On-page & contentIntermediate

Auto-Generated Content

Auto-generated content is text produced programmatically. It violates Google's spam policies when made at scale mainly to manipulate rankings.

Also called: Automatically generated content, Machine-generated content

Auto-generated content is page text produced by software rather than written by a person: from templates, data feeds, scraped sources, automated translation or AI models. Automation is not banned. Google's policies target automated content generated primarily to manipulate search rankings, without adding value for users. Since March 2024, Google covers this under its scaled content abuse policy.

Acceptable and abusive automation

Plenty of useful pages are generated automatically: weather pages, sports scores, product specs from a database, documentation generated from code. What makes them useful is that the underlying data is real and the page serves a need.

  • Problematic: text stitched together from search results or other pages
  • Problematic: automated translation published without human review
  • Problematic: rewritten copies of other content using synonyms or obfuscation
  • Problematic: pages that contain search keywords but make no sense to a reader
  • Usually fine: pages generated from your own structured data that answer a real need

The policy history

Google's guidelines long named "automatically generated content" as a spam tactic. In March 2024, Google broadened this into the scaled content abuse policy, which applies however content is produced: automation, humans or a mix. The test is purpose and value, not method.

Checking your own generated pages

Read a random sample of generated pages as a visitor would. Ask whether each page gives information a person could not get as easily from another page on your site, whether its data is accurate, and whether you would publish it if search engines did not exist.

Example

A template library with real data

Hypothetical spreadsheet tool Plumcourt generates a page for each template in its library from a database: a preview, the columns included, formulas used and a sample filled-in version. The pages are generated, but each one describes a real, usable template. A plan to auto-generate thousands of pages for keyword variations with no matching template was dropped.

Common mistakes

Assuming any automation is safe if the text reads fluently

Fluent text with no substance still fails the value test. Quality, not grammar, is the bar.

Questions

Programmatic pages are generated from templates and data, so technically yes. They are fine when each page has unique, useful data; they become spam when the pages exist only to target keywords.

Know the terms. Now win the searches.

We find the searches your buyers make, then build, publish and improve the pages that win them.