Obility Editorial · · 4 min read
A snippet rule is now an AI answer rule
Most AI visibility checklists start with robots.txt. A second gate sits on the page itself, in the robots meta tag or the X-Robots-Tag HTTP header, and it gets less attention because it was written for search snippets. Google and Microsoft have both tied these rules to their AI answers. A page that blocks or trims its snippet can be crawled and ranked normally and still be left out of AI Overviews, AI Mode or Copilot, or reduced to a bare link inside them.
These rules usually arrive without anyone deciding anything about AI. A legal review asked for snippet limits on a pricing page years ago. An SEO plugin setting was copied into a page template. A CDN rule adds a header to every PDF. Each was a search decision when it was made. It is now also a decision about how AI answers describe you.
Google applies nosnippet and max-snippet to AI Overviews and AI Mode
Google’s robots meta tag specification says nosnippet applies to all forms of search results, naming AI Overviews and AI Mode, and that it also prevents the content from being used as a direct input for those features. Google’s documentation on AI features lists nosnippet, data-nosnippet, max-snippet and noindex as the controls for limiting what is shown from a page, and keeps robots.txt for Googlebot as the control over crawling.
max-snippet is the subtler one. The same specification says a character limit also limits how much of the content may be used as a direct input for AI Overviews and AI Mode. A value of 0 is equivalent to nosnippet. A value such as 50 or 160 keeps the page eligible but hands the model a short excerpt, which is a thin basis for describing you accurately in an answer that compares you with competitors. A value of -1 lets Google choose the length. Google notes one exception: the limit does not interrupt uses the publisher has separately permitted, such as in-page structured data or a licence agreement.
When rules conflict, Google’s specification says the more restrictive one applies. A page carrying max-snippet:50 in its template and nosnippet in a header is treated as nosnippet.
Microsoft splits the decision between NOCACHE and NOARCHIVE
Microsoft reuses two older directives. In its September 2023 announcement for what was then Bing Chat, Microsoft said content tagged NOCACHE may still be included in chat answers, but only its URL, title and snippet will be shown, while content tagged NOARCHIVE will not be included in chat answers or linked from them. A page carrying both is treated as NOCACHE. Both kinds of page still appear in ordinary Bing results, which is why the problem goes unnoticed: the blue link looks fine.
Bing’s current Webmaster Guidelines carry the same rules into Copilot. They state that NOARCHIVE prevents content from being used in Copilot responses and grounding results, that NOCACHE limits Copilot to the URL, title and snippet and reduces citation depth, and that NOSNIPPET and DATA-NOSNIPPET may limit Copilot citation quality. Their advice for content meant for Copilot is to avoid NOCACHE.
Plenty of sites still carry noarchive, added years ago to stop search engines showing cached copies of a page. Under Microsoft’s rules, that old tag now keeps those pages out of Copilot answers.
The rule you need to find may not be in the page source
Viewing the page source finds a meta tag. It does not find an X-Robots-Tag HTTP header, which Google’s specification accepts for the same rules and which is often set by a server, CDN or plugin rather than in the template. Check the response headers too, in your browser’s network panel or with a command line request, on one sample page per template: home, product, pricing, comparison, documentation, blog and any PDFs you want cited.
Then check what the crawler received rather than what your browser shows. Google’s documentation on AI features points to the URL Inspection tool in Search Console for the HTML that Googlebot got. This matters most for data-nosnippet. Google’s specification says the attribute may be read before or after rendering, so adding or removing it with JavaScript gives uncertain results, and that it works on span, div and section elements. Its own example shows that an unclosed div carrying data-nosnippet covers all the content after it, which can turn a small exclusion around a cookie notice into an exclusion of the whole page body.
Decide per template, then wait for the recrawl
The goal is not to remove every restriction. Some content should stay out of AI answers: licensed material, content under contract, or a page that reads badly out of context. The goal is that each rule is a choice someone made with AI answers in mind. For the pages that carry your positioning, pricing explanation and comparisons, that usually means no snippet restriction for Google and no NOCACHE or NOARCHIVE for Bing. Where one block on a page must stay out, data-nosnippet on that block is narrower than a page-level rule.
Changes take effect slowly. Google’s documentation says recrawling can take from several days to several months, depending on how often its systems refresh a page, and that you can request a recrawl. Write down the date of each change so a later shift in citations can be read against it.
Lifting a restriction makes a page eligible, and nothing more
These rules gate eligibility. Removing nosnippet does not make Google cite the page, and removing NOCACHE does not make Copilot recommend you. Whether you are named still depends on how well the page answers the question and on what other sources say about you. The rules above are the ones Google and Microsoft have documented for their own products, so do not assume another assistant reads them the same way.
In Obility, Agent Analytics keeps crawler requests separate from citations because a request shows that a page was fetched, not that it was used. A page with steady crawler requests that never appears among the sources Answer Engine Insights records for your buying questions is a page worth checking for a snippet rule, before anyone rewrites it.