected as one of the sources the Overview is built from. Miss the first two and you are not even a candidate.
This is why Google can honestly say that optimizing for AI Overviews is, from its perspective, still SEO. There is no separate AI Overview index and no secret submission process. If you are invisible to Google, you are invisible to its AI. We cover the underlying retrieve-and-cite mechanism in How LLMs Retrieve and Cite Sources.
What Google says works
In May 2026 Google published guidance on optimizing for its generative AI features, which is the most authoritative source there is on this, since it is the company that builds AI Overviews. Its advice is deliberately ordinary:
Create genuinely useful, reliable, people-first content. Google's example is pointed: a first-hand review beats a summary of other people's reviews. Unique, experience-based content is the biggest lever it names. Make sure the site is technically healthy and crawlable, so the content can be found and understood. And keep information accurate and well-structured, so it can be extracted cleanly.
That is the whole official recipe. It rewards depth and originality, not tricks.
What Google says you do NOT need
This is the part worth reading twice, because it contradicts a lot of what is sold as AI Overview optimization. Google explicitly lists tactics you do not need:
You do not need to chunk your content into small pieces. Google's systems understand multiple topics on one page. You do not need to rewrite content specifically for AI, because models handle synonyms and intent. You do not need llms.txt or other special machine-readable files. And you do not need special schema markup to appear in AI features, though valid structured data that matches your visible text is still fine to keep for other reasons.
If an agency is selling you chunking, an llms.txt file, or schema as the key to AI Overviews, they are selling something Google has said in writing does not work.
The practical steps that actually move the needle
Here is the how-to, in the order that matters.
Get indexed and snippet-eligible
This is the floor. Confirm the page is indexed in Google Search Console, renders cleanly, loads fast, and is not accidentally blocking crawlers or suppressing snippets (a restrictive nosnippet or a very low max-snippet setting can exclude you). If you are not eligible for a snippet, you cannot appear in an AI Overview.
Answer the question directly, near the top
AI Overviews are assembled by lifting clean, self-contained passages. Put the direct answer in the first sentence under each heading. If the query is "how to prevent inventory loss," the passage should start with the answer, not with a paragraph of throat-clearing. This is the core of answer engine optimization.
Structure content so passages stand alone
Use question-style headings that mirror how people actually ask. Keep each section self-contained, so it makes sense lifted out of context, because that is exactly what happens to it. Use short lists and tables where they genuinely clarify, since these extract cleanly. Well-structured FAQ content still helps here, even though Google deprecated FAQ rich results in May 2026: the format is easy to quote, you just should not expect the old rich-result dropdown.
Publish genuine, first-hand content
This is Google's biggest stated lever and the hardest to fake. Original data, real experience, a specific point of view, and information other pages do not already have. The 2023 GEO research found the same thing from the other direction: content that adds new information, quotations and statistics is more likely to be cited. Commodity content that restates the consensus rarely gets picked.
Keep structured data accurate, but do not lean on it
Add valid schema (Organization, Article, and so on) because it is sound technical practice and Google advises that structured data should match your visible text. Just do not treat it as the lever. A controlled test found schema behaves as an SEO signal, not an AI-citation one. Do it, then move your effort to content and structure.
Earn rankings and off-site authority
AI Overviews still favor content Google already trusts, but the link is loosening: only 38% of AI Overview citations come from top-ten pages, down from 76% eight months earlier, so pages outside the top ten are increasingly cited too. Strong organic performance helps you get retrieved; off-site authority helps you get trusted. Brand mentions correlate with AI visibility about three times more strongly than backlinks, which is the generative engine optimization side of the work.
Target the queries where AI Overviews actually appear
AI Overviews trigger most on informational and question-shaped queries, less on pure navigational ones. Build content around the real questions your buyers ask before deciding, and phrase headings as those questions. That is where the Overviews are, and where inclusion is worth having.
A realistic expectation
One honest caveat, because it changes how you judge success. Being included in an AI Overview does not send the traffic a top ranking used to. Click-through to the page below an Overview falls by around 58% when one appears. So AI Overview inclusion is primarily a visibility and authority win: your brand and claim are put in front of the user and associated with the answer, even if fewer of them click. Measure inclusion and citation, not just sessions, or you will conclude, wrongly, that it is not working.
Frequently asked questions
Want to know which queries in your category trigger an AI Overview, and whether you are in it? Get a free AI Visibility Audit.
Sources
- Google Search Central, "Optimizing your website for generative AI features on Google Search" (May 2026) — https://developers.google.com/search/docs/fundamentals/ai-optimization-guide
- Ahrefs, "Update: 38% of AI Overview Citations Pull From The Top 10" (March 2026) — https://ahrefs.com/blog/ai-overview-citations-top-10/
- Ahrefs, "76% of AI Overview Citations Pull From the Top 10" (July 2025 baseline) — https://ahrefs.com/blog/search-rankings-ai-citations/
- Ahrefs, "Update: AI Overviews Reduce Clicks by 58%" (300K keywords, Dec 2025 data) — https://ahrefs.com/blog/ai-overviews-reduce-clicks-update/
- Ahrefs, "An Analysis of AI Overview Brand Visibility Factors (75K Brands)" (mentions vs backlinks) — https://ahrefs.com/blog/ai-overview-brand-correlation/
- Otterly, "Does Schema Markup Really Impact AI Search?" — https://otterly.ai/blog/schema-markup-real-impact-ai-search/
- Search Engine Journal, "Google Drops FAQ Rich Results From Search" (deprecation, May 2026) — https://www.searchenginejournal.com/google-drops-faq-rich-results-from-search/574429/
- Aggarwal, Pranjal, et al. (2024), "GEO: Generative Engine Optimization" (information gain), arXiv:2311.09735 — https://arxiv.org/pdf/2311.09735