Four Google controls that people keep mistaking for each other

Four different controls sit near each other in the same mental space, and they are routinely used as though they were interchangeable. They are not. Each one is documented by Google, and each does something specific.

Google-Extended

A robots.txt token with no user agent of its own. Google's crawler documentation describes it as a control publishers use to manage whether crawled content may be used for training future generations of the Gemini models, and for grounding in Gemini Apps and Grounding with Google Search on Vertex AI.

The documentation then says the part people miss: Google-Extended does not affect a site's inclusion in Google Search, and is not used as a ranking signal.

So disallowing it is a decision about Gemini as a product. It is not a way out of AI Overviews.

The Search generative AI control

This is the one that governs AI Overviews, AI Mode and generative AI features in Discover. It lives in Search Console under Settings, and it has three states: include, exclude, or inherit from the parent property.

Include is the default for every property. Exclude prevents your links and content appearing in those features, and prevents your content being used to ground responses in them.

Google is explicit about the cost. Its own wording is that you will not receive any traffic or impressions from these features. This is not a hygiene setting, it is a decision to give up a surface.

It is also explicit about the limits. The control is not used as a ranking or inclusion signal elsewhere in Search. It does not override Merchant Center or Google Ads participation. It does not affect training, which is what Google-Extended is for. Changes take a day or two to take effect, and sometimes longer through caching.

Inheritance catches people out. A property inherits from its closest parent that has stopped inheriting, defaulting all the way up to the top-level domain property. Set it on a domain property and every child follows unless someone has overridden it.

Nosnippet and max-snippet

These are page-level and element-level directives that stop Google showing a text preview of your content. Because an AI Overview is built from passages, a page carrying nosnippet cannot contribute one.

The reason this is rarely the right lever is that the same directive removes your ordinary search snippet and your featured snippets. You would be paying for AI exclusion with the description that sells your normal result. For most commercial sites that trade is bad.

Noindex

The blunt one. Noindex removes the page from Google Search altogether, which removes it from the generative features as a side effect, because eligibility for those features requires the page to be indexed and eligible to show with a snippet in the first place.

Worth stating plainly so nobody reaches for it as an AI control. It is a Search control that happens to have an AI consequence.

The order to think in

Decide first what you are actually trying to stop: model training, appearing inside an AI answer, having text lifted, or being in Search at all. Those are four different objectives and each has exactly one correct control.

For most lead-generation businesses the answer is that none of them should be touched. Appearing inside an AI answer is how a business gets shortlisted in a market where fewer people click through to compare. The businesses with a real case for exclusion are the ones whose product is the content itself.

What this means for an operator

Check what your robots.txt and your Search Console generative AI control currently say, because one of them is probably set from an assumption nobody rechecked. Then leave them alone unless your product is the content.