Scaled Content Abuse
Google’s spam-policy name for generating many pages primarily to manipulate rankings rather than help users, regardless of whether a human, an AI, or a mix produced them.
Definition
The official definition is in Spam policies for Google web search: many pages generated for the primary purpose of manipulating search rankings and not helping users, typically large amounts of unoriginal content with little or no value, however it is created. Using generative AI to produce many pages without adding value is a named example. Google Search's guidance about AI-generated content makes the same intent-based point earlier: automation is not spam by itself; automation whose primary purpose is ranking manipulation is. The adjacent doorway abuse policy covers multiple city or region pages that exist to rank and funnel users to one destination.
The policy itself is established (it is official and unambiguous). What is not established is where any given page set falls, which is why it matters for Programmatic SEO, the Core 30, and 8-Pass AI Writing Pipeline.
It Works Until It Doesn't: AI Content Strategies That Backfire adds the practitioner picture. Lily Ray monitored 220+ sites showcased by AI content vendors and reports a consistent “Mount AI” shape (a label she credits to Glenn Gabe): 6–12 months of rising pages, a traffic peak 3–6 months later, then a decline that erases most of the gain within a year and often falls below the prior baseline (54% lost ≥30% of peak, 39% lost ≥50%, 22% lost ≥75%). These are third-party estimates (Ahrefs, Sistrix), and Ray explicitly claims no causation. She ties the pattern to the September 2023 helpful content update, the March 2024 core update, and this policy, formalized alongside it. She also reports an unconfirmed late-January 2026 update: 40+ sites down 40–95%, centered on self-promotional listicles. Caleb Ulku relays the study accurately in They Lied About AI Content (This Proves It) and adds a local-SEO reading.
Key Properties
- Intent, not authorship: human, AI, or mixed production is irrelevant. (official)
- Works before it fails: every at-risk template ranks and earns AI citations at first (“Indeed they do!”), which is why the failure is delayed and expensive. (Ray)
- Eight templates seen on declining sites (Ray; most declining sites use three or four, the most aggressive all eight): comparison pages at scale; “what is X” glossary pages; “best X for Y” listicles; self-promotional listicles; competitor-alternative pages; programmatic location and language pages; FAQ farms; off-topic content at scale.
- Two reach local SEO (Caleb’s extension; Ray does not discuss local): FAQ farms built from People Also Ask, and service pages multiplied across cities the business is not in.
- Footprint theory (Ray’s stated belief, expanded by Caleb; not tested by her data): formulaic, competitor-repeatable templates become a detectable signature once enough sites publish them.
- Spillover: impact often covers the whole blog or subfolder, sometimes the full domain, not just the offending pages. (Ray)
- SEO risk is AI-search risk, “largely because of RAG.” (Ray)
- Pre-publish test (Ray, one of five questions): could a competitor publish a near-identical page tomorrow with the same prompt? See Information Gain.
- Recovery pattern (Ray): sites that came through 2023–2024 best “prioritized quality, originality, and topical focus over scale”; brands still growing in her data generally do not match the eight templates.
Examples from Sources
| Example | Source |
|---|---|
| Generative AI used to create many pages without adding value (named policy example) | Spam policies for Google web search |
| City/region pages that funnel to one page (doorway abuse) | Spam policies for Google web search |
| ”Plumber Houston, plumber Katy, plumber Sugar Land”: one template, city name swapped, 200 pages | They Lied About AI Content (This Proves It) |
| B2B software sites publishing jokes and baby names for keyword volume | They Lied About AI Content (This Proves It) |
In the Sources
| Source | Context |
|---|---|
| Spam policies for Google web search | Official definition; AI-at-scale named; doorway abuse adjacent. |
| Google Search's guidance about AI-generated content | AI is fine; AI used to manipulate rankings is spam. |
| Creating helpful, reliable, people-first content | Who / How / Why self-check, including disclosure of automation. |
| It Works Until It Doesn't: AI Content Strategies That Backfire | Primary: 220-site pattern, eight templates, January 2026 update, safe-use guidance. |
| They Lied About AI Content (This Proves It) | Accurate relay of Ray, plus local implications. |
Tension
- Where does the Core 30 sit? Caleb says it is safe because it is bounded by the GBP and targets only the GBP’s city. The same creator’s earlier videos promote scaled neighborhood and landmark pages (Programmatic SEO). He does not reconcile the two.
- Duplicate Content (for Service-Area Pages): James Dooley and Joy Hawkins report copied service-area pages outranking rewritten ones. That is a claim about what ranks today; this policy is about what Google says it acts against. Both can be true for a while, which is exactly the delay Caleb describes.
Related
Concepts: Information Gain, Programmatic SEO, Core 30, Duplicate Content (for Service-Area Pages), 8-Pass AI Writing Pipeline, E-E-A-T (Experience, Expertise, Authoritativeness, Trust), GEO / AEO (Getting Recommended by AI) Entities: Google, Lily Ray, Glenn Gabe, Caleb Ulku, Core30 AI Tool, Ahrefs