Snippet Controls Are Your AI Visibility Off Switch

One meta robots tag can remove your pages from Google's AI answers while every ranking you track looks completely fine. Google's published eligibility bar for AI Overviews and AI Mode has three parts: a page must be indexed, crawlable by Googlebot, and eligible to appear in Search with a snippet. Snippet controls, meaning the nosnippet and max-snippet directives and the data-nosnippet attribute, govern what Google may show and quote from a page, not where it ranks. That means a page can hold position three for its money query, blue link intact, while being ineligible for every AI answer Google builds. This post covers the mechanism, the four specific ways sites switch themselves off, how to check your own site, and the fix for each case.

Why a preview setting became an AI answer setting

Google's AI features do not answer from memory. AI Overviews and AI Mode are retrieval-augmented generation over the ordinary Search index: the system runs multiple related queries in a process Google calls query fan-out, retrieves ranked pages, and assembles a cited answer from their content. The assembly step is the one that matters here, because the content an AI feature may use is the content Google is allowed to show as a snippet. A page that forbids snippets has opted out of the raw material AI answers are made of.

“To be eligible to be shown in generative AI features on Google Search, a page must be indexed and eligible to be shown in Google Search with a snippet.”

Read that sentence twice, because the failure mode hides in what it does not say. A nosnippet directive does not demote a page. It does not deindex it. Classic rankings and classic traffic dashboards stay green. The page simply leaves the pool of quotable content, and the AI answer gets built from whatever remained in that pool, which usually means a competitor. There is no error, no warning, and nothing in a rank tracker that would ever surface it.

The four ways a site switches itself off

• Nosnippet, the full off switch. A meta robots tag (on either the robots or googlebot meta name) containing nosnippet tells Google to show no text preview for the page at all. No snippet eligibility means no AI eligibility: the page fails Google's third condition outright. max-snippet:0 is the same switch in different spelling, since a zero-character budget and a blanket ban are identical in effect.
• Max-snippet between 1 and 99, the dimmer. max-snippet:N caps previews at N characters. The page stays technically eligible, but a 25-character budget leaves room for half a sentence, which starves the answer engine of anything worth quoting. ClarAI's audits flag caps below 100 characters as restrictive.
• Data-nosnippet, the surgical carve-out. An HTML attribute placed on a span, div, or section that excludes exactly that passage from previews. It is the one control designed for precision, and used deliberately it is legitimate. Used carelessly, a data-nosnippet wrapped around your pricing table, your comparison chart, or your product summary carves your most quotable material out of every AI response while the rest of the page stays eligible.
• Noindex pages sitting in your sitemap, the contradiction. A noindex directive fails the first condition, indexed, so the page is disqualified before snippets even come up. When that same URL is listed in your sitemap, you are advertising a page to Google that you have simultaneously banned from the index. One of the two signals is wrong, and it is usually the noindex: a staging setting that shipped, or a plugin toggle nobody remembers.

Why the off switch is usually an accident

These directives were not designed with AI answers in mind, because they predate AI answers by years. Google introduced max-snippet and data-nosnippet in 2019 as content preview controls, and nosnippet is older still. Whatever a site's preview settings were then, they usually still are, and nobody configuring a preview in 2019 was pricing in that the preview would one day become the answer.

The delivery mechanism is almost always a setting rather than a decision. SEO plugins expose robots directives as toggles and can apply them site wide, so one checkbox set years ago can stamp nosnippet or a snippet cap onto every template. Themes and boilerplate get copied between projects with old meta tags intact. Directives can also arrive through the X-Robots-Tag HTTP header, set in server config or at the CDN, where no template search will ever find them. And noindex has its own classic: the staging-site setting that launched with the site. None of these are exotic failures. They are settings that outlived the person who set them.

How to check your own site

• View source on your money pages and search for meta tags named robots and googlebot. Read the full content value: nosnippet, max-snippet, and noindex are all tokens in the same place.
• Check response headers too. Fetch a few key URLs with curl -I and look for X-Robots-Tag, which carries the same directives at the server or CDN level.
• Open your sitemap, then spot-check the listed URLs for noindex. Any URL carrying both is a contradiction to resolve.
• If you run WordPress or a similar CMS, open your SEO plugin's settings and look for site-wide robots, snippet, or preview options before auditing page by page.
• In Search Console, URL Inspection shows the indexing state Google actually holds for a page, which settles any argument about what Googlebot saw at last crawl.

Or let tooling do the sweep. ClarAI's free AI Readiness Grader at /grade checks snippet eligibility on your domain as its single heaviest-weighted check, 25 of the 100 points, alongside Googlebot access, AI crawler policy, and structure signals. The report is free with no signup. Inside the app, the Site Audit crawls your sitemap and checks every page: a snippet-blocked page is raised as a critical finding, a noindexed page still listed in your sitemap as high, a restrictive max-snippet cap as medium, and data-nosnippet usage as a low-severity note asking you to confirm it is intentional. Each finding is cross-referenced against the AI prompts you monitor, so a blocked page shows up next to the questions it was supposed to win.

The fix, case by case

1. Nosnippet: edit the meta robots tag and delete the nosnippet token. If the tag contains nothing else, remove the whole tag. No directive is the default, and the default is fully snippet-eligible.
2. Max-snippet: remove the directive, or set max-snippet:-1, which explicitly means no limit. If some cap is genuinely required, anything from 100 characters up clears the threshold ClarAI flags as restrictive, though any cap still limits what an answer can quote.
3. Data-nosnippet: remove the attribute from the passages you want AI answers to quote, your pricing, your comparisons, your product descriptions, and keep it only on content you never want quoted out of context.
4. Noindex in sitemap: decide which signal is telling the truth. If the page should be findable, remove the noindex. If it genuinely should not be indexed, remove it from the sitemap so you stop advertising it.
5. Sweep the header layer: check server config and CDN rules for X-Robots-Tag directives that a template search will not show you.
6. Then wait for a recrawl. Directive changes take effect when Googlebot next fetches the page, so request reindexing of key URLs through Search Console's URL Inspection and expect days rather than minutes.

When the off switch is the right call

Honesty requires the other side: these controls exist for good reasons, and the goal is not zero snippet directives. Paywalled content, legal disclaimers, and text that would mislead when quoted out of context are all legitimate candidates for nosnippet or data-nosnippet, which is exactly why ClarAI's audit copy says to remove a directive unless the exclusion is intentional. The standard to hold your site to is simple: every snippet control should be a decision someone can defend today, not a setting that outlived its author. The off switch is fine. The off switch nobody remembers flipping is the problem.