Missing From Gemini? Check Google Index, Snippets, Extended

Gemini is not one place

When people say "Gemini does not mention us", they can mean any of three different things, and each one draws on Google's systems in a slightly different way. Separating them is the first job, because the fix for one is not always the fix for another.

The practical consequence is simple. If you are missing from all three, the problem is almost certainly in how Google Search sees your site. If you appear in AI Overviews but never in the Gemini app, look at Google-Extended and at whether the app is answering from training rather than a live search. If you appear in ordinary results but never in AI Overviews, look first at snippet controls and then at whether your page gives a quotable answer.

Narrowing down the cause

Work through these questions in order. Each one rules out a whole class of causes, and most sites find their answer in the first three.

1. Does the page appear in ordinary Google results for the question? Search the exact question and a few variations. If your page is nowhere in the first few pages, Gemini has little reason to pick it, and the work is ordinary search visibility rather than anything AI-specific.
2. Is the page indexed at all? Use URL Inspection in Search Console. "URL is not on Google" ends the investigation: nothing downstream can use a page Google has not indexed.
3. Does your result show a text snippet under the title? If Google shows only the title and address, or a very short fragment, a snippet rule is probably in force. That rule also limits AI Overviews and AI Mode.
4. Is the answer visible in the rendered HTML Google stored? In URL Inspection choose View crawled page, then the HTML tab, and search for a sentence from your answer. If it is missing, Google indexed a shell.
5. Does Google know who you are? Search your business name. A knowledge panel with the right address, website and description is a good sign. A panel for a different company with a similar name, or none at all, means entity work is needed.
6. Does the Gemini app answer from search? Ask it something recent that it could only know from the web, such as this month's opening hours on your site. If it cannot, it is answering from training for your topic, and the gap is about how often and how widely you are written about.

The AI Visibility Tracker helps here because it asks the same set of customer questions repeatedly and records whether your brand is cited, mentioned, recommended or missing. A pattern across many prompts tells you far more than a single test in your own browser.

Snippet rules: the setting that most often hides a page

Google's documentation on AI features is explicit that the normal snippet controls also govern AI Overviews and AI Mode. If a page tells Google not to show a snippet, Google will not quote it in an AI answer either. These rules are often added years ago for unrelated reasons, or switched on by a plugin setting, and then forgotten.

Look for them in three places, because the page source alone can mislead you:

<!-- 1. In the HTML head -->
<meta name="robots" content="index, follow, max-snippet:-1, max-image-preview:large">

<!-- 2. In the HTTP response headers (check with developer tools, Network tab) -->
X-Robots-Tag: nosnippet

<!-- 3. Wrapped round content in the body -->
<div data-nosnippet>...your main answer...</div>

The first line above is what you want on a page you would like quoted: max-snippet:-1 means no limit. A header rule wins even when the meta tag looks correct, and header rules are usually set in the server configuration, a CDN rule or a security plugin rather than in the page template.

What Google-Extended does, and what it does not

Google-Extended is a robots.txt product token, not a separate crawler. Googlebot does the fetching; the token tells Google whether content it has fetched may be used for Gemini. Google's crawler documentation describes it as controlling use for training future Gemini models and for grounding answers in the Gemini apps and in Google's grounding service for developers.

It does not control Google Search. Google states that Google-Extended has no effect on a site's inclusion or ranking in Search, and AI Overviews and AI Mode are part of Search. So:

• Blocking Google-Extended can reduce how often your content is used in the Gemini app, but it will not remove you from AI Overviews or AI Mode.
• Allowing Google-Extended will not by itself get you into AI Overviews. That depends on indexing, snippets and the quality of your answer.
• Blocking Googlebot to keep content away from AI removes the page from Google Search entirely, which is almost never the intent.

# robots.txt: allow Gemini to use the site as well as Search
User-agent: Googlebot
Allow: /

User-agent: Google-Extended
Allow: /

# or, to opt out of Gemini training and grounding only:
# User-agent: Google-Extended
# Disallow: /

Because Google-Extended is a token rather than a crawler, you will not see it in your access logs. The only way to confirm what you have told Google is to read the live robots.txt file at the root of the domain, and at the root of every subdomain that serves content.

The fix, step by step

Where the blocking settings hide, by platform

WordPress

SEO plugins such as Yoast SEO, Rank Math and All in One SEO write the robots meta tag. Check each affected page's advanced settings for "no snippet" or a snippet length limit, and check the plugin's global defaults for the post type. Rank Math, for example, exposes advanced robots settings including max snippet length. Some security and performance plugins add X-Robots-Tag headers, which will not appear in the SEO plugin at all. The Settings, Reading option "Discourage search engines from indexing this site" adds noindex to every page and is still left switched on after launches more often than it should be.

Shopify

Shopify generates robots.txt from a template, robots.txt.liquid, which you can edit in the theme code. Apps that promise to "block AI" may add rules there or in the theme's theme.liquid head. Search both for Google-Extended, nosnippet and max-snippet. Product descriptions loaded by a third-party app after the page loads may also be missing from the rendered HTML, which Step 5 will reveal.

Cloudflare and other CDNs

Cloudflare offers a managed robots.txt option that can add Disallow rules for AI training crawlers to the start of your file, and bot management settings that can challenge or block automated visitors. Check what your live robots.txt contains, not the copy on your server, because the CDN may be rewriting it. Transform rules and response header rules can also add X-Robots-Tag to every response.

JavaScript frameworks

React, Vue and Angular sites that render in the browser can index as near-empty pages when rendering fails or times out. Server-side rendering or static generation, available in Next.js, Nuxt and Angular's own tooling, puts the answer in the first HTML response. Frameworks also let developers set robots meta tags per route, so check the route config as well as the base layout.

Cases where the usual fixes are not enough

• The query does not trigger an AI Overview. Google only shows one when its systems judge it helpful. For many short, navigational or commercial searches there is none to appear in. Check that the question you are testing actually produces an AI Overview for anyone before treating your absence as a fault.
• You appear in AI Overviews but not in the Gemini app. The app may answer your question from training alone, which reflects what was written about you before its cut-off date. Being mentioned more widely on other sites is the long-term fix; see the guide to weak brand mentions.
• Gemini names you but links to a directory. The directory page answered the question more directly than yours. Compare the two and bring the missing facts, such as prices, areas covered or opening hours, onto your own page.
• Gemini gives wrong facts about you. That is an accuracy problem rather than visibility, covered in fixing wrong attribution.
• A new site. Google may index a new site within days but take much longer to trust it enough to use in answers. Nothing in robots.txt speeds that up; useful pages and genuine mentions do.
• Workspace and enterprise Gemini. Answers inside company accounts may be configured to use internal documents only, so a missing website there may be an administrator setting, not a problem with your site.

Keeping Google AI features able to use your site

• Record your robots rules. Keep a short note of every robots.txt line, robots meta default and header rule, who added it and why. Unexplained rules are how most of these problems begin.
• Check after every plugin, theme or CDN change. Load one key page, look at the robots meta tag and the response headers, and read the live robots.txt. It takes two minutes.
• Watch the Page indexing report. A sudden rise in "Excluded by noindex tag" in Search Console is an early warning that a template change has spread a rule across the site.
• Keep answers in the HTML. Make server rendering a requirement for any page you want quoted, and include a rendering check in release testing.
• Track the same prompts monthly. Run your prompt set on a schedule so a drop shows up within weeks rather than when a customer mentions it.

For the other assistants, see optimising for Gemini, Claude and Copilot, and for crawler rules in general, fixing AI crawler access.