# Scaled content abuse: what Google's spam policy bans

> What Google's scaled content abuse policy bans, the examples it gives, where AI fits, how bulk pages in a store can trip it, and how to audit your site.

URL: https://clione.ai/glossary/scaled-content-abuse/
Lang: en

If your store publishes pages in bulk (one per city, one per color and size combination, hundreds of descriptions written in one go by a model), this is the Google policy to read before you hit "publish." It does not punish a big catalog: a catalog with thousands of products has thousands of legitimate [product pages](/glossary/product-page/). It punishes **manufacturing pages to rank, not to answer anyone**. Google's name for it is _scaled content abuse_, and that is the name you will see in audits and in Google's own documentation.

## What it prohibits, exactly

Google defines scaled content abuse as generating many pages with the primary purpose of manipulating Search rankings rather than helping users. It gives five examples:[^google-spam-policies]

- Using generative AI tools, or similar tools, to generate many pages without adding value for users.
- Scraping feeds, search results or other content to generate many pages, including through automated transformations such as swapping in synonyms, translating or other ways of disguising it, when they add little value.
- Stitching or combining content from different web pages without adding value.
- Creating multiple sites to hide the fact that the content is made at scale.
- Creating many pages whose content makes little or no sense to a reader but contains search keywords.

The word that decides everything is **purpose**. None of the five examples bans a technique: each one pairs a quantity ("many pages") with a lack of value for the reader.

The policy has had this name since March 5, 2024. Before that, Google went after the use of automation to generate low-quality or unoriginal content at scale; since that day, the focus is producing content at scale to boost rankings, **whether automation, humans or a combination of the two are involved**.[^google-march-2024] With this and other measures announced at the same time, Google expected to reduce low-quality, unoriginal content in its results by 40%; it later added to the announcement that, as of April 19, 2024, the reduction was 45%.[^google-march-2024]

### Where AI fits

Google does not ban generative AI: it says it can be especially useful for researching a topic and for adding structure to original content. What it warns is that using it to generate many pages without adding value for users may break this policy.[^google-genai-content] The same guidance asks you to check the accuracy of everything generated by hand before publishing, because models can make up facts (known as [AI hallucination](/glossary/ai-hallucination/)), and it extends that review to metadata: the `<title>` element, the [meta description](/glossary/meta-description/), the [structured data](/glossary/structured-data/) and the [alt text](/glossary/alt-text/) of images.[^google-genai-content]

For stores, the guidance points to the [Google Merchant Center](/glossary/google-merchant-center/) policies on AI-generated content: generated images must carry the IPTC `DigitalSourceType` metadata with the value `TrainedAlgorithmicMedia`, and generated product data, such as the title and description, must be submitted separately and marked as AI-generated.[^google-genai-content]

### Where AI search fits

Google's guide to its [AI search](/glossary/ai-search/) features applies the policy to a new temptation. Creating separate content for every possible variant of a search, for example for the related queries the model launches with [query fan-out](/glossary/query-fan-out/), **breaks this policy if it is done mainly to manipulate rankings or generative AI answers** in Search. It adds that it does not work in the long run either, because having many pages does not make a site higher quality or more relevant.[^google-ai-guide]

## How it is detected and what happens

Google detects violations with automated systems and, when needed, with human review, which can end in a manual action. A site that breaks its spam policies **can rank lower in the results or not appear in them at all**.[^google-spam-policies]

The manual action is the visible part. Google applies one when a human reviewer decides that some pages on the site do not comply with its spam policies, notifies you in the manual actions report in [Search Console](/glossary/google-search-console/) and in its message center, and it can affect part of the site or all of it.[^gsc-manual-actions] Scaled content abuse is named explicitly under the action for pure spam, alongside cloaking; thin affiliate pages, scraped content and doorway pages have their own action, the one for thin content with little or no added value.[^gsc-manual-actions] Because detection is also automatic, **an empty report does not prove everything is fine**.[^google-spam-policies] To ask for a review, Google wants the request to explain the exact problem, the steps you took to fix it and the result.[^gsc-manual-actions]

## In a store

**Fictional example.** A shoe store with a few hundred models weighs five ways to get more pages. The table crosses them with the policy's examples. Matching an example is not an automatic conviction, because what decides is purpose, and only the store knows that.

| What the store is considering                                                         | Which Google example it matches                                                           | Why                                                         |
| ------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------- | ----------------------------------------------------------- |
| One page per model, with measurements, materials and original photos                  | None                                                                                      | Each page describes a product that exists and can be bought |
| "Running shoes in" plus the name of every city in the country, with the same text     | Many pages that make little sense to a reader but contain keywords[^google-spam-policies] | The text does not change; only the keyword does             |
| Importing the supplier feed and running the descriptions through a synonym rewriter   | Scraping feeds and transforming them with synonyms[^google-spam-policies]                 | They are the same listings other stores have, in disguise   |
| Descriptions for the whole catalog generated with AI from the title alone, unreviewed | Generative AI for many pages without adding value[^google-spam-policies]                  | The model knows nothing that is not in the title            |
| One page for every question an assistant might ask, all with the same answer          | Separate content for each search variant or fan-out query[^google-ai-guide]               | Google names it explicitly in its AI guide                  |

The last four rows have something in common: **the page exists for the query, not for the customer**. The way out is not to stop growing but to make every new page carry something the one next to it does not: a product fact, a measurement, a photo, an answer the store knows and the rest do not.


Believing the policy is aimed at AI-written text, or that human-written text is safe. Neither is true: since March 2024 Google looks at purpose and value, not the tool, and **human writing churned out in bulk to rank falls under the same policy**.[^google-march-2024] AI used to research or to structure original content, on the other hand, Google considers useful.[^google-genai-content]


## How to check

1. **Open the manual actions report in Search Console.** If there is one, the description says which pattern of pages it affects, whether part of the site or all of it.[^gsc-manual-actions]
2. **List the pages that come from a template** (by city, by filter, by question) and write down, for each type, what each page carries that the one next to it does not. If the answer is "the keyword," you are in Google's last example.[^google-spam-policies]
3. **Read a sample of what AI generated before publishing it**, metadata included: the title, the meta description, the structured data and the alt text.[^google-genai-content]
4. **If you use supplier descriptions**, compare one with the same listing on another store. If only a few synonyms differ, the page adds little.[^google-spam-policies]
5. **If you send your catalog to Merchant Center**, check that titles and descriptions generated with AI are submitted separately and marked as such.[^google-genai-content]

## Related concept: duplicate content

[Duplicate content](/glossary/duplicate-content/) gets confused with this policy because both involve many similar pages. Duplicate content is the same content on several URLs: each region's version, the mobile and desktop versions, the `http` and `https` versions, or the ones generated by a category's filters and sorting. Google says that **some duplicate content on a site is normal and does not violate its spam policies**, and it resolves it by choosing a representative address, the [canonical URL](/glossary/canonical-url/).[^google-canonicalization] Scaled content abuse, by contrast, is many pages manufactured with the purpose of ranking.[^google-spam-policies] The first gets sorted out with a tag; the second is fixed with what each page carries.

To go further, the [catalog enrichment](/glossary/catalog-enrichment/) entry explains how to give each listing the data that sets it apart, and the [AI search reading path](/learn/ai-search/) puts the rest of the terms in order.