Google's AI Overviews and AI Mode: what site owners can actually control
AI-generated answers now sit at the top of many Google searches, and Google's AI Mode offers a conversational search experience built on the same technology. Site owners have reasonable questions: how is my content used, can I control it, can I measure it, and do I need to do anything differently?
Google has published direct answers, most recently in its guide to optimizing for generative AI features on Google Search, last updated in July 2026. This article summarizes what that guidance says, what Search Console now offers, and where the popular advice goes wrong.
How Google's AI features choose sources
According to Google, its generative AI features are rooted in its core Search ranking and quality systems. They don't use a separate index. When an AI Overview or an AI Mode answer is generated, it draws on pages that are already indexed and links to sources so people can read more. Google has also described a technique called query fan-out, where several related searches are run behind the scenes to gather a broader set of results.
The practical consequence: the pages that are eligible and competitive in regular Search are the same pages AI features can draw on. There is no side door.
The eligibility requirements
Google lists two requirements for a page to appear in its generative AI features:
- The page must be indexed and eligible to be shown in Google Search with a snippet. Anything that prevents indexing or snippets, such as
noindexornosnippet, also prevents use in AI features. - The site must be included in Search generative AI features in Search Console. Search Console now has a Search generative AI control with three options: include, exclude, or inherit from a parent property. Including is the default for all properties, and Google rolled the control out to all websites worldwide on August 31, 2026.
Google notes that the control is not a ranking signal, doesn't affect indexing or the rest of Search, and doesn't control model training. It is simply the switch for whether your links and content appear in AI Overviews, AI Mode and generative AI features in Discover.
Meeting the requirements makes a page eligible. It doesn't guarantee it will be used, any more than being indexed guarantees a top ranking.
The controls you have
| If you want to… | Use |
|---|---|
| Keep your whole site out of Google's generative AI features | Search Console's generative AI control, set to exclude |
| Stop a page's text being shown as a snippet or in AI features | nosnippet robots meta tag |
| Keep specific passages out of snippets and AI features | data-nosnippet attribute on those elements |
| Limit how much text can be shown | max-snippet robots meta tag |
| Keep a page out of Google entirely | noindex |
| Limit use of your content for training and grounding in some of Google's other AI systems | Google-Extended in robots.txt |
Two clarifications are worth making. Google-Extended doesn't affect Google Search or your inclusion in AI Overviews and AI Mode; Google says robots.txt rules for Googlebot are the control for Search access. And every one of these controls has a cost: limiting snippets also limits how your page is presented in regular results.
How to measure it
There are two places to look in Search Console:
- The Performance report, Web search type. Google says clicks and impressions from AI features such as AI Overviews and AI Mode are included in overall search traffic, within the Web search type.
- The generative AI performance report. A dedicated report showing how often links to your site were shown in generative AI features on Google Search. It currently reports impressions only, for AI Overviews and AI Mode, grouped by page, country, date and device. Discover has its own separate report.
Because the dedicated report shows impressions without clicks, pair it with your analytics: look at landing pages that gained or lost AI feature impressions and check how their organic sessions and conversions changed.
Other assistants, such as ChatGPT, Claude and Perplexity, show up in analytics as referral traffic when people click through. Manual checks of how assistants answer questions in your field add useful context, but they are samples, not measurement.
What Google says you don't need to do
Google's guide includes a list of things you can ignore. You don't need to:
- create llms.txt files or other new machine-readable files, AI text files or special markup;
- chunk your content into small pieces for AI;
- write in a special way or add phrasing variations just for generative AI search;
- chase inauthentic mentions of your brand across the web;
- add special structured data. Structured data isn't required, and there's no special schema.org markup for these features.
If a vendor is selling any of these as the key to AI visibility, ask how that squares with Google's own documentation.
What does help
Google's recommendations are the fundamentals of helpful content:
- Unique, non-commodity content. Google contrasts generic advice with content based on first-hand experience. Pages that add something other pages don't are the ones worth citing.
- Clear organization. Descriptive headings and well-structured sections help readers and machines find answers.
- High-quality images and videos where they genuinely help.
- Technical eligibility. Crawlable, indexable pages that are eligible for snippets.
- No scaled shortcuts. Mass-producing pages to cover every variation of a question can break Google's scaled content abuse policy.
Don't forget the other AI assistants
Other AI companies run their own crawlers, and most separate search from training. OpenAI uses OAI-SearchBot for ChatGPT search and GPTBot for training. Anthropic uses Claude-SearchBot for search and ClaudeBot for training. Blocking a training crawler doesn't remove you from that company's search features; blocking its search crawler does.
Check two things: that your robots.txt reflects a deliberate decision for each, and that your CDN or firewall bot protection isn't blocking the search crawlers you meant to allow. Bot protection settings are the most common reason a site is invisible to AI assistants without anyone having decided that it should be.
The short version
AI features have changed how search results look, but not what makes a page worth showing. Make sure your important pages are indexable and snippet-eligible, check your Search Console settings and bot protection, write content that adds something only you can add, and measure with the reports Google now provides. That is still SEO, which is exactly Google's point.
