How to get cited in Google AI Overviews and AI Mode
Google takes the links in AI Overviews and AI Mode from its ordinary search index, so there is nothing extra to submit, install or mark up. A page can be shown as a supporting link once it is indexed and eligible to appear in Search with a snippet, provided the site has not opted out through Search Console's Search generative AI control. Google-Extended plays no part in that decision. The snippet rules do: nosnippet, data-nosnippet and max-snippet limit what Google may show or use. Beyond eligibility, the links go to pages that hold up in ordinary search across the related searches Google fans each question out into, and since 31 August 2026 Search Console's Generative AI performance report has been open to every site.
The Rank Master works on Google's two AI features and the five other engines buyers ask first, for B2B SaaS teams, with published pricing and no annual lock-in. Book a 30-minute call to look at your category.
How Google builds the answer and picks its links
Google describes both features as built on its core ranking and quality systems. Its guide to generative AI features explains that the model works from pages Google's ranking systems retrieve from the Search index, a method it calls retrieval-augmented generation or grounding, and then shows links to the pages that support the response. Along the way the model may run query fan-out: a batch of related queries, issued at the same time, to fetch more results for the different parts of the question.
For a SaaS buying question, fan-out is the mechanism to plan around. A query like "best payroll software for a remote-first startup" can fan out into searches about payroll pricing, paying contractors abroad, compliance by country and named comparisons. Each of those searches has its own results, and the answer's links can come from any of them. A single long guide has to hold its own in each of those searches against pages written for that one question, while a pricing page, an international payments page and a comparison page are each written for theirs.
The two features do not behave the same way. Google says AI Overviews appear only when its systems judge them to add something to the classic results, so many queries never show one, while AI Mode is aimed at questions that call for exploration, reasoning or comparison. They may run on different models and techniques, which is why the same page can be linked in one and missing from the other (Google).
The eligibility checklist
Google's position is that AI features add no technical requirements of their own. What applies is the ordinary Search baseline, plus one newer setting.
| What to confirm | What Google says | Where to check |
|---|---|---|
| Indexed and snippet-eligible | Pages need to be in the index and allowed to show a snippet | URL Inspection in Search Console |
| Crawlable by Googlebot | Robots.txt rules for Googlebot are the control for Search, AI features included; your CDN or host must allow crawling too | robots.txt, server logs, URL Inspection |
| Site included in Search generative AI features | Required for display; the control reached every site on 31 August 2026, and "include" is the default | Search Console: Settings, then Search generative AI |
| No snippet limits on the content you want used | nosnippet and max-snippet restrict what AI Overviews and AI Mode can use; data-nosnippet removes the sections it wraps | Page source, or the HTML Googlebot received, in URL Inspection |
| Important content in text | Text first, with images and video supporting it, and structured data that matches the visible page | The rendered page |
Meeting every line still guarantees nothing: Google says plainly that it may not crawl, index or serve a page that qualifies.
The controls, and what each one changes
| Control | What it changes in AI Overviews and AI Mode | Worth knowing |
|---|---|---|
| robots.txt rules for Googlebot | Whether Google can crawl the page for Search, AI features included | Google calls "robots.txt directives for Googlebot" the control for how a site is crawled for Search |
| noindex | Removes the page from Search, so from both features | The bluntest option |
| nosnippet | No text snippet, and the content cannot be a direct input to either feature | Applies to every kind of Google result, not only the AI ones |
| max-snippet:[number] | Caps how much text can be shown and used as a direct input | Does not cut across uses you permitted separately, such as in-page structured data or a licence agreement |
| data-nosnippet | Keeps the wrapped span, div or section out of snippets | The rest of the page stays usable |
| Search generative AI control, set to exclude | Removes your links and content from both features, including as grounding | Also covers generative AI features in Discover; does not affect AI training or ranking elsewhere in Search |
| Google-Extended token in robots.txt | Not whether you appear: Google says it does not affect inclusion in Google Search | It covers Gemini training and grounding in Gemini Apps and Vertex AI |
The snippet rules come from Google's robots meta tag documentation, the control from its Search Console help and the Google-Extended row from its crawler documentation.
A SaaS site rarely wants any of these on its commercial pages, so the risk is accidental use. Look for a nosnippet left in a template, a data-nosnippet wrapper around a pricing table, or a Google-Extended block that someone added expecting it to keep the site out of AI Overviews. It will not do that, though it does affect Gemini. Changes also take time. Google says an exclusion set in the control generally applies within a few days: content is excluded within one to two days of the control going live, though caching can hold some of it longer. Page-level rules wait for a recrawl, which Google says can run from days to months.
What to publish so the fan-out searches find you
Start from the sub-questions behind each buying query rather than the query itself. For a B2B SaaS product the usual set is pricing and plan limits, integrations, security and compliance, migration from a named rival, use cases by team or industry, and head-to-head comparisons. Each deserves a page that answers it completely, in text a reader can verify.
Google's guidance favours what it calls non-commodity content: material that brings first-hand knowledge instead of restating what is already on the web. For a software company that means real pricing with its limits, screenshots of the actual product, implementation steps from your own onboarding and numbers from your own product data. A page that repeats what every vendor in the category says gives Google no reason to pick it over the others.
Two limits apply. Google warns that producing content for every variation of how people search, fan-out queries included, mainly to manipulate rankings or AI answers, breaches its scaled content abuse policy; write the pages buyers need, not one per phrasing. And Google says its AI features can show what is said about a product across blogs, videos and forums, but that planting mentions does little. The mentions that count are earned on the review sites and roundups your buyers already read, which our guide to getting a SaaS product recommended by AI engines covers across all seven engines.
What Google says you can skip
Google's position is that optimising for its generative AI features is still SEO, and its guide lists several popular tactics that Google Search does not need:
- llms.txt and other AI text files, special markup or Markdown versions of pages. Google Search ignores them, so they make no difference there either way.
- Chunking pages into small pieces. Google says its systems can find the relevant part of a page that covers several topics.
- Rewriting content for AI or chasing every long-tail phrasing, since the systems understand synonyms and what a searcher means.
- Schema.org markup added in the belief that AI features need it. None is required, though structured data still helps a page qualify for rich results.
Google also warns against tools that promise rankings or say they draw on Google's internal metrics, because, in its words, "No third-party tool has access to our internal ranking or AI systems". We agree that the foundations have not moved, and we set out what does change when buyers ask instead of search.
Measuring AI Overviews and AI Mode in Search Console
Google announced Search Console's generative AI performance reports on 3 June 2026 and, after testing them with a subset of sites, made them available to every site on 31 August 2026 (Search Central blog). The Search report counts impressions from AI Overviews and AI Mode; Discover has a report of its own. According to the help article, each appearance of a link to your site in one of these features counts as an impression. The documented ways to slice the data are page, country, date and device, with a filter that separates text searches from multimodal ones such as Google Lens.
Three limits are worth knowing before you report on it:
- Impressions only. The report shows how often you were linked, not whether you were named or recommended, and its documented dimensions do not include the search query. Google says it may add metrics over time.
- Property-level counting in the chart. In the chart, your site counts once per answer however many of its pages are linked, so table totals can differ from the chart.
- A missing report usually means too few impressions so far, or a site excluded through the control.
The same data also sits inside the standard Performance report under the Web search type, so AI feature visibility is part of your overall search totals. Pair the report with your analytics to see what those visitors go on to do, and with your own checks of the buying questions in AI Mode, where you can see which brands are named and which pages are linked. ChatGPT is a separate system with its own crawler and no Search Console data; our guide to getting cited by ChatGPT covers it.
Questions people ask about AI Overview citations
Do you need schema markup to appear in AI Overviews?
No. Google says AI Overviews and AI Mode need no special schema.org markup, and that its generative AI features do not require structured data. It is still worth using for rich results, as long as it matches what the page visibly says.
Does blocking Google-Extended remove a site from AI Overviews?
No. Google says the Google-Extended token has no bearing on whether a site appears in Google Search; it covers Gemini training and grounding in Gemini Apps and Vertex AI. To leave AI Overviews and AI Mode, use the Search generative AI control for the whole site, or snippet rules for single pages.
How do you stop Google using a page in AI Overviews and AI Mode?
For the whole site, set the Search generative AI control in Search Console to exclude. For a single page, nosnippet stops the content being used as a direct input, max-snippet caps how much can be used and data-nosnippet protects specific sections. A noindex rule removes the page from Search altogether.
Are AI Mode citations different from AI Overview citations?
Often, yes. Google says each feature can run on its own models and techniques, so one may link pages the other does not. AI Overviews also appear only when Google judges them useful, while AI Mode is built for nuanced, comparative questions, so track the two separately.
How quickly do changes show up?
A change to the Search generative AI control usually takes effect within a few days. Page-level rules take effect after Google recrawls the page, which can take from several days to several months; you can ask Google to recrawl a page you have changed.
Do AI Overviews replace SEO?
Not in Google's view. Google says its generative AI features rest on its core ranking and quality systems, and that optimising for them is still SEO. What changes is how well each page has to answer one part of a question on its own.
If you would rather have a team run this alongside ChatGPT, Perplexity, Gemini, Claude and Grok, The Rank Master's published plans start at $2,000 a month for 4 revenue pages and 2 earned citations, with custom programmes from $1,500 a month. Your pages, briefs, data and contacts stay yours, and after the 90-day initial term it is month-to-month. Book a 30-minute call.