AI Mode SEO checker

What Google says AI Mode requires

Eligibility is one sentence in Google's documentation: a page must be indexed, and eligible to be shown with a snippet. Google adds that there are no additional technical requirements, and no special optimisations.

The full wording is worth having in front of you, because it is narrower than most advice implies: to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet, fulfilling the Search technical requirements.

The second half of that sentence is the part sites miss. Being indexed is not enough on its own — a page can rank perfectly well and still be excluded from AI Mode because a directive on it forbids a snippet. Snippet eligibility is a separate switch, and it is usually flipped by accident.

AI Mode and AI Overviews are not the same surface. Google says they may use different models and techniques, so the set of responses and links they show will vary. The eligibility rule is shared; which pages get chosen is not.

The four directives that silently remove you

Each of these is a one-line change someone made for a good reason years ago, and each one costs AI Mode eligibility today. Google states the snippet rules limit how much content may be used as a direct input for AI Overviews and AI Mode.

There is also a fifth control, and it is newer than most advice on this subject. Search Console now carries a search generative AI control with Include and Exclude settings for the whole property. Exclude prevents your content appearing in AI Overviews, AI Mode and generative features in Discover, while leaving ordinary Search untouched — Google states it is not used as a ranking or inclusion signal affecting other parts of Search.

1. View source on the page and search the head for a robots meta tag, then read its full content string — not just the word noindex.
2. Check the HTTP response headers too, because X-Robots-Tag carries the same directives and never appears in the HTML.
3. Search the body for data-nosnippet, which is usually applied to a wrapper by a template rather than by whoever wrote the page.
4. Open Search Console and confirm the search generative AI control is set to Include for the property.
5. Re-scan the URL afterwards and confirm the indexability checks pass.

What Google says makes no difference

This is the half of the category worth being blunt about, because a great deal of published advice contradicts the vendor's own documentation.

Google states that you do not need to create new machine-readable files, AI text files, markup or Markdown, and that llms.txt files neither harm nor help your site's visibility in its generative features. It says structured data is not required for generative AI search. It says there is no requirement to break your content into tiny pieces, and no need to write in a specific way just for generative AI search.

Google-Extended is the most consistently misreported of these. It governs Gemini apps and the Vertex AI API, and Google has confirmed it does not affect Google Search — so blocking it does not remove you from AI Mode, and allowing it does not admit you. If you want out of AI Mode specifically, the search generative AI control is the switch; Google-Extended is a different question about model training.

None of this makes structured data or a clean information architecture pointless. It makes them non-specific: they help other answer engines and they help Search generally, and they are not a Google AI Mode lever. A checker that tells you otherwise is selling you a signal Google has publicly denied.

Query fan-out changes which queries matter

AI Mode does not run your query once. Google describes a fan-out technique that issues multiple related searches across subtopics and data sources, then assembles an answer from a wider and more diverse set of links than a single query would return.

The practical consequence is that you surface for questions you never targeted. A page that answers one question well and its three obvious follow-ups badly will appear for the one and lose the other three to whoever covered them. Coverage of the neighbouring subtopic is the lever, not another variation of the head term.

It also means impressions arrive before clicks do, and often at positions far below the first page. Seeing a phrase in Search Console with impressions and no clicks is not a tracking fault — it is the fan-out finding you for a subtopic on which you rank badly.

• Answer the follow-up questions on the same page rather than scattering them across three.
• Give each subtopic its own heading, phrased the way the question is asked.
• Open each section with a sentence that answers its heading without the surrounding paragraphs.
• Keep one checkable specific per section — a figure, a threshold, a named directive.

What this checker reads, and what it cannot

Maveriko fetches the URL you give it and reads the page a crawler would receive. It reports whether the page is indexable, whether the robots rules admit the crawlers that matter, whether the content is in the HTML rather than painted by JavaScript, and whether the prose is shaped so a single passage can be lifted out and quoted.

It does not tell you whether AI Mode has ever named you, and no tool that reads your HTML can. That would require running live prompts repeatedly and watching answers that move as the model and the index change. Google's own Search Console generative AI performance report is the closest first-party measure of that, and it is the one worth watching.

The honest split is between the inputs you control and the outputs you do not. Eligibility and extractability are inputs, they are measurable from the page, and they are what this checker scores.

1. Scan the URL and clear anything under Indexability first — nothing else matters while the page is ineligible.
2. Read the snippet directives yourself, since a page can be perfectly indexable and still barred from being quoted.
3. Fix the section openings next: one sentence under each heading that answers that heading on its own.
4. Then widen coverage to the neighbouring questions, because fan-out retrieves subtopics rather than your head term.

Frequently asked questions

Does blocking Google-Extended remove my site from AI Mode?

No. Google-Extended governs Gemini apps and the Vertex AI API, and Google has confirmed it does not affect Google Search. Blocking it changes nothing about AI Mode or AI Overviews. The control that does affect them is the search generative AI setting in Search Console, which has Include and Exclude options for the property.

Do I need llms.txt or extra structured data for AI Mode?

Not for Google's AI features. Google states that llms.txt files neither harm nor help visibility there, and that structured data is not required for generative AI search. Both still earn their keep elsewhere — schema makes your facts unambiguous for other engines and for rich results — but neither is a Google AI Mode lever, and anyone selling one as such is contradicting Google's documentation.

How do I stop appearing in AI Mode without losing Search?

Use the search generative AI control in Search Console and set the property to Exclude. Google says it prevents your content being visible in AI Overviews, AI Mode and generative features in Discover, and that it is not used as a ranking or inclusion signal affecting other parts of Search. Using nosnippet instead also works, but it strips your ordinary search snippets too, which is a much higher price.

Is AI Mode SEO different from optimising for AI Overviews?

The eligibility rule is identical — indexed and snippet-eligible — but the surfaces are not. Google says the two may use different models and techniques, so the responses and links they show will vary. In practice you do the same work and should expect to appear in one without the other.

Why do I get AI Mode impressions but no clicks?

Usually because fan-out found you for a subtopic on which you rank far down. AI Mode issues several related searches and pulls from a wider set of links, so a page can collect impressions at position eighty without ever being seen. Check your average position before treating it as a tracking problem: impressions at that depth are a coverage signal, not a traffic one.

Read next