Snippet and content reuse controls

Snippet and content reuse controls explained

Snippet and content reuse controls are page- or section-level instructions limiting what supporting platforms can display or use from a resource. Examples include nosnippet, max-snippet, data-nosnippet, and Microsoft’s page-level ai-grounding preference, announced on October 30, 2026, which will progressively replace its AI-related use of noarchive and nocache. Microsoft announcement.

These controls address a different question from indexability: a page may remain discoverable while some uses of its content are restricted. The exact meaning depends on the platform. A familiar directive name should not be treated as a universal AI policy.

Impact details

Impact: High. Some settings can remove direct content input or restrict participation in particular AI answer experiences. Others affect only selected passages. The rating reflects the potential consequence of an unintended broad restriction, not a prediction that longer previews always generate more citations.

Google documents these rules:

• nosnippet prevents direct content input to AI Overviews and AI Mode.
• max-snippet limits that input; 0 is equivalent to nosnippet, while -1 leaves length selection to Google. Separately permitted uses can be exceptions.
• data-nosnippet excludes selected text from snippets, using supported HTML elements.
• Google Search ignores noarchive and nocache.

Google robots specifications.

Google’s AI guidance includes snippet controls among the ways to limit information in its Search AI features. Supporting-link eligibility requires an indexed page eligible for a snippet. Google’s AI features guidance.

Microsoft’s legacy guidelines say NOARCHIVE prevents content use in Copilot responses and grounding results; NOCACHE limits Copilot to the URL, title, and snippet. They describe NOSNIPPET and DATA-NOSNIPPET as caption restrictions that may reduce citation quality. Bing Webmaster Guidelines.

Microsoft will begin honoring AI Grounding on February 1, 2027, as pages are crawled. Allow is the default; DisallowExtended permits grounding in Microsoft-operated experiences but excludes Microsoft grounding services for third parties; Disallow prevents grounding use. Microsoft warns that restrictions reduce AI visibility. These preferences do not change eligibility for traditional Bing search results and are independent of AI Training. Microsoft announcement.

The selected influences reflect an editorial interpretation: Understanding & Retrieval covers what material can enter supported answer pipelines; Mention & Recommendation covers possible downstream citation effects. These instructions do not provide a mechanism for commanding a positive recommendation.

Proof & consensus details

Proof: High for platform-specific controls. Operators explicitly document the relevant behavior. Microsoft additionally states that text marked with data-nosnippet is excluded from Bing snippets and AI summaries while remaining indexed and available for ranking. That distinguishes display restrictions from removal of the page. Bing’s October 2025 announcement.

Consensus: Strong on the existence and importance of these controls within their documented scope. This is not a claim that every provider recognizes the same syntax, or that independent experiments have established universal compliance. The reviewed sources agree that publishers can restrict particular uses; implementation differences are product differences rather than conflicting evidence about one shared rule.

Microsoft’s original 2023 announcement specified that, when both NOCACHE and NOARCHIVE are present, it treats the combination as NOCACHE. It also distinguished prospective training restrictions from answer display. This historical detail matters when diagnosing inherited settings; this is legacy behavior, not the interpretation of the new AI Grounding preference. Microsoft’s original announcement.

Announced behavior, not yet an observed rollout: Microsoft's October 30, 2026 announcement and Bing Webmaster Tools FAQ document the new scope, syntax, and transition. As of this review, implementation starts on February 1, 2027; these sources establish the announced policy, not independently verified enforcement.

No cited study measures a universal citation increase from removing these controls. The evidence also does not show that a setting erases previously learned information, prevents third-party descriptions of your business, or governs every user-requested fetch. Avoid treating a single successful answer or missing citation as a compliance experiment.

Recommendation

Define the intended tradeoff before editing: public discoverability, permitted excerpts, and use of full content are separate decisions. Remove accidental broad restrictions from pages intended to supply evidence in AI answers, but retain restrictions chosen for editorial or commercial reasons.

Audit shared templates before individual pages. A legacy cache-related setting deserves review because its significance differs between Google and Microsoft. Record the provider, directive, affected content, and desired outcome together; a generic “AI allowed” checkbox obscures meaningful differences.

For Microsoft, choose AI Grounding separately from training and publish it per page through a Bingbot meta tag or HTTP header. There is no robots.txt grounding preference. To preserve existing grounding restrictions, sites using NOARCHIVE/NOCACHE as of December 15, 2026 must publish page preferences and record a temporary Bing Webmaster Tools preference before March 1, 2027. Legacy tags stop expressing training or grounding preferences after that date. The temporary grounding choice lasts until recrawl discovers the new page preference or September 30, 2027, whichever comes first. Transition guidance.

Prefer selective exclusions when only one passage should be omitted. Bing specifically supports keeping the remainder of a page available when sections are marked.

Do not select a numeric snippet limit as an assumed AI ranking trick. Even for ordinary featured snippets, Google gives no universally sufficient length, and a low value does not guarantee exclusion. That feature is distinct from AI Overviews, so it is not evidence of an optimal AI character count.

AI platforms

Google Search AI features: use the documented Search controls; do not extrapolate them to all Gemini services. Google recommends checking the crawler-visible implementation and allowing recrawling after changes. Google.

Bing and Copilot: AI Grounding governs use in generated responses separately from traditional indexing and ranking. Set, for example, <meta name="Bingbot" content="ai-grounding:disallow"> or X-Robots-Tag: Bingbot: ai-grounding:disallow; use ai-grounding:disallowextended to retain Microsoft-operated grounding while excluding third-party grounding services. Unset means Allow. New preferences take effect on crawl from February 1, 2027; publishing alone does not activate them. They progressively replace NOARCHIVE/NOCACHE for AI uses, not their original traditional-search meanings. An Allow preference does not override a Bingbot crawl block. Microsoft implementation guide.

Gemini: Google-Extended controls specified model training and grounding uses. It is separate from Google's Search snippet controls, so do not assume those controls have the same effect in Gemini Apps. Google's crawler reference.

ChatGPT: OpenAI's crawler guidance does not establish support for Google's specific nosnippet or max-snippet directives in ChatGPT answers. Use OpenAI's documented controls for the intended use. OpenAI's crawler documentation.

Claude: Anthropic documents that noindex can keep content out of search-partner results supplied to Claude web search, while other access routes may remain. This does not establish identical handling of every Google snippet directive. Anthropic's removal guidance.

Perplexity: its crawler guidance does not establish equivalent support for every Google or Bing snippet directive. Check the relevant provider's current policy before relying on a restriction. Perplexity's crawler documentation.

Audit instructions

1. Collect representative URLs. Select pages whose full content should be reusable and pages with intentional restrictions. Write down the desired outcome before assessing compliance.
2. Inspect both locations. Check HTML robots meta tags and X-Robots-Tag response headers.
3. Inspect section boundaries. Find data-nosnippet and check how much content it wraps.
4. Interpret by provider. Record nosnippet, numeric limits, legacy Microsoft noarchive/nocache, and page-level ai-grounding values separately; distinguish grounding from training and traditional search eligibility. Flag intentional restrictions distinctly from publishing mistakes.
5. Check for template spread. Compare at least one page from each major content type and language. A correct homepage does not establish that article or document templates are correct.
6. Check the Microsoft transition. Record whether legacy tags are in use as of December 15, 2026, the new page preference, and the temporary Bing Webmaster Tools choice where applicable. Verify the March 1 deadline and September 30 expiry against the official FAQ.
7. Validate after processing. With owner access, compare implementation dates with the last crawl time. Track resulting excerpts and citations separately; a configuration change is not proof of a visibility gain.