noindex vs nosnippet vs data-nosnippet: what each one removes from AI answers

All three controls tell a search engine to show less of your content. They differ in how much they remove and what stays eligible. Google and Bing document them as the on-page controls for limiting what appears in AI features.

Here is what each looks like in the page source:

<meta name="robots" content="noindex">
<meta name="robots" content="nosnippet">
<meta name="robots" content="max-snippet:0">
<div data-nosnippet>Internal pricing note</div>

The definitions of nosnippet, max-snippet and data-nosnippet cover the syntax in more detail.

Google's AI features documentation (last updated December 2025) sets one rule for AI Overviews and AI Mode. A page must be indexed and eligible to be shown in Google Search with a snippet to appear as a supporting link. Google adds that there are no additional technical requirements.

That rule makes the two page-level controls easy to reason about:

• noindex fails the first condition. The page leaves the index, so it can't be a supporting link.
• nosnippet and max-snippet:0 fail the second condition. The page stays indexed and can still appear as a plain result, but it is not snippet-eligible, so it can't be a supporting link in AI Overviews or AI Mode.

Eligibility is not inclusion, though. Google also states that indexing and serving isn't guaranteed. Removing a stray nosnippet makes a page eligible again. It does not make Google pick it.

Two Google controls sit outside these tags. Google-Extended in robots.txt governs Gemini Apps training and grounding, and Google says it does not affect inclusion in Search. And Search Console has a generative AI inclusion control that is opt-out by default and can inherit from a parent property. Google's help doc still described it as rolling out to a subset of owners, so check whether your property has it.

data-nosnippet is the only control that works below the page level. You put the attribute on a specific element, and engines skip that element when they build snippets and AI summaries.

Bing added support on October 15, 2025. Its announcement says content marked with data-nosnippet "is still indexed normally, but it will be excluded from snippets and AI summaries." The same page says the tagged content "is available for ranking." So the marked text still helps the page match queries. It just never appears as quoted text. Bing's data-nosnippet announcement, digested covers the details.

This makes data-nosnippet the right tool for a single block you don't want reused: a draft price, a legal disclaimer you don't want quoted out of context, or internal notes on a public page. If you want Copilot specifically, see whether data-nosnippet keeps content out of Copilot answers.

There is no opposite attribute. Nothing marks a section as more quotable. data-nosnippet is a suppression tool only.

The meta tags above are documented by Google and Bing. The evidence we work from has no first-party statement from OpenAI, Perplexity or Anthropic saying their search features honor noindex or nosnippet. Don't assume they do.

For those engines, the documented control is robots.txt, set per bot:

• OpenAI says sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers, though they can still appear as navigational links. Blocking GPTBot only signals that content should not be used for training. OpenAI states each setting is independent of the others.
• PerplexityBot handles Perplexity's search index. Claude-SearchBot handles Claude's search quality.
• User-triggered fetchers are different. OpenAI says robots.txt rules may not apply to ChatGPT-User, and Perplexity says Perplexity-User generally ignores robots.txt.

So noindex is not a reliable way to keep a page out of every AI answer. Whether noindex keeps a page out of ChatGPT and Perplexity walks through each engine.

These controls usually go wrong by accident, not by choice. A typical case: a developer adds data-nosnippet or nosnippet to hide a cookie banner, the tag lands in a shared template, and every page built on that template drops out of snippets and AI answers. A staging noindex that ships to production does the same at a larger scale.

The failure is silent. Nothing in your CMS warns you, and the pages still load normally for visitors. Page-level controls can also arrive through an X-Robots-Tag HTTP header, which never shows in the HTML. Read how X-Robots-Tag headers affect AI search if your server or CDN sets headers.

What to check:

1. Fetch the raw HTML of one page per template and search for noindex, nosnippet, max-snippet and data-nosnippet.
2. Check the response headers of the same URLs for X-Robots-Tag.
3. For every hit, confirm someone chose it on purpose.
4. Re-check after template or plugin changes.

The site-wide snippet controls audit turns this into a repeatable routine.

Pick the smallest control that meets the goal.

• The page should not be in search at all (thank-you pages, internal search results, staging): use noindex. It removes the page from that engine's index and therefore from its AI Overviews and AI Mode.
• The page can rank, but no text should be quoted (paywalled or licensed content): use nosnippet or max-snippet:0. In Google, this also removes AI Overviews and AI Mode supporting-link eligibility.
• You only want to limit how much is quoted: use max-snippet with a character count.
• One block on a public page must not be quoted: use data-nosnippet on that element. The page and the section stay indexed and rankable.
• You want out of ChatGPT, Perplexity or Claude search: disallow OAI-SearchBot, PerplexityBot or Claude-SearchBot in robots.txt, and accept that user-triggered fetchers may not follow it.
• You want content in AI answers: remove any accidental controls. That restores eligibility. It does not guarantee anyone cites you.

For a step-by-step approach across engines, see how to keep specific content out of AI answers.