Google documents how nosnippet, max-snippet, and Google-Extended control your brand in AI results
Google updated its Search Central documentation on AI Overviews and AI Mode, detailing which tags control whether and how a site's content appears summarized within AI-generated responses.
On December 31, 2025, Google updated its Search Central page dedicated to explaining how AI Overviews (AI-generated summaries at the top of search results) and AI Mode (conversational answers for complex queries) work.
The document doesn't introduce a new feature: it organizes, in one place, controls that already existed in robots.txt and in snippet meta tags, now explicitly applied to the context of generative AI search. For those who handle technical SEO, this is official confirmation that content governance for AI is done with the same tools as always, except with one new, separate control for model training.
What changes in search mechanics
According to the documentation, both AI Overviews and AI Mode use a technique called "query fan-out": instead of answering a single search, the system fires off several related searches on different subtopics and sources to put together the answer. This means a page can be pulled in to make up a summary even if it wasn't the main result for the original query the user typed. In practice, a site's exposure surface increases: it can show up cited in questions adjacent to its core topic, not just in the queries it was directly optimized for.
Google also states that AI Overviews only appears when the system calculates that the summary adds value beyond classic search, which is why it doesn't show up for every query. This matters for anyone measuring impact: you can't predict 100% coverage of a site's keywords, because the decision to generate the summary or not belongs to the model, not to SEO.
The technical requirements are the same as always
To be eligible as a supporting link in AI Overviews or AI Mode, a page only needs to be indexed and qualified to appear with a snippet in regular search, meeting Search's standard technical requirements. There is no additional technical requirement specific to AI. The best practices Google lists are the same as those in any technical SEO audit:
- Making sure crawling is allowed in
robots.txtand in the hosting or CDN infrastructure - Making content discoverable through internal links
- Offering a good page experience (Core Web Vitals fall under this)
- Making sure important content exists in text format, not just as image or video
- Supplementing text with quality images and videos when it makes sense
- Keeping structured data consistent with the visible text on the page
- Keeping Merchant Center and Business Profile information up to date
In short: there is no parallel "SEO for AI" checklist. Whoever already does solid technical SEO is already in the race; the differentiator now is actively deciding what to leave out.
The tags that decide visibility in AI
The document separates two types of controls. One limits what appears in Search overall, including AI summaries; the other specifically limits the use of content to train and ground ("grounding") other Google AI systems.
| Control | What it does | Where it applies | Expected effect |
|---|---|---|---|
nosnippet | Blocks any text, image, or video snippet as a page preview | Meta tag or HTTP header on the page | Page can remain indexed, but without a snippet or AI summary |
data-nosnippet | Marks a specific HTML section so it isn't used as a snippet | Attribute on HTML elements (span, div) | Protects sensitive paragraphs without taking the whole page down |
max-snippet | Sets a character limit for the displayed snippet | robots meta tag | Reduces how much literal text can be reused in a summary |
noindex | Removes the entire page from results | Meta tag or HTTP header | Page doesn't appear in classic search, AI Overviews, or AI Mode |
Google-Extended | Restricts the use of content for AI training and grounding | Token in robots.txt | Doesn't affect crawling or ranking in Search itself |
The most important line in the table, for anyone thinking about GEO, is the distinction between the first four rows and Google-Extended. The first four control what Google can show as a search result, including AI-generated summaries. Google-Extended is a different contract: it affects model training and grounding for products like Gemini, without interfering with indexing or Search ranking position. These are independent decisions and can be combined: you can allow appearing in AI Overviews while, at the same time, blocking the use of content to train a model, or vice versa.
How to measure the effect in Search Console
The documentation confirms a point that generates recurring doubt among growth teams: sites that appear in AI Overviews and AI Mode are counted within overall search traffic in Search Console, listed in the performance report under the "Web" search type, alongside traditional results. There is no separate "clicks from AI" report within Search Console today. This means a drop or rise in impressions/clicks attributed to changes in AI Overviews needs to be investigated by proxy: filtering queries where the feature is known to appear, cross-referencing with Analytics data, and observing visit duration.
Google itself notes that when a click originates from a results page with AI Overviews, quality tends to be higher, with users spending more time on the site. This is data straight from the source, not an independent measurement, and it works as a hypothesis: before optimizing to "appear" in the summary, it's worth measuring whether the click that remains after the summary converts or engages better than the site's average click, using Analytics segments by landing page.
Troubleshooting and propagation time
When preview controls have already been implemented and the content keeps showing up in AI features, Google recommends three steps before opening a support ticket: confirm, using the URL Inspection tool, that the HTML received by Googlebot actually contains the applied tag; wait for the re-crawl, which can take anywhere from a few days to several months depending on that page's crawl frequency; and, if necessary, request an explicit re-crawl through Search Console.
Only after that does it make sense to post in Google's help community. For SEO teams, this reinforces that blocking a snippet isn't instant: the effect only shows up after the next crawl, which can skew any A/B test run in a short window.
The trade-off left for the decision-maker
Blocking a snippet with nosnippet or max-snippet also removes the page from classic search with a snippet, not just from AI summaries: there is no option to "appear normally on Google, but disappear only from AI Overviews." This is a real organic traffic trade-off that the growth team needs to consciously accept before applying the tag in production. Google-Extended, on the other hand, is the opposite: it only affects training/grounding, with no cost to Search visibility, making it the natural choice for anyone who wants to protect the content's intellectual property without losing search traffic.
Translated from the Brazilian Portuguese original · Read the original
Search Console Doesn't Separate AI Overviews Traffic, Google Documentation Confirms
Google Search Central's official documentation, updated in December 2025, confirms that AI Overviews and AI Mode fall into the same Web search type data bucket in Search Console. For those doing GEO in Brazil, this changes how before-and-after measurement works.
