Diese Leitfäden zur Behebung sind in englischer Sprache verfasst.

KI-Direktiven

Diese Prüfung trägt 6 % des Scores.

How the page-level AI directives work: noindex, nosnippet, max-snippet, noarchive, nocache, data-nosnippet, and why Google-Extended does not do what most people think.

These are the only page-level controls Google and Bing document as actually honoring for AI answers. Everything else in this area is a declaration; these change behaviour. That cuts both ways: several of them remove you from AI answers, and sites set them by accident.

Homepage is marked noindex

A noindex page is invisible to every search-grounded assistant, because they all cite from a search index. Not ranked lower: absent.

If it is deliberate, nothing else in this check matters. If it is not, it is almost always a staging directive that shipped:

<!-- remove -->
<meta name="robots" content="noindex, nofollow">

<!-- or the header form, which is easier to miss -->
X-Robots-Tag: noindex

nosnippet, and max-snippet:0

Google documents that nosnippet prevents your content being used as a direct input for AI Overviews and AI Mode, as well as suppressing the search snippet. max-snippet:0 has the same effect by another route: a zero-character limit is a refusal.

Remove it if you want to be quoted. Keep it if you do not. This is one of the few settings in the whole audit where both answers are legitimate and the site owner is the only one who knows which.

A short max-snippet limit

A low character limit leaves an assistant too little to quote usefully, so it tends to paraphrase from the title instead. max-snippet:-1 means no limit and is the setting for a page you want cited accurately.

noarchive removes you from Copilot answers

Microsoft documents that noarchive means a page is not included in Copilot answers and not linked from them. It is a heavier instrument than most people setting it realise, and it is often inherited from an old SEO template.

nocache is the lighter option

With nocache, Copilot may use your URL, title and snippet but not the body. You stay linkable and stop being quotable. Remove it to allow full grounding.

data-nosnippet wraps the main content

Scoping data-nosnippet around a cookie banner, a byline or a price that changes hourly is housekeeping. Wrapping <main>, <article> or <body> removes the page from AI answers as surely as nosnippet does, without looking like it.

<!-- housekeeping -->
<div data-nosnippet>Cookies? Yes / No</div>

<!-- the same thing as nosnippet, by accident -->
<main data-nosnippet> … the whole page … </main>

noimageindex

Your images are excluded from indexing, so an assistant citing your text has nothing visual to show beside it. Remove it if you want images cited alongside the words.

Google-Extended does not do what you think

Disallowing Google-Extended in robots.txt governs Gemini training and grounding. It does not remove you from AI Overviews, which are built from the ordinary search index and controlled by the page-level directives above.

The pair people mean to set

To stay in search and out of AI answers: allow Googlebot, disallow Google-Extended, and set nosnippet. Setting only the robots.txt half is the common mistake, and it leaves the page quoted in the surface most people were trying to leave.

robots.txt Google-Extended: Gemini training and grounding.
nosnippet / max-snippet:0: AI Overviews and AI Mode.
noarchive: Copilot answers.

Homepage HTML unavailable

The check could not read the page, so it cannot report on directives it never saw. This is a symptom of something earlier: the site did not answer, answered with something that was not HTML, or required JavaScript to produce any. See the HTML Rendering and Agent Access guides.