# Forty pages for generated queries and what Google says about it | Illia BizTech

> Query fan-out is real. A separate page for each generated query is what Google names as a breach of its scaled content policy.

Source: https://illia-biztech.de/en/forum/vierzig-seiten-fur-fragen-die-niemand-eintippt-und-die-ki-selbst-erzeugt

1. [Home](/en)
2. [Forum](/en/forum)
3. Forty pages for questions nobody types and the AI generates itself

# Forty pages for questions nobody types and the AI generates itself

One page for every extra question the model is assumed to generate, forty in total.

* ##### Asked by  
BoldWrench665
* ##### Asked on  
September 27, 2026

We run a dog training school with online courses. An agency explained that the AI asks further questions in the background for every search, and now wants to build us one page for each of forty such assumed questions.

The argument sounds technically grounded. Forty pages for questions nobody types still strikes me as odd.

#### Answers

![Portrait of Illia Belichenko, technical specialist at Illia BizTech](/_ipx/f_webp,q_75/https://cms.illia-biztech.de/uploads/small_Illia_Belichenko_author_81f88864f0.png)

##### Illia Belichenko · Founder & Technical Lead

✓ Accepted answer September 28, 2026 

The technical half of that argument is correct. The conclusion drawn from it is described by Google as a violation, and both statements sit on the same page.

Google names two techniques. The first is grounding, described as a way to improve the quality, accuracy and freshness of AI responses by relying on its core Search ranking systems to retrieve relevant, up-to-date pages. The second is query fan-out, described as a set of concurrent related queries the model generates to gather more information.

Your agency turns that into a page per generated query. Google writes that doing so primarily to manipulate rankings or generative responses violates its scaled content abuse spam policy. Forty pages of that kind are a paid-for risk.

Something useful survives in the idea, and it is coverage. If a search generates further questions in the background, the page that handles the whole topic properly is the one that wins. In your case that is probably three to five good pages. Before anything gets built, look at where you stand in answers like these. An [AI visibility report](https://illia-biztech.de/en/ki-sichtbarkeitsbericht) from Illia BizTech sets that out.

##### 

#### [A provider wants to put us in the AI overview permanently and brought a contract](/en/forum/ein-anbieter-will-uns-dauerhaft-in-die-ki-ubersicht-bringen-und-legt-einen-vertrag-vor)

September 27, 2026

[](/en/forum/ein-anbieter-will-uns-dauerhaft-in-die-ki-ubersicht-bringen-und-legt-einen-vertrag-vor)

#### [Assistants name us without linking us, and the report says zero](/en/forum/assistenten-nennen-unseren-namen-verlinken-uns-aber-nicht-und-im-bericht-steht-null)

September 25, 2026

[](/en/forum/assistenten-nennen-unseren-namen-verlinken-uns-aber-nicht-und-im-bericht-steht-null)

#### [Our AI requests rise every month and they are counted by the user agent name](/en/forum/unsere-ki-zugriffe-steigen-jeden-monat-gezahlt-wird-nach-dem-namen-im-user-agent)

September 25, 2026

[](/en/forum/unsere-ki-zugriffe-steigen-jeden-monat-gezahlt-wird-nach-dem-namen-im-user-agent)

#### [A monthly report counts Google-Extended, and the log knows no such name](/en/forum/ein-bericht-zahlt-google-extended-das-logfile-kennt-diesen-namen-nicht)

September 24, 2026

[](/en/forum/ein-bericht-zahlt-google-extended-das-logfile-kennt-diesen-namen-nicht)

#### [Our contractor says the block has nothing to do with the lost mentions](/en/forum/unser-dienstleister-sagt-die-sperre-habe-mit-den-verlorenen-nennungen-nichts-zu-tun)

September 24, 2026

[](/en/forum/unser-dienstleister-sagt-die-sperre-habe-mit-den-verlorenen-nennungen-nichts-zu-tun)

**Share** 

**Topics** 
* Content
* AI Overviews

Google names two techniques behind its AI features. Grounding retrieves relevant, up-to-date pages through the core Search ranking systems. Query fan-out is a set of concurrent related queries the model generates itself to gather more information than the typed question contained.

The substitution happens at the next step. Observing that many queries are generated does not create a mandate to build a page for each one. Google names that approach as a violation once it serves manipulation, and points to a page that covers its subject completely.