What You Are Avoiding
Google defines scaled content abuse as "when many pages are generated for the primary purpose of manipulating search rankings and not helping users", and says it applies "no matter how it's created". Every example Google gives includes the condition "without adding value". So the job is not to hide that pages were made at scale. It is to make sure each one adds something. These six gates do that.
Gate 1: One Page per Entity
Decide what a page is about, and publish one page for it. Not a page per keyword variation. If you find yourself planning thirty pages about one product, you have one page and twenty-nine duplicates. Variations belong as sections on the one page.
Gate 2: Facts Before Words
Fill in the data for a page before anything is written. If a record has no fact unique to that page, do not generate the page. This is the gate that stops most problems, because it is applied before the cheap part.
Gate 3: The Swap Test
Put two pages from the same template side by side. If swapping the names would leave the rest unchanged, the template produces filler. Fix the data or drop the pattern.
Gate 4: Check Every Claim
Match each number, name and date to a source. Remove what cannot be matched. Google's guidance on generative AI content says to fact-check and review AI-generated content, and to apply the same care to titles, descriptions and structured data, since those appear in search results. See how to fact-check AI-written pages.
Gate 5: A Person Approves It
Someone reads each page before it is published. For a large batch, that means the batch is sized to what can be read properly. If nobody has time to read the pages, the batch is too big.
Gate 6: Publish in Batches and Watch
Ship a small batch, then look at how many pages are indexed and whether they earn impressions. A pattern where most pages stay unindexed is telling you something before you repeat it a thousand times.
A Checklist to Run on Any Page
| Question | Pass looks like |
|---|---|
| What search does this page answer? | One specific query a real person types |
| What is on this page that is on no other? | At least one sourced fact or finding |
| Where does each claim come from? | A source you can name for every figure |
| Can the reader finish here? | The answer is on the page, not a click away |
| Who read it? | A named person |
| When was it last checked? | A date, and a plan to check again |
Things That Do Not Help
- Rewording to look different. Google's examples include automated changes such as swapping synonyms. Variety in wording is not value.
- Spreading pages across several sites. Google lists "creating multiple sites with the intent of hiding the scaled nature of the content" as an example of abuse.
- Adding a paragraph of generic text to every page to raise the word count.
- Avoiding AI entirely. The policy is about value, not tools. Thin hand-written pages fail the same way.
If You Already Published Thin Pages
- Sort them. Pages with impressions or clicks, pages indexed with none, pages not indexed.
- Improve the ones with a reason to exist by adding the facts they lack.
- Merge near-duplicates into one stronger page and redirect the others.
- Remove the rest from the index. Google's policy says such content should be excluded from Search.
How We Apply This
On presenc.ai we keep one page per entity, hold a sourcing record with a confidence grade for every claim in a batch, and have corrected or removed our own pages when a figure could not be supported. A redirect replaced one page about a product that turned out never to have shipped. The process in full is in how to build programmatic SEO pages.
If You Want This Done for You
Presenc AI runs a team of AI agents that does this work, and a person reviews every page before it ships. Request access.
Sources
Quotations are from Google Search spam policies and Google's guidance on generative AI content, read November 13, 2026. The gates are our interpretation and practice, not a Google checklist, and following them is not a guarantee of any ranking outcome.