1. How detection works
- 1
Pattern matching
Deterministic rules — keyword density, hidden-text CSS, doorway-page redirect chains, cloaking mismatches. Fast, cheap, high-precision.
- 2
Structural analysis
DOM-level checks — visibility mismatches, unnatural link density, template-clone detection across the site.
- 3
AI quality assessment
LLM-based scoring on the page as a whole: does it read as helpful content, or as generated filler for ranking?
- 4
Composite score 0–100
Score bands: Clean 0–20 · Notice 21–40 · Warning 41–60 · Critical 61–80 · Severe 81–100 (blocks publish, triggers auto-fix).
2. The eight major policies
Scaled content abuse
Many pages produced primarily to rank, not to help. Applies whether written by humans, AI or a mix.
Site reputation abuse
Third-party content hosted on your domain to piggyback on your ranking signals. See below.
Cloaking
Showing different content to Googlebot than to users. Includes user-agent, IP and referrer-based swaps.
Doorway pages
Pages built to funnel users to a single destination they'd rather have reached directly.
Hidden text / links
Text or links invisible to users but readable by crawlers.
Keyword stuffing
Unnatural repetition of keywords past the point of readability.
Link spam
Manipulative inbound or outbound links — paid, exchange, PBN, comment.
Expired-domain abuse
Buying an expired domain to redirect its authority into unrelated content.
3. Site reputation abuse (the newest and most misunderstood)
Announced in 2024 and enforced 2024–2026, site reputation abuse targets third-party content published on a host's domain to benefit from the host's ranking signals — the classic pattern of a news site renting subfolders to coupon affiliates.
The policy applies even when the host has editorial oversight, even when the content is high quality, if the primary purpose is to piggyback on the domain's authority. Publishers who host guest content on their own domain need explicit editorial involvement and clear demarcation.
4. How WBP blocks each pattern
- Scaled abuse: page contracts refuse rows that don't meet entity, uniqueness and first-hand-data thresholds.
- Cloaking: WBP renders one HTML per URL. No user-agent branching in the theme or the plugin surface.
- Doorway pages: every WBP-generated page must serve its own primary intent — templates that funnel to a single destination fail the intent-diversity check.
- Hidden text / links: the DOM audit fires on save; visibility mismatches are blocked.
- Keyword stuffing: template density check flags any repeated term above the readable threshold.
- Link spam: outbound-link diversity gate on every template; the Affiliate Manager forces rel=sponsored on paid links.
- Expired-domain abuse: WBP surfaces domain-history signals on setup and refuses to migrate a template into a domain whose historical content doesn't match.
5. Recovering from a policy hit
If Search Console shows a manual action or you detect an algorithmic policy hit, the recovery path is: identify the pattern that triggered it (usually one of the eight above), fix the pattern at the template layer so it cannot recur, remove or rewrite the offending pages, then request reconsideration or wait for the algorithm to re-crawl.
WBP's audit dashboard maps every URL against every spam policy so the trigger is identifiable, not guessed.