What Google actually says about AI Overviews and AI Mode

Key takeaways

• Google says existing SEO practices remain relevant, and no special AI file or schema.org markup is required for AI Overviews or AI Mode. [1][2]
• Google says its dedicated Generative AI performance report and Search generative AI control reached worldwide availability by Aug. 31, 2026, but the report may not appear without enough impressions, and control settings can be inherited. [5][6]
• Google describes its Oct. 21, 2026, documentation update as adding existing control options, not introducing a new control. [4]

About this guide: This guide describes Google’s documentation as of Oct. 29, 2026. Sources were checked against the pages as published on Oct. 29, 2026.

What does Google mean by AI Overviews and AI Mode?

Google describes AI Overviews as an AI-generated snapshot with information and links inside search results. [13] AI Mode supports AI-generated responses, follow-up questions, and links for further exploration. [14] According to its help page, AI Mode divides a question into subtopics and searches for them simultaneously. [14]

As of Oct. 29, 2026, Google’s availability lists for both features include the US, the UK, Germany, France, and Australia, and both list English as a supported language. [13][14] Both help pages warn that settings described there may not match every user’s experience while Google updates Search services settings. [13][14]

The two features need not return the same responses or links. [1] Google’s developer documentation says: “AI Mode and AI Overviews may use different models and techniques, so the set of responses and links they show will vary.” [1, “How AI features work in Search”] Google also says AI Overviews appear only when its systems determine that they add to classic Search. [1]

Google’s documentation updates log records the addition of a separate AI Overviews page on May 14, 2024, and new documentation on AI features and websites on May 21, 2025. [4] These are documentation dates, not a claim about when every related feature became available.

What does it take to be eligible?

According to Google’s documentation as of Oct. 29, 2026, a page must be indexed and eligible to appear in Search with a snippet. [1][2] Its optimization guide also requires the site to be included in Search generative AI features through Search Console. [2]

The AI features page states: “There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary.” [1, introduction] The optimization guide, last updated July 10, 2026, separately documents the Search Console inclusion requirement. [2] Both pages caution that meeting requirements and following best practices do not guarantee crawling, indexing, or serving. [1][2]

Google’s May 21, 2025, guidance tells site owners to ensure that Googlebot is not blocked, the page returns a successful HTTP response, and the page contains indexable content. [11] It summarizes the relationship this way: “Meeting the technical requirements covers you for search generally, including AI formats.” [11, “Ensure we can access your content”]

Do not assume that an untouched property is included. Google’s control help page identifies Include as the default, but says properties with a parent default to inheriting their parent’s control. [5] A child property can therefore inherit an exclusion, and its owner can override the inherited setting. [5]

Google also identifies several things that are not required:

• New machine-readable files, AI text files, or special schema.org markup are unnecessary for appearing in these features. [1][2]
• Google says maintaining llms.txt files neither helps nor harms Search visibility or rankings because Search ignores them. [2]
• There is no requirement to divide content into small pieces for AI systems, and Google recommends choosing page length for the audience and subject. [2]
• The guide discourages pursuing inauthentic mentions and says structured data is not a special requirement for generative AI search. [2]

What about answer engine optimization, or AEO, and generative engine optimization, or GEO? Google treats optimization for its generative AI search experience as SEO. [2] Its guide also states: “No third-party tool has access to our internal ranking or AI systems.” [2, “Measure visibility in generative AI features with Search Console”]

How does query fan-out work, according to Google?

Google uses query fan-out to describe multiple related searches across subtopics and data sources that can contribute to a response. [1][2] Its AI features page says both AI Overviews and AI Mode may use this technique. [1] According to Google, these searches allow its systems to identify additional supporting pages and present a broader set of links than a classic web search. [1]

The optimization guide also describes retrieval-augmented generation, or RAG, as grounding responses with pages retrieved from Google’s Search index using its core Search ranking systems. [2] Google says its systems review information from those pages to generate a response with supporting links. [2]

That is not an instruction to create a page for every possible related query. Google says creating separate content for query variations primarily to manipulate rankings or generative AI responses violates its scaled content abuse policy. [2] Its spam policies expressly include attempts to manipulate generative AI responses in Search. [12]

The cited guides do not specify a typical number of fan-out queries, a complete query-selection procedure, or market-by-market differences. [1][2]

How can site owners measure AI Overviews and AI Mode?

As of Oct. 29, 2026, Google documents a dedicated Generative AI performance report for impressions in AI Overviews and AI Mode. [6] AI-feature clicks are also included in the standard Search Console Performance report’s overall Web traffic. [1][8] These are different reporting views, not interchangeable metrics.

Google announced the dedicated reports on June 3, 2026, initially for a subset of websites. [7] Its announcement and help page say worldwide rollout was complete by Aug. 31, 2026. [6][7] However, the help page says a missing report usually means the site has not received enough impressions in the relevant features. [6]

The dedicated Search report measures how often links to a site were shown in supported generative AI features. [6] It supports grouping by page, country, date, and device, plus filters for text-based and multimodal Web searches. [6] Dates use Pacific Time, and most page-level performance data is assigned to the canonical URL. [6]

Aggregation matters. If multiple results from the same site appear in a generative AI feature, the property-level chart can count them as one impression, while the page table aggregates by page. [6] Google says these differences can produce discrepancies between chart and table totals. [6] Search Labs experiments are excluded, and Discover has a separate generative AI report. [6]

Google’s counting documentation distinguishes clicks, impressions, and position:

• Clicking an external-page link in AI Mode or an AI Overview counts as a click. [8]
• Standard impression rules apply to both features. [8]
• An AI Overview occupies one position, with all its links assigned that position. [8]
• A follow-up question in AI Mode counts as a new query, with the new response’s metrics attributed to it. [8]

The dedicated report described in Google’s help page shows impressions, not a separate AI-feature click metric. [6] The ordinary Performance report includes AI-feature traffic within overall Web data. [1]

Google also makes a claim about visits after clicks. Its AI features page says users clicking from results pages with AI Overviews are more likely to spend more time on the destination site. [1] The cited page provides no underlying dataset or methodology, so that page alone does not allow an assessment of the claim. [1]

What controls does Google offer?

As of Oct. 29, 2026, Google documents controls for crawling, indexing, snippets, inclusion in Search generative AI features, and specified training and grounding uses. [1][3][5][9][10] Their scopes and effects differ. [3][5][9][10]

Control	Scope	Documented effect	Source
robots.txt rules for Googlebot	Specified paths or site	Manage crawling for Google Search, including Discover and other Search features; this is not a substitute for a readable noindex rule.	[9][10]
noindex	Page or resource	Prevents the content from appearing in Google Search results.	[3][9]
nosnippet	Page	Prevents text snippets and video previews, and prevents content from being used as direct input for AI Overviews and AI Mode.	[9]
max-snippet	Page	Limits text-snippet length and the amount usable as direct AI Overview and AI Mode input, subject to separately permitted uses.	[9]
data-nosnippet	Specified text within a page	Excludes marked text from snippets.	[9]
Search generative AI control	Search Console property and its subpages	Includes or excludes links and content from supported features, with inheritance and child-property overrides.	[5]
Google-Extended	Paths covered by robots.txt rules	Controls specified training and grounding uses; Google says it does not affect Search inclusion or ranking.	[10]

How does the Search generative AI control work?

Google places the control under Settings > Search generative AI in Search Console. [5] Its help page says the control reached worldwide availability by Aug. 31, 2026. [5] The documented feature list includes AI Overviews, AI Mode, and generative AI features in Discover. [5]

Excluding a property prevents its links and content from appearing in those features and makes its crawled content ineligible as input for their responses or previews. [5] Google says the excluded site receives no impressions or traffic from those features. [5] Content from other sites can still appear, including content that resembles the excluded site’s material. [5]

The control does not serve as a ranking or inclusion signal for other parts of Search, according to Google. [5] It also does not override separate participation choices for services such as Merchant Center and Google Ads. [5]

Scope requires care. A child property can follow its parent’s setting or use its own setting for that property and its subpages. [5] Owners should therefore check the effective setting rather than relying only on a top-level configuration.

Google says exclusion generally takes a few days after a change. [5] Its help page specifies exclusion within one to two days after the control goes live, while warning that caching and propagation can delay some content. [5]

What changed on Oct. 21, 2026?

Google updated its control-sharing documentation on Oct. 21, 2026, to include robots meta tags, data-nosnippet, X-Robots-Tag, and the Search generative AI control. [3][4]

The updates log describes these as existing options and says the purpose was to refresh the page’s coverage of available controls. [4] That supports a narrow conclusion: this documentation update was not an announcement of a new control. [4]

Although the overview page labels the Search generative AI control site-wide, the dedicated help page documents property inheritance and child-property overrides. [3][5] Use that more detailed scope when reviewing an implementation.

What exceptions and timing limits apply to snippet rules?

Google’s robots specification says nosnippet applies across Search result types, including AI Overviews and AI Mode, and prevents content from being used as direct input for those two features. [9]

The specification says max-snippet also limits how much content can be used as direct input, but separately granted permissions are exceptions. [9] Google gives in-page structured data and licensing agreements as examples of permitted uses that the limit does not interrupt. [9] Structured data can remain usable even when declared inside a data-nosnippet element. [9]

Google must be able to crawl a page to discover its robots meta tags or X-Robots-Tag headers. [9] Blocking that URL in robots.txt can therefore prevent Google from reading those rules. [9] When robots rules conflict, Google applies the more restrictive rule. [9]

Changes to preview controls require recrawling and processing. [1] Google says recrawling can take from several days to several months, depending on its systems’ assessment of how often the page needs refreshing. [1]

Why is Google-Extended a separate decision?

Google-Extended is a control token, not a separate HTTP request user agent. [10] Google’s crawler documentation, last updated Oct. 15, 2026, describes it as controlling whether crawled content can be used for future Gemini model training and specified grounding uses in Gemini Apps and Vertex AI. [10]

That page says Google-Extended does not affect a site’s inclusion in Google Search and is not a Search ranking signal. [10] However, the Search generative AI control help page also directs publishers to Google-Extended to limit training of models used to generate responses in Search generative AI features. [5]

The distinction is between training and display eligibility. Google explicitly says the Search generative AI control does not affect AI training. [5] The crawler page’s product description and the Search control page’s training instruction should therefore both be considered, rather than describing Google-Extended simply as an outside-Search control. [5][10]

What do the cited pages leave unclear?

Google’s documentation describes retrieval through Search ranking systems, but does not provide a complete selection procedure for the sources linked in a particular AI response. [1][2]

It also leaves several narrower questions unresolved:

• The cited guides do not specify a typical fan-out query count or how that count varies by market or language. [1][2]
• The dedicated report’s help page documents impressions, not a separate click metric for its supported AI features. [6]
• The AI features page does not provide the dataset or methodology behind Google’s statement about longer visits after clicks. [1]
• The Google-Extended crawler description names Gemini Apps and Vertex AI uses, while the Search control help page also connects the token to training models used in Search AI features without explaining that relationship in detail. [5][10]

These are limits of the cited pages, not claims that Google has never addressed the questions elsewhere.

What do third-party studies say about clicks?

Pew Research Center and Ahrefs report lower traditional-result click measures associated with AI summaries or AI Overviews in their respective samples. [15][16] Their methods are observational, not controlled experiments establishing that an AI Overview caused the difference. [15][16]

Pew Research Center examined participants’ browsing activity and reconstructed Google results after the browsing period to identify pages with AI summaries. [16] Its article reports less frequent clicking on traditional results when a summary appeared, while cautioning that summaries can change over time. [16]

Ahrefs, a commercial SEO software vendor, used aggregated desktop Search Console data and compared keywords with AI Overviews against informational keywords without them. [15] Its estimate compares observed click-through rates with a forecast based on changes in the comparison group. [15]

These studies examine clicking, while Google’s cited quality statement concerns time spent after a click. [1][15][16] Neither measure, by itself, establishes the economic value of a visit or the net effect on a particular site’s conversions.

What should site owners do now?

The following steps reflect Google’s documentation as of Oct. 29, 2026, rather than AVJ test results.

1. Check eligibility together with inclusion. Review indexing, snippet eligibility, Search policies, and the effective Search generative AI setting. [1][2][5]
2. Inspect inherited settings. Open Settings > Search generative AI for the relevant property and check whether a parent setting or child override applies. [5]
3. Review preview restrictions carefully. Check nosnippet, max-snippet, and data-nosnippet, including separately permitted structured-data uses and whether Google can crawl the affected pages. [9]
4. Keep reporting views distinct. Use the dedicated report for AI-feature impressions and the ordinary Performance report for overall Web clicks; do not treat the latter as an isolated AI-feature click report. [1][6][8]
5. Consider outcomes after visits. Google’s site-owner guidance recommends examining conversion indicators such as sales, sign-ups, and other forms of engagement. [11]
6. Evaluate Google-Extended separately. Consider both its crawler documentation and the Search control page’s explicit instruction about limiting training of models used in Search AI features. [5][10]
7. Check third-party claims against official guidance. Google warns against promises of ranking success or access to internal ranking and AI systems. [2]

What remains to be measured?

The cited click studies do not provide a direct comparison of AI Mode and AI Overviews across the US, UK, German, French, and Australian markets. [15][16] The documentation also does not establish whether a particular site’s AI-feature visibility will translate into more conversions. [1][6][11]

This guide contains no AVJ Citation Index results.

How this was made: AVJ used AI tools to help compare claims with the cited sources and to line-edit this guide, as described in our AI Use Policy. Nhi Phan reviews the published version.

Related on AVJ

• AI crawler directory: who crawls your site for AI, and how to control it
• Glossary: AI Overviews, AI Mode, query fan-out, grounding
• News: Google: robots.txt governs crawling, not AI use. What does?

Sources

[1] Google Search Central. AI features and your website. Last updated Dec. 10, 2025; accessed Oct. 29, 2026.

[2] Google Search Central. Optimizing your website for generative AI features on Google Search. Last updated July 10, 2026; accessed Oct. 29, 2026.

[3] Google Search Central. Control what you share with Google Search. Last updated Oct. 21, 2026; accessed Oct. 29, 2026.

[4] Google Search Central. Latest documentation updates. Historical entries and Oct. 21, 2026, entry cited; accessed Oct. 29, 2026.

[5] Google. Search generative AI control. Search Console Help. Includes Aug. 31, 2026, rollout note; accessed Oct. 29, 2026.

[6] Google. Generative AI performance report (Search). Search Console Help. Includes Aug. 31, 2026, rollout note; accessed Oct. 29, 2026.

[7] Hillel Maoz and Moshe Samet. Introducing Search Generative AI performance reports in Search Console. Google Search Central Blog. June 3, 2026; includes Aug. 31, 2026, rollout note; accessed Oct. 29, 2026.

[8] Google. What are impressions, position, and clicks?. Search Console Help. No publication or update date shown on the page; accessed Oct. 29, 2026.

[9] Google Search Central. Robots meta tag, data-nosnippet, and X-Robots-Tag specifications. Last updated March 24, 2026; accessed Oct. 29, 2026.

[10] Google. List of Google’s common crawlers. Google Crawling Infrastructure. Last updated Oct. 15, 2026; accessed Oct. 29, 2026.

[11] John Mueller. Top ways to ensure your content performs well in Google’s AI experiences on Search. Google Search Central Blog. May 21, 2025; accessed Oct. 29, 2026.

[12] Google Search Central. Spam policies for Google web search. Last updated Aug. 28, 2026; accessed Oct. 29, 2026.

[13] Google. Find information in faster & easier ways with AI Overviews in Google Search. Google Search Help. No publication or update date shown on the page; accessed Oct. 29, 2026.

[14] Google. Get AI-powered responses with AI Mode in Google Search. Google Search Help. No publication or update date shown on the page; accessed Oct. 29, 2026.

[15] Ryan Law. Update: AI Overviews Reduce Clicks by 58%. Ahrefs Blog. Vendor research. Feb. 4, 2026; accessed Oct. 29, 2026.

[16] Athena Chapekis and Anna Lieb. Google users are less likely to click on links when an AI summary appears in the results. Pew Research Center. July 22, 2025; accessed Oct. 29, 2026.