How to Control What Google and AI Can Quote From You

Somebody on a forum told you to add nosnippet so Google stops 'stealing your content.' Somebody else told you Google-Extended blocks AI Overviews. Both pieces of advice are wrong in ways that cost you money, and they are wrong in opposite directions.

These four directives are not variations of the same switch. They target different systems, they are read at different moments, and only one of them is safe to use on a page that earns you leads. Here is what each one actually does, straight from Google's robots meta tag specification rather than from forum folklore.

What do nosnippet, max-snippet, data-nosnippet, and Google-Extended each actually do?

They control four separate things: all snippets, snippet length, one element, and Gemini training. Only the first three touch what Google shows on the SERP.

Two of these live in the page HTML or HTTP header. One lives on a single element. One lives in robots.txt and never touches Search at all. Mixing them up is how an owner ends up invisible in the exact SERP feature they were trying to win.

Does nosnippet also remove you from AI Overviews?

Yes. Google's robots meta tag documentation says nosnippet applies to all forms of search results — web search, Google Images, Discover, AI Overviews, AI Mode — and 'will also prevent the content from being used as a direct input for AI Overviews and AI Mode.'

So the popular advice to 'add nosnippet to keep AI from using your content' does exactly that, and takes your featured snippet, your regular SERP description, and your AI Overview citation with it. You do not get a quiet, invisible opt-out. You get deleted from the description text of your own listing.

Google's AI features documentation is blunt about this: the controls available are nosnippet, data-nosnippet, max-snippet, and noindex. There is no AI-Overviews-only opt-out. Opting out of AI features means opting out of Search snippets. That is the trade, and there is no version of it where you keep the snippet and lose the AI citation.

Does Google-Extended stop you appearing in AI Overviews?

No. Per Google's crawler documentation, Google-Extended is a robots.txt token that manages whether crawled content is used for training future Gemini models and for grounding answers in Gemini Apps and Vertex AI — and Google states plainly that 'Google-Extended does not impact a site's inclusion in Google Search nor is it used as a ranking signal in Google Search.'

AI Overviews are a Google Search feature. They are served off the same index Googlebot builds. Disallowing Google-Extended does not remove you from that index, so it does not remove you from AI Overviews. Owners add it, feel protected, and are still being summarized the next morning.

OpenAI splits its bots the same way. OpenAI's crawler docs state that GPTBot crawls content that may be used to train its foundation models, while OAI-SearchBot surfaces sites in ChatGPT's search features — and that sites opted out of OAI-SearchBot 'will not be shown in ChatGPT search answers.' The settings are independent. Blocking GPTBot is a training decision. Blocking OAI-SearchBot is a visibility decision, and it is the one that costs you customers.

If you want the full argument on which crawlers a service business should and should not block, we wrote it up in should you block AI crawlers, and the related question of whether a manifest file helps in llms.txt explained.

How do you hide one paragraph without killing the whole page's snippets?

You wrap that one paragraph in data-nosnippet. Google's spec allows the attribute on span, div, and section elements only — put it on a custom tag or a paragraph tag and it is ignored.

This is the surgical tool almost nobody uses. A pricing table you do not want lifted out of context, a proprietary comparison chart, an internal rate card on a public page: wrap it, and the rest of the page stays fully eligible for snippets, featured snippets, and AI Overview citations.

Four gotchas from the spec, in the order they bite:

• The attribute is boolean. Any value you set is ignored — even data-nosnippet='false' hides the content. There is no off switch except removing the attribute.
• An unclosed div swallows everything after it. Google's docs spell this out: an unclosed element will include all content afterwards, so a broken tag can silently mute the back half of your page.
• Do not add or remove the attribute with JavaScript. Rendering is not guaranteed, extraction can happen before or after render, and the result is a coin flip.
• nosnippet outranks data-nosnippet. If both appear on a page, Google's featured snippets documentation says nosnippet takes priority and no snippets show at all.

One useful exception: structured data stays usable for search results even when it is declared inside a data-nosnippet element. Which schema types actually earn you anything is a separate question — we cover it in schema markup for GEO.

When should a service business ever suppress a snippet?

Almost never — and when you do, use data-nosnippet on one element, not nosnippet on the page. In every case we would sign off on, the thing being protected is a specific block of proprietary data, not the page that sells the service.

Legitimate reasons we would accept: a rate table competitors scrape and undercut hourly, a proprietary dataset you license, a client-only resource that must stay indexed for navigational search but must not be quoted, or a legal disclosure that reads badly out of context.

Not legitimate: 'AI is stealing my content.' Your service pages exist to be found by people who need a plumber, a lawyer, or a dentist this week. A page nobody can be shown a preview of is a page nobody clicks. If you want to opt out of being quoted at all, the honest version of that decision is noindex, and you should say so out loud before you do it.

If you want to appear in featured snippets rather than hide from them, max-snippet is worth knowing in reverse. Google's featured snippets documentation says the shorter your max-snippet value, the less likely the page appears as a featured snippet — and it explicitly declines to publish an exact minimum length, because the minimum is variable and depends on factors including the information in the snippet, the language, and the platform. Setting max-snippet to -1 hands the length decision back to Google.

What happens to your brand impressions if you opt out?

You drop into the worst bucket on the SERP: present but uncited. Seer Interactive analyzed 5.47 million queries across 53 brands and found that being cited inside an AI Overview delivers 120% more organic clicks per impression than appearing on the same SERP uncited — though still 38% fewer clicks than a SERP with no AI Overview at all (Seer Interactive, 2026).

Read that carefully, because the second half is the part people miss. AI Overviews cost you clicks whether you are cited or not. Citation is not a bonus; it is damage control. Volunteering to be uncited via nosnippet is choosing the losing side of that comparison on purpose.

The behavioural data says the same thing from the user's side. Pew Research Center tracked the actual browsing of 900 US adults across 68,879 Google searches and found that when an AI summary appeared, users clicked a search result on just 8% of visits — versus 15% when no AI summary was present (Pew Research Center, March 2025 data). Fewer clicks are available. Being the cited source is how you keep a share of the ones that remain.

The scale is real but not apocalyptic: Ahrefs analyzed 146 million SERPs and found AI Overviews appear on 20.5% of keywords — roughly one in five (Ahrefs, September 2025 desktop data). Four SERPs in five are still classic blue links, where a suppressed snippet just makes your listing look broken.

How do you check whether a directive is already live on your site?

Check three places, in this order, and check the rendered response rather than the plugin settings screen. Directives arrive from more places than most owners expect: an SEO plugin's advanced tab, a theme, a staging config that shipped to production, or a CDN rule nobody documented.

• The page HTML. Search the rendered source for a robots meta tag and for data-nosnippet. Do not trust the CMS preview — read the source Google reads.
• The HTTP response headers. X-Robots-Tag can carry every rule the meta tag can, and it is often set at the server or CDN, which is exactly why nobody finds it. Fetch the headers for a live URL and look for it.
• robots.txt. Look for Google-Extended, GPTBot, and OAI-SearchBot user-agent groups. A blanket disallow that someone pasted from a blog post can be quietly cutting you out of ChatGPT search answers.
• Search Console URL Inspection. It reports what Googlebot actually saw, which settles arguments that source-code guesswork cannot.

One trap that catches good developers: if a URL is disallowed in robots.txt, Google never crawls it, so it never sees your robots meta tag at all. Google's spec states that indexing and serving rules on a disallowed URL will not be found and will therefore be ignored. Blocking the crawl does not enforce your directive — it deletes it. Same logic applies to noindex, which is why a disallowed page can still show up in results.

If you have never audited what your site is emitting, a rogue nosnippet is one of the cheapest, highest-value things to find. Our technical SEO service starts with exactly this sweep, and the DIY version is in our technical SEO audit walkthrough.

The short version: leave your service pages fully quotable, use data-nosnippet on the one table you genuinely need to protect, and stop treating Google-Extended as an AI Overview kill switch — it isn't one. If you want a second pair of eyes on what your site is currently telling Google and the AI engines, Get my free audit.