Google AI Mode and AI Overviews Optimization: The Complete Guide for Businesses

Search for AI Mode optimization advice and you’ll find plenty of similar lists. Google has published guidance that answers many of these questions directly. I’d start there, rather than rely on another blog’s version.

Let’s work out which Google document to use, which takes priority, and what each one says you can do.

Which Google page should you actually be reading?

Google publishes a small set of pages that bear directly on AI Mode and AI Overviews, plus a set of Search Console Help documents underneath them. They aren’t equivalent, they don’t perfectly agree with each other, and most of the commentary online still points at the oldest one.

AI features and your website is the page most often cited. It’s stamped Last updated 2025-12-10 UTC. That date matters: it predates the single largest change to AI-feature measurement by roughly eight months. It remains Google’s canonical statement on eligibility, and this guide cites it throughout - but treat anything you read elsewhere derived from it as describing the platform as of December 2025.

Google’s Guide to Optimizing for Generative AI Features on Google Search is the document that supersedes it in substance. Google added this guide on 2026-05-15 per its own documentation changelog, and its footer reads Last updated 2026-07-10 UTC. It contains a mythbusting section, a definition of query fan-out with a worked example, and the eligibility language that the older page doesn’t have. If you read one thing, read this one.

Two more sit at the edges. Preferred sources was last updated 2026-09-18 UTC and is genuinely current; it explains a mechanism: a “preferred” badge in AI Mode and AI Overviews. Google’s common crawlers, last updated 2026-10-15 UTC, is the authority on Google-Extended.

Here is the hierarchy I would actually work from, in descending order of how much weight each claim deserves:

1. The AI optimization guide (2026-07-10) - the current statement of intent and of what to ignore.
2. Search Console Help - the operational truth about what is measured and what is switchable. Updated 2026-08-31 for the two most important pages.
3. blog.google product announcements - what is shipping or being tested right now. See the 2026-10-15 post for the recent announcements.
4. AI features and your website (2025-12-10) - still authoritative on eligibility basics, stale on everything downstream.
5. Third-party SEO blogs, including most of the ones ranking for your target keywords. Last, and mostly not at all.

Google’s own guidance on third-party SEO tools and advice (2026-06-05) makes the fourth of those concrete. It names “AEO” and “GEO” tools by category and states plainly that Google doesn’t evaluate third-party services, that using such a tool “doesn’t guarantee ranking success,” and that third-party tools “don’t have access to our internal ranking data.” If a tool claims to show you your exact AI citation share, it’s modelling something, not measuring it.

What AI Mode and AI Overviews actually do

Google describes its generative features as retrieval-augmented generation sitting on top of conventional Search ranking. In the AI optimization guide’s phrasing, RAG relies “on our core Search ranking systems to retrieve relevant, up-to-date web pages from our Search index,” after which the systems “review the specific information from those retrieved pages to generate a more reliable and helpful response.”

Retrieval uses the same Search index. The systems that choose candidate pages are the ranking systems you’ve been optimising for already. There’s no separate AI index to join. That’s why much of this work still looks like ordinary SEO.

The second mechanism is query fan-out. Google defines it in three places, and the definitions differ.

The AI features page says both features “may use a ‘query fan-out’ technique - issuing multiple related searches across subtopics and data sources - to develop a response.” The Search Console methodology documentation puts it more operationally: “AI Mode groups the user’s question into subtopics and searches for each one simultaneously.” The AI optimization guide is the most concrete, and supplies its own worked example - if the user asks “how to fix a lawn that’s full of weeds,” fan-out queries might include “best herbicides for lawns,” “remove weeds without chemicals,” and “how to prevent weeds in lawn.”

Hold on to that example. I will come back to it.

One thing Google is careful about, and which the commentary usually flattens: AI Mode and AI Overviews aren’t the same system with different sizes. Google’s wording is that they “may use different models and techniques, so the set of responses and links they show will vary,” and that AI Overviews appear only when Google’s systems determine they’re “additive to classic Search,” and “as such, often don’t trigger.” So neither volume nor visibility can be modelled from classic query counts, and a page appearing in one tells you almost nothing about the other.

Scale, for context: at I/O 2026 Google said AI Mode had passed one billion monthly users, with queries “more than doubling every quarter since launch,” and moved Gemini 3.5 Flash in as the default model for AI Mode globally. That’s Google’s own figure, published without methodology. Treat it as a directional signal about investment, not a forecast you can plan budgets against.

The eligibility rule, and the one gate you can actually switch off

Here the two Google pages genuinely differ, and the difference is worth understanding rather than papering over.

The 2025-12-10 page states the rule and stops: to be eligible to be shown as a supporting link, “a page must be indexed and eligible to be shown in Google Search with a snippet, fulfilling the Search technical requirements. There are no additional technical requirements.”

The 2026-07-10 guide restates the same core rule and then adds a sentence the older page doesn’t contain:

In addition to the technical requirements for Search, a site must be included in Search generative AI features in Search Console to be eligible for display in generative AI features on Google Search.

That sounds like a new barrier. It isn’t - but it’s a real gate, and it’s the one control on this list you can flip. Google documents a Search generative AI control under Settings → Search generative AI in Search Console, rolled out worldwide on 2026-08-31. It offers three states:

• Include my site’s links and content in Search generative AI features - the default for all properties.
• Exclude my site’s links and content - content “is prevented from being visible to users in Search generative AI features, including being linked to within these features and helping with grounding AI responses in these features. You won’t receive any traffic or impressions from these features.”
• Inherit control from parent - the default where a parent property exists, cascading down unless overridden.

Three details in that page matter operationally. First, the control “only affects whether your content can appear in certain Search generative AI features; this control isn’t used as a ranking or inclusion signal affecting other parts of Search” - switching it off won’t cost you your organic rankings, only your AI-feature presence. Second, it “doesn’t affect AI training”; Google-Extended is the separate lever for that. Third, it doesn’t override your Merchant Center or Google Ads participation settings, which operate independently.

Your page needs to be indexed, eligible for snippets, and included through the default-on control. There’s no extra technical build. For most sites, I’d start with a thirty-second check in Search Console. A larger project is needed if you’ve excluded yourself or manage properties that inherit an exclusion from a parent.

Changes take a few days. Google’s guidance: content is excluded “within 1-2 days after the control goes live, but some content may take longer due to caching and propagation across Google systems.”

The query fan-out surface map

Google gives us three usable facts. Fan-out decomposes one prompt into concurrent related queries. AI Mode searches for each sub-topic simultaneously. AI Overviews occupy a single position, with every link inside assigned that same position. Google does not publish how a response’s retrieved pages are matched to particular parts of the response.

Here’s how I’d think about the page design. If retrieval happens by sub-topic, each section has a job to do. A page with fifteen loosely organised sections may be partly relevant to fifteen different needs. Clear sections give the engine a better-defined answer to the needs you do cover. That suggests a different page design from trying to answer the whole prompt at once.

So: build a map of the fan-out slots your topic generates, then check whether your page has a self-contained section that answers each slot.

Step one - enumerate the slots. Take a real prompt and decompose it the way Google says it decomposes. Use the Queries tab of the main Performance report, not the generative AI report - the latter has no query dimension at all, which is the measurement constraint discussed later. Google gives us the lawn example, so let me run it out. Prompt: how to fix a lawn that’s full of weeds. Google’s own fan-out set:

Slot	Sub-query	Slot type	What a winning section must do
1	best herbicides for lawns	Comparison / ranking	Name options and give a defensible basis for choosing between them
2	remove weeds without chemicals	Constraint	Serve the method-avoiding reader as a first-class case, not a footnote
3	how to prevent weeds in lawn	Prevention / verification	Show the end state, not just the fix

Three slots, three distinct reader intents. Note that slot 2 is a constraint slot - the reader has already excluded the obvious answer, and a page that only covers the obvious answer is invisible to it no matter how authoritative it is.

Step two - audit the section map. Now look at your page. If it has an H2 that a reader could scroll to and be satisfied by, you have a slot candidate. If your structure is a nine-screen narrative with bold lead-ins instead of headings, you have one undifferentiated blob competing against every slot at once.

Against the lawn example, a single “Lawn Care: A Complete Guide” page with no headings covers none of the three slots cleanly. A page carrying the same subject matter but broken into a herbicide-comparison section, a chemical-free removal section treated as a full method rather than an aside, and a prevention section lands directly on slots 1, 2 and 3. Identical subject. Different retrieval surface.

Step three - score coverage against density. Coverage is how many slot types the page addresses at all. Density is how many sections compete for the same slot. Five sections all arguing “why our product is the right herbicide” is dense and narrow: strong on one slot, absent from the rest. Spreading sections thinly across slots you can’t actually satisfy is worse than covering fewer slots well. The AI optimization guide’s own framing - write “non-commodity content,” for example “Why We Waived the Inspection & Saved Money: A Look Inside the Sewer Line” rather than commodity content like “7 Tips for First-Time Homebuyers” - is the same principle applied to substance. Google is asking you to have a position on each slot, not to have a heading.

Step four - map to surface. Fan-out sets aren’t one undifferentiated list, because Google has told us responses have distinct link surfaces. The 2026-10-15 announcement describes four: inline links embedded in the response text, which Google says it has increased and moved “higher up in responses more frequently”; a dedicated section at the top of the response, still being tested, for when “there are only one or two official places on the web to complete a specific task or get the exact info you need”; prominent carousels for timely articles, rolling out to more locations and languages; and the links at the end of a response, which Google is making more prominent with image thumbnails.

Which part of the response could your page fit into? A live status page, current pricing table, or official spec sheet could fit the top-of-response section being tested. A timely article could fit the carousel. A detailed page without a recency advantage is competing for inline links, where more sources compete. Choose the place you’re aiming for, then decide how much freshness and specificity matter.

The rail on this framework. It describes how to structure pages you already need. It isn’t a licence to manufacture one page per fan-out sub-query, which is the single most common scam in this topic and which Google explicitly forbids. The AI optimization guide warns that while it may be tempting “to create separate content for every possible variation of how people might search (for example, by focusing on other queries that people have asked, or fan-out queries), doing so primarily to manipulate rankings or generative AI responses in Google Search violates Google’s scaled content abuse spam policy,” and adds that it “is also an ineffective long-term strategy.” The spam policies documentation defines scaled content abuse as generating “many pages … for the primary purpose of manipulating search rankings and not helping users,” explicitly “no matter how it’s created.”

Use the surface map to improve the pages you have. If your ten-page pillar is missing an answer, the fix is usually a section on an existing page. Create a distinct page when there’s a distinct reader need. You rarely need eleven pages to solve that problem.

What Google says you can safely ignore

The AI optimization guide has a section titled “Mythbusting generative AI search: what you don’t need to do,” and it’s the most directly useful page Google has written on this. Here is the whole of it, with the framing preserved.

llms.txt and other “special” markup. Google: “You don’t need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search (including its generative AI capabilities), as Google Search itself doesn’t use them.” Google adds an important nuance that gets lost in the retellings: “Google may discover, crawl, and index many kinds of files in addition to HTML on a website: this doesn’t mean that the file is treated in a special way.” So an llms.txt on your site isn’t a penalty and not a boost. It’s a file Google may fetch and then ignore.

The canonical, dated confirmation is a changelog entry: 2026-06-15, “Clarifying guidance on llms.txt files”, which records that Google “added a note to the AI optimization guide clarifying Google Search’s usage of llms.txt files,” because the files “aren’t needed for Google Search (and won’t negatively or positively impact your visibility or rankings).” If Google reverses this, it will show up in the changelog first - check there rather than in a blog post.

“Chunking” content. “There’s no requirement to break your content into tiny pieces for AI to better understand it. Google systems are able to understand the nuance of multiple topics on a page and show the relevant piece to users.” Google then says something worth hearing: “There’s no ideal page length, and in the end, make pages for your audience, not just for generative AI search.”

Rewriting content specifically for AI systems. “AI systems can understand synonyms and general meanings of what someone is seeking, in order to connect them with content that might not use the same precise words. This means you don’t have to worry that you don’t have enough ‘long-tail’ keywords.” The corollary, for anyone who has been trained to fear keyword variance: you aren’t failing some coverage check.

Seeking inauthentic mentions. Google notes its features can surface what is being said about products across blogs, video and forums, but says seeking inauthentic mentions “isn’t as helpful as it might seem,” because core ranking focuses on quality while “other systems block spam; our generative AI features depend on both.” Brand-mention campaigns aren’t neutral - they’re an attempt to manufacture the signal, and Google is describing that as a technique rather than a tactic.

Overfocusing on structured data. “Structured data isn’t required for generative AI search, and there’s no special schema.org markup you need to add.” Google doesn’t say stop using it - it says continue, “as it helps with being eligible for rich results on Google Search.” That’s the correct framing: structured data is a rich-result and commerce mechanism, not an AI-citation mechanism.

One correction to widely repeated advice, while we’re here. If your playbook tells you to add FAQPage schema to win AI answers: Google’s FAQ rich result documentation was removed on 2026-06-15, with the changelog stating “The FAQ rich result feature is no longer shown in Google Search results.” FAQPage markup isn’t harmful and remains valid, but it buys you nothing in Search and nothing in AI features.

What Google actually asks you to do

Strip out the debunking and the AI optimization guide’s positive guidance is short enough to state plainly.

Non-commodity content with a point of view. Google distinguishes first-hand experience - “a first-hand review provides a unique perspective based on personal experience, whereas a summary of existing content simply restates information already available elsewhere” - from commodity content that “could originate from anyone.” Then the line that should stop a lot of editorial plans cold: “Don’t just recycle what others on the internet have already said, or could easily be produced by a generative AI model.”

Organised for readers. Google’s phrasing is human-first - “web pages are organized by paragraphs and sections, along with headings that provide a clear structure to navigate content.” That isn’t a ranking signal description; it’s Google telling you what it believes good pages look like. It happens to be exactly what the surface map in the previous section operationalises.

Important content available in textual form. This one is repeated on both Google pages and is the most concrete technical instruction either gives. The AI features page lists “Making sure that important content is available in textual form” among its SEO best practices. If your pricing, specifications, availability or policy lives only in an image, a PDF, or a canvas, it isn’t available to retrieval in the form the system needs.

Structured data matching visible text. The AI features page lists “Making sure your structured data matches the visible text on the page.” Read that as a consistency requirement, not an enrichment strategy: what your markup asserts should be what a reader can verify on the page.

Crawlable, internally linked, good experience. Allowed in robots.txt and by CDN or hosting infrastructure; findable through internal links; solid page experience. Google’s stated mechanism is blunt - its generative models “use publicly accessible, crawlable content.” A page behind a login, a blocked CDN path, or an internal-link orphan isn’t a candidate, and no amount of content quality substitutes.

Merchant Center and Business Profile current. Both Google pages call this out, and for businesses specifically it’s probably the most useful item on the list. The AI optimization guide is direct that “Where appropriate, generative AI responses can include product listings, product information, and information about local businesses,” and that Merchant Center and Business Profiles “can help your products and services to be visible in both AI responses and other Google Search results.” Google’s business details documentation covers the mechanics - claim your Business Profile, verify ownership in Search Console, keep your knowledge panel accurate. Stale price and availability data is a liability in a response that may quote it.

Measurement reality: impressions without clicks

This is where most published advice on this topic is now out of date, and where the biggest practical change has happened.

The December 2025 position, which you’ll still read everywhere, is that AI-feature traffic “is included in the overall search traffic in Search Console. In particular, they’re reported on in the Performance report, within the ‘Web’ search type.” That remains true as far as it goes. It stopped being the whole truth on 2026-08-31.

Google announced Search Generative AI performance reports on 2026-06-03 to a limited set of sites, and both the Search and Discover reports now carry the note: “As of August 31, 2026, we’ve rolled out these insights to all websites worldwide.” The Search generative AI performance report covers AI Overviews and AI Mode. There’s a parallel Discover report.

So AI visibility is now directly measurable. Here is the ceiling, and it’s lower than the announcement implies.

The report shows impressions. That’s all it shows. The default view “shows impression data.” The metrics available are impressions. The dimensions are Pages, Countries, Dates and Devices. There’s no clicks metric, no CTR, no average position, and - importantly - no Queries dimension. You can see that a page is surfacing. You can’t see what it’s surfacing for.

That last point is the operational crux. On 2026-09-24 Google added web multimodal Search performance reporting, which brought a new “Web: multimodal” search type covering Lens, Circle to Search, image uploads and Chrome’s “Search this image.” Compare that with the main Performance report’s search-type filter, which offers Web text-based, Web multimodal, Image, Video and News - and no AI Mode or AI Overviews option anywhere. The advanced filtering documentation says you can filter and compare by Search appearance, but there’s no filter to isolate AI features from classic results.

How to think about this. Treat the generative AI report as a top-of-funnel visibility instrument and the main Performance report as your conversion instrument, and stop looking for a report that does both. The generative AI report tells you which pages are being surfaced. The main report tells you whether those pages convert. Google states the generative AI report “includes data from the Web search type in the Performance report,” so the two views describe overlapping impressions from different angles.

The position fact that should change how you benchmark. From the Search Console methodology documentation: “An AI Overview occupies a single position in search results, and all links in the AI Overview are assigned that same position.” If your page is cited in an AI Overview alongside seven others, your average position is identical to theirs. There’s no position-1 AI Overview. Beating a competitor into an AI Overview can’t be measured as a ranking improvement, and any tool that reports one is reconstructing something the underlying data doesn’t contain.

AI Mode behaves differently - “Position in AI Mode follows the same methodology as a Google Search results page” - so if you need relative ranking signal, that’s where it lives. One more reporting quirk: “If a user asks a follow-up question within AI Mode, they are essentially performing a new query,” so follow-up-driven impressions attribute to the follow-up, not the original prompt.

A note on the older “higher quality clicks” claim. The 2025 page says “when people click from search results pages with AI Overviews, these clicks are higher quality (meaning, users are more likely to spend more time on the site).” Google publishes no methodology for this and no effect size, so treat it as an unquantified first-party assertion. It sits in direct tension with the best independent measurement available: Pew Research Center found that users who encountered an AI summary clicked a traditional result in 8% of visits, against 15% for those who didn’t, and clicked a link inside the summary itself in just 1% of visits.

The one durable thing Pew measured that survives the staleness problem is its relationship with query shape: 53% of searches of ten words or more produced an AI summary, against 8% of one- or two-word searches. Longer, question-shaped prompts are where these features live. That’s consistent with what Google is building - at I/O it described the new Search box as designed “to anticipate your intent,” helping users “formulate their question with AI-powered suggestions that go beyond autocomplete.” The direction of travel is toward more prompts triggering fan-out, not fewer.

Controls, and what Google-Extended does not do

Two different levers get conflated constantly. Separating them is the whole job.

In-Search preview controls. The AI features page is clear that “AI is built into Search and integral to how Search functions, which is why robots.txt directives for Googlebot is the control for site owners.” Four controls apply, and the robots meta tag specification spells out exactly how each behaves with respect to AI features:

• nosnippet - “This applies to all forms of search results (at Google: web search, Google Images, Discover, AI Overviews, AI Mode) and will also prevent the content from being used as a direct input for AI Overviews and AI Mode.”
• max-snippet:[number] - likewise “will also limit how much of the content may be used as a direct input for AI Overviews and AI Mode.” Note the carve-out: the limit “does not apply in cases where a publisher has separately granted permission for use of content. For instance, if the publisher supplies content in the form of in-page structured data or has a license agreement with Google.” Special values are 0 (equivalent to nosnippet) and -1 (let Google choose).
• data-nosnippet - a boolean HTML attribute on span, div and section elements to exclude specific parts of a page. Google’s advice: don’t add or remove it on existing nodes through JavaScript, to avoid extraction order depending on rendering.
• noindex - removes the page from Google Search entirely, AI features included.

The operational trap: these only work if the page is crawlable. Google’s specification states that if a URL is disallowed in robots.txt, “any information about indexing or serving rules will not be found and will therefore be ignored.” Blocking a page and expecting noindex to be honoured doesn’t work.

If a control doesn’t appear to take effect, Google gives a troubleshooting path: confirm the control is correct and visible to Googlebot using the URL Inspection tool to inspect the HTML Googlebot received, then wait - “crawling can take anywhere from several days to several months” - and request a recrawl.

Google-Extended, and the thing it doesn’t do. This is the most consistently misreported item in the entire topic. Google-Extended is a standalone robots.txt product token. Per the crawler documentation, last updated 2026-10-15 UTC, it’s “a standalone product token that web publishers can use to manage whether content Google crawls from their sites may be used for training future generations of Gemini models that power Gemini Apps and Vertex AI API for Gemini and for grounding (providing content from the Google Search index to the model at prompt time to improve factuality and relevancy).”

It has no effect on Google Search. Verbatim: “Google-Extended does not impact a site’s inclusion in Google Search nor is it used as a ranking signal in Google Search.” The control page says the same thing from the other direction - the Search generative AI control “isn’t used as a ranking or inclusion signal affecting other parts of Search.”

If you want visibility in AI Mode and AI Overviews, blocking Google-Extended doesn’t change what you can observe in Search. If you want to control Gemini training or grounding, Google-Extended is the relevant control. Blocking it won’t make you disappear from AI Overviews. The systems are separate, even though grounding uses content from the Search index.

One honest caveat on enforcement. Google-Extended has no separate HTTP user agent - “Crawling is done with existing Google user agent strings” - so you can’t identify its traffic in your logs or block it at the network layer. Google describes the token as being used “in a control capacity,” which is softer language than the word “block.”

For search visibility specifically, the order of preference runs: nosnippet or max-snippet for excerpt control, noindex to leave Search, and the Search generative AI control to leave AI features while staying in classic Search.

The business-specific layer

For a business site, three things sit outside ordinary on-page work.

Merchant Center and Business Profile. Called out on both Google pages, as above. In AI responses that can quote product and local information, stale data is worse than absent data.

Preferred Sources. Google rolled this into AI Overviews and AI Mode on 2026-05-27 per the changelog, and the documentation (2026-09-18) is current. When a user selects your site as a preferred source, “your content can be highlighted with a ‘preferred’ badge” in both AI Overviews and AI Mode. Two constraints catch people out: eligibility in AI features requires the site to be included in Search generative AI features in Search Console - the same control as above - and only domain-level and subdomain-level sites qualify. https://www.example.com/blog doesn’t. Implementation is a two-line script plus a container element, with a deeplink fallback if you can’t run JavaScript. Google reported in October 2026 that usage of the feature “has grown by more than 7x since the start of the year” - a relative figure with no stated baseline, so don’t model a number from it.

Timeliness, if you publish news or commentary. Google’s 2026-10-15 post describes prominent carousels for timely articles rolling out to additional locations and languages, expanding their reach “across a wider range of queries.” If freshness is genuinely part of your business model, this is a surface worth understanding.

Where to spend the effort: a decision tree

Six questions, in order. Stop when you get a “no.”

1. Is the property’s Search generative AI control set to include? Check Settings → Search generative AI. If it’s excluded, or inheriting an exclusion, nothing below matters. Thirty seconds.

2. Are priority pages indexed and snippet-eligible? If they aren’t eligible for a snippet in classic Search, they can’t be a supporting link in AI features - this is the entire eligibility rule. Fix indexing before anything else.

3. Is anything important trapped in non-textual form? Prices, specs, availability, policy. Move to text.

4. Do you have a page per genuine reader need, or a page per keyword variant? If the second, you have a scaled content exposure, not an AI problem. Google’s spam policy names generative content generated “without adding value for users” explicitly.

5. Pick your three highest-value prompts. Run the surface map on each. Not every page - three, because the map is only useful if you’ll act on it. For each, decompose the prompt into likely sub-queries, then check your page against the slots. Missing slots usually mean a missing section, and frequently mean a heading problem rather than a content-volume problem.

6. Are Merchant Center and Business Profile accurate? For most business sites this beats any content change available to you in the same week.

If you complete all six, the honest remaining answer is that there’s no seventh step. That isn’t a cop-out - it’s Google’s stated position, stated twice, in its own words: “you don’t need to accomplish everything in this guide in order to succeed on Google Search.”

Sources and References

• Google’s Guide to Optimizing for Generative AI Features on Google Search - Google Search Central, last updated 2026-07-10 UTC
• AI Features and Your Website - Google Search Central, last updated 2025-12-10 UTC
• Latest Google Search Documentation Updates - Google Search Central changelog, last updated 2026-10-16 UTC
• Search generative AI control - Search Console Help, rollout dated 2026-08-31
• Generative AI performance report (Search) - Search Console Help, rollout dated 2026-08-31
• Generative AI performance report (Discover) - Search Console Help, rollout dated 2026-08-31
• Introducing Search Generative AI performance reports in Search Console - Google Search Central Blog, 2026-06-03
• Announcing web multimodal Search performance reporting in Search Console - Google Search Central Blog, 2026-09-24
• What are impressions, position, and clicks? - Search Console Help
• Performance report (Search results): Overview and basic setup - Search Console Help
• Performance report (Search results): Advanced filtering and comparison - Search Console Help
• Robots meta tag, data-nosnippet, and X-Robots-Tag specifications - Google Search Central
• How to Write Meta Descriptions - Google Search Central, last updated 2026-04-20 UTC
• Guide to Preferred Sources in Google Search for Web Publishers - Google Search Central, last updated 2026-09-18 UTC
• Google’s common crawlers (Google-Extended) - Google Search Central, last updated 2026-10-15 UTC
• Google Search’s Guidance on Third-Party SEO Tools & Advice - Google Search Central, last updated 2026-06-05 UTC
• Search Essentials spam policies - Google Search Central, last updated 2026-08-28 UTC
• Add and manage your business details on Google Search - Google Search Central, last updated 2025-12-10 UTC
• More ways to connect with the web from Search - Google, 2026-10-15
• A new era for AI Search (I/O 2026) - Google, 2026-05-19
• Google AI Mode - a new way to search - Google
• Do people click on links in Google AI summaries? - Pew Research Center, 2025-07-22 (data collected March 2025)