Nosnippet and data-nosnippet control previews without blocking a page because they are presentation directives, not access rules. A nosnippet robots meta tag suppresses the text snippet for a whole page while the URL can still be indexed and shown. The data-nosnippet HTML attribute excludes only the marked text inside span, div or section elements. Crawlers must still be allowed to fetch the page for either setting to be read.
What nosnippet and data-nosnippet actually do
Google's robots meta tag, data-nosnippet and X-Robots-Tag specifications describe two levels of control. Page-level settings go in a robots meta tag in the HTML head or in an X-Robots-Tag HTTP header. Text-level settings use the data-nosnippet attribute on HTML elements inside the page.
The distinction matters because the two tools answer different questions. Nosnippet answers: should this page show a text snippet at all? Data-nosnippet answers: which parts of this page should never be used as snippet text?
Neither tool blocks a page. A noindex rule tells search engines not to show the page in results. Nosnippet keeps the page eligible while limiting its preview. That is why nosnippet is often described as a presentation control rather than an indexing control.
The same document notes that these settings can only be read and followed if crawlers are allowed to access the pages that include them. If robots.txt blocks Googlebot, the directive is never seen. If a page is behind a login, the crawler cannot read the attribute. Access comes first; presentation control comes second.
How does data-nosnippet differ from nosnippet?
Data-nosnippet is a text-level attribute. Google's documentation says it can be placed on span, div and section elements. The marked text is excluded from snippets, but the rest of the page remains eligible for snippet generation.
Nosnippet is page-level. It applies to the whole page and suppresses the text snippet. The page can still appear as a result, and other preview elements may still be shown depending on the directives in place.
A useful way to think about it: data-nosnippet is a scalpel, nosnippet is a switch. If only a price table, a legal disclaimer or a user-generated comment block should stay out of previews, data-nosnippet is the narrower tool. If the whole page should not produce a text snippet, nosnippet is the page-level option.
Both are read at crawl time. If the page is already indexed, a change to the directive is not applied instantly. Google must recrawl the page before the new setting takes effect. The delay depends on crawl scheduling, page importance and how often the URL is refreshed. There is no guaranteed timeframe in the official documentation.
What about max-snippet, max-image-preview and max-video-preview?
These are page-level directives that set limits rather than removing previews. Max-snippet controls the maximum length of a text snippet. Max-image-preview controls the size of an image preview. Max-video-preview controls the length of a video preview.
They are often used together with nosnippet logic. For example, a publisher might allow a short text snippet but limit image previews. Another site might allow a full snippet but restrict video preview length. The official specification lists these as separate values, and conflicting rules favour the most restrictive combination.
A practical diagnostic is to check which directive is actually present in the rendered HTML and in the HTTP header. A CMS may inject a robots meta tag that overrides a template-level setting. An X-Robots-Tag header may add a rule that is not visible in the page source. Both need to be checked.
Do these settings stop AI features or model training?
This is where the evidence is narrower than many summaries suggest. The official Google documentation covers how Google presents content in search results. It describes page-level and text-level settings for search presentation. It does not state that nosnippet or data-nosnippet prevents a page from being used for model training, nor does it list every AI feature that might use page content.
The documentation does note that a noindex rule applies to search engine crawlers, and that blocking non-search crawlers such as AdsBot-Google may require rules targeted to the specific crawler. That is a reminder that different crawlers can have different rules. It is not a statement that nosnippet controls training or all AI systems.
So the honest position is: nosnippet and data-nosnippet are documented controls for search result previews. Their effect on AI features and training is not established by the cited specification. Anyone claiming otherwise should be asked for a source that names the specific system and the specific rule.
A diagnostic for preview control
Use this sequence when a page shows a preview you do not want, or when you want to change one.
- Confirm crawl access. Check robots.txt and any server-level blocks. If Googlebot cannot fetch the page, the directive cannot be read.
- Inspect the rendered HTML. Look for a robots meta tag in the head and for data-nosnippet attributes on elements.
- Inspect HTTP headers. Check for an X-Robots-Tag header that may add or override a rule.
- Identify the level. Is the problem the whole page or a specific block? Choose nosnippet for the page, data-nosnippet for the block.
- Check for conflicts. If multiple rules apply, the most restrictive one wins according to the specification.
- Request a recrawl if the page is already indexed. Use the appropriate search console tool for the property. Expect a delay, not an instant change.
- Recheck the live result after recrawling. Compare the preview against the directive you set.
Example (hypothetical): a UK retailer marks a delivery-estimate block with data-nosnippet because the dates change weekly. The rest of the product page remains eligible for snippets. If the retailer instead wanted no text snippet for the whole page, nosnippet would be the page-level choice. These are illustrative scenarios, not measured outcomes.
How does this connect to commerce and AI channel work?
Preview control is one part of a wider data-quality picture. If a product feed or catalog sends inconsistent data to an AI channel, the search preview is only one surface where that inconsistency appears. Our analysis of Shopify adding Meta as an AI channel looks at catalog data quality checks that sit alongside preview decisions. Our breakdown of AI Max previews and the Search Ads journey report examines what data is promised in that reporting context. Neither replaces the need to verify the actual robots directive on the page.
For teams in the UK and US, the mechanics are the same because the directives are part of the HTML and HTTP specification, not a country-specific rule. Local obligations such as tax or consumer information are separate questions and should not be confused with preview control.
What remains uncertain
Recrawl timing is not guaranteed. The official documentation does not promise a specific interval between changing a directive and seeing the result. Crawl budget, page importance and site health all influence scheduling.
AI feature behaviour is not fully specified in the cited document. It describes search presentation controls. It does not enumerate every AI surface or training pipeline. Treat claims about training exclusion as unverified unless a named official source supports them.
CMS behaviour varies. A platform may expose a search engine settings page rather than raw HTML editing. The rendered output is what matters, not the template intention.
Follow-up questions
Does data-nosnippet stop a page from being indexed?
No. Data-nosnippet excludes marked text from snippets. It does not remove the page from the index. A noindex rule is the directive that addresses indexing.
How long does a nosnippet change take to appear?
The official documentation does not state a fixed timeframe. The page must be recrawled before the new directive is read. The delay depends on crawl scheduling and is not guaranteed.