Restore / Research / On-Page Spam Example On-Page Spam Reference On-page query stuffing. The signatures Google reads and the manual-action categories they trigger.
Unnatural term density, hidden text via CSS positioning or font-color matching, anchor-text repetition. Each pattern carries a distinct manual-action category in the Search Console report. The diagnostic reads the rendered DOM against the indexable text before scoping the remediation.
The three signature patterns Google's reviewers flag
The Search Essentials spam-policy reference names a small, stable signature set. The patterns evolve in surface (CSS techniques shift, on-page templates change), the underlying signature does not.
Term density inflation. The same target query repeated across the page beyond what reader-first prose would carry. The signature reads on a per-page term-frequency analysis: a target phrase recurring with significantly higher density than the same phrase carries on competing pages that rank for the same query cluster. The page reads as written to a ranking target rather than to a reader.
Hidden text via CSS or markup. Text placed where a human reader cannot see it but a crawler can index it. The historical patterns: white text on a white background, font-size set to zero or near-zero, off-screen positioning via position: absolute; left: -9999px;, text inside elements with display: none; or visibility: hidden;, text covered by an absolutely-positioned image. Modern detection runs at the rendered-DOM layer, which means the technique has to fool both the indexable-text extraction and the rendering pass; both layers compare.
Anchor-text repetition on internal or outbound links. The same exact-match anchor phrase repeated across many internal or outbound links pointing at the same destination. The signature crosses with the Penguin link-quality surface for outbound anchors; on internal anchors it reads as page-level on-page manipulation. The healthy distribution is mixed (exact, partial, branded, generic) per the standard anchor-profile pattern; over-concentration on the exact-match form is the flag.
The detection surface, as a layered read
Google detects on-page query stuffing across three layers that compose:
Crawl-time indexing extraction. The standard text-extraction pass that pulls indexable text from the served HTML. Density analysis runs at this layer; the term-frequency signature is a per-page computation against the indexable-text surface.
Rendered-DOM comparison. The headless-browser pass that reads the painted output. The rendered text is compared against the indexable text; significant divergence flags hidden-text patterns. White-on-white, off-screen positioning, font-size-zero all show as mismatches between the painted state and the indexable state.
Manual review. Sites flagged at the automated layers route to the spam-action team for human review. The reviewers work from the Quality Rater Guidelines surface, which names spam patterns in operational language. The manual review produces the action category that lands in the Search Console Manual Actions report.
The three layers run continuously. A site can carry an action from any layer without the others firing; the on-page spam category and the Helpful Content algorithmic surface are independent signals that read the same on-page input. Google's HCU covers the algorithmic surface in detail.
The Restore diagnostic for on-page spam patterns
The diagnostic step runs the same shape regardless of whether the Manual Actions report is populated. We pull the affected pages, render each one through a headless browser, and read the rendered DOM against the indexable text. The output is a per-page table: density values for the target query cluster, hidden-element inventory (positioning, font-color, display state), anchor-text distribution for internal and outbound links.
From the table the routing is direct. Light density inflation routes to a rewrite engagement against reader-first prose. Hidden-text inventory routes to a markup audit and removal. Over-concentrated anchor distribution routes to the anchor-rebalance pass that runs against the homepage anchor profile (the seed set per SEO Penalty Removal doctrine).
Where a manual action is involved, the reconsideration request follows the three-element structure: exact quality issue (the specific signature pattern identified), specific steps taken (per-page edits with file paths and commit references), outcome documentation (rendered-DOM verification on the remediated pages). Reviews typically take days to weeks; partial remediation is the most common rejection cause, so the engagement covers every affected page before the request goes out. The authoring engagement is authoring the reconsideration request; the wrapper engagement is manual penalty removal.
Adjacent references: Google manual action categories for the complete taxonomy this page's signatures sit inside, how to do a reconsideration request for the operational requirements that govern the submission, spam link patterns for the outbound anchor side. The diagnostic itself runs at the forensic SEO diagnostic.
What operators ask when a density or hidden-text pattern surfaces.
- 01.
What is on-page query stuffing as Google defines it?
Loading a page with repeated terms or numbers intended to manipulate ranking rather than communicate to readers. The Search Essentials spam-policy reference names three signatures: lists of phone numbers or cities without substantive context, blocks of text repeating the same target term across the page, repetition that reads as nonsensical to a human reader. The pattern is page-level; the surface that detects it is sitewide pattern analysis plus manual review.
- 02.
What is hidden text and how does Google detect it?
Text intentionally placed where readers cannot see it but crawlers can: white text on a white background, off-screen positioning via CSS, font-size set to zero, text behind an image, content hidden inside non-displayed elements. Detection runs at the rendered-DOM layer. Google's headless rendering reads the painted output and flags mismatches between the rendered visibility state and the indexable text. The violation category is hidden text within the broader spam-policy surface.
- 03.
What manual-action category does this trigger?
Two routes depending on severity. Light density inflation routes into thin-content-with-little-added-value. Heavy, deliberate manipulation routes into pure-spam. The hidden-text pattern carries its own category line in the Search Console Manual Actions report. A site can carry multiple categories at once when the on-page spam pattern overlaps with link manipulation or scaled-content abuse. The complete category set is at Google manual action categories.
- 04.
Does this overlap with algorithmic demotion?
Yes. The Helpful Content System signals (post-March 5, 2024 redistributed across core ranking systems) read the same surface that on-page query stuffing inflates. A site can carry an algorithmic demotion on the quality signal even when the Manual Actions report is empty. The diagnostic correlates the date of organic-traffic-loss against the named-update calendar to read which signal is involved. Google's HCU covers the algorithmic side.
- 05.
How does Restore remediate this?
Forensic audit first: pull the affected pages, read the rendered DOM against the indexable text, surface the density and hidden-text signatures. Remediation: rewrite the on-page surface against reader-first language, remove the hidden elements, normalize anchor-text distribution. Where a manual action is involved, the reconsideration request is authored to the three-element structure: exact quality issue, specific steps taken, outcome documentation. Manual penalty removal covers the engagement.
If your Search Console flagged an on-page spam category or the density and hidden-text audit needs running, book the diagnostic.
We render the affected pages through a headless browser, read the rendered DOM against the indexable text, surface the density and hidden-element signatures, and scope the remediation per page. Where a manual action is involved, the reconsideration request is authored to the three-element structure with the per-page evidence package.