Topic
Quality & Prediction
The largest class, and the one that runs before ranking rather than during it: a set of models that predict how a human rater would score a page or a site. Most of what people mean by “E-E-A-T signals” lives here — and so does the calibration layer that decides what the predictions are trained against.
01Systems8
- 06BQ* (Balance)Q* is a summary quality score for the entire website. It works like a 'balance' made up of the site's authority (siteAuthority), a low-quality flag (lowQuality), and the overall NSR value. The score changes slowly: internal Google documents describe it as 'largely static'. It is a sluggish foundation, not a day-to-day signal.
- 07B/CNSR (+ Fallback inheritance)NSR (Normalized Site Rank) is the real quality core. It is a normalized quality and authority value, computed for individual 'chunks' (subsections) of a website. New sections initially inherit the average of their neighboring chunks ('fallback inheritance'). The historicized signal (predictedDefaultNsr) shows development over time: the trajectory matters, not today's single value.
- 08Bchard (+ YMYL/Hoax)chard rates the content quality of individual pages — calibrated against the judgments of human quality raters that Google trains according to strict guidelines (QRG). A variance score shows how confident the model is in its verdict. chard is especially sharp on YMYL content (Your Money or Your Life — topics with real life consequences such as health, safety, severe weather) and on suspected misinformation.
- 09BcontentEffortA language model estimates how much editorial effort and depth went into a page. The system distinguishes between shallow, machine-generated content and pages with recognizable research, structure and substance. The value is not a hard measurement but a model estimate, one that increasingly reliably separates thin AI content from genuine editorial engagement.
- 10B/Ctofu / keto / Rhubarb / SubchunksThese codenames stand for refinement signals that fine-tune the quality estimate at a very granular level: individual subsections (subchunks) of a page. Their exact function is not fully documented in the available material. They are not a standalone main system. They refine chard and NSR at the section level.
- 11B/COriginalContentScoreThis score measures how much first-hand material a piece of content contains: information found only there, that an AI could not reconstruct from other sources. The higher the score, the more original the content. It is the closest machine equivalent to the E-E-A-T dimension 'Experience' (lived experience and direct observation).
- 12AIS-Calibration (Rater/QRG)Not a live signal used daily in ranking — but the foundation on which all quality models are calibrated. The IS-Score (Information Satisfaction) summarizes how well a page meets a user's information need, rated by trained human raters. The QRG (Quality Rater Guidelines) are the public rulebook by which these raters judge.
- 13BTopic-Embeddings (Focus/Radius)Google computes a kind of 'topic map' from the entire text content of a website (embedding, a mathematical vector representation of the topic space). siteFocusScore measures how sharply the site's topic field is defined. siteRadius measures how widely it scatters thematically. A small radius means a tightly focused site. The system measures thematic consistency, not subject-matter competence.
02Deep Dives10
- Q* and P*: The Two Pillars of Google's Ranking ArchitectureThe DOJ trial revealed that Google's ranking ultimately reduces to two top-level signals — Quality (Q*) and Popularity (P*). This is what the evidence shows about how they work.
- IS-Calibration: The Human Rater System That Trains Google's Quality ModelsThe hardest evidence in the entire matrix — DOJ/sworn material confirms the rater system that calibrates every quality signal Google uses
- chard: Google's Page-Level Content Quality ModelThe leak field that rates individual page quality — calibrated by human raters, feeding into NSR, and why YMYL pages face stricter standards
- OriginalContentScore: How Google Measures Content OriginalityThe leak field that measures how much original material a document contains — information that is only found there and that an AI couldn't reconstruct from other sources
- contentEffort: How Google Estimates Editorial EffortThe AI model that estimates how much editorial work went into a page — and why 'effort' is becoming a ranking signal
- Topic-Embeddings: How Google Understands What Your Page Is AboutThe focus vectors and radius that determine topical relevance — and why a weather site needs a narrow focus to rank for weather queries
- tofu, keto, Rhubarb: The Subchunk Refinement SignalsCodenames for refinement signals at the most granular level, the individual subsections of a page. They are the weakest evidence in the matrix.
- Site Reputation Abuse: What Google Suspended in Europe — and What It Did NotWhat Google suspended in the EEA is the manual action — the replacement removes exactly what parasite SEO pays for · also in Penalty & Spam
- What Good Content Is: contentEffort and Google's Soft MetricsOne leaked field, one rater definition, twelve official questions, and a checklist that marks which parts are evidence and which are our reading.
- Links That Count: Five Link Tactics Against the Leak and the PatentsA popular link guide reads PageRank buckets, trusted anchors and context hashes out of Google's leak. The fields are real. The trusted sources are news sites, the bucket counts onsite anchors, and the cap is 200, not zero. · also in Penalty & Spam
03Leak fields14
- B
siteAuthorityAuthority distilled from quality_nsr. It is applied in Q*.Q* / NSR - B
lowQualityFlag for low quality, converted from quality_nsr.NsrData.Q* - B
predictedDefaultNsrBaseline quality score, stored with its history (VersionedFloatSignal), so the trajectory counts.NSR - B
nsrConfidenceConfidence in its own NSR judgment (deprecated, meaning marked as obsolete).NSR - B
chardScoreContent-quality value per document, calibrated against the judgments of human raters.chard - B/C
chardVarianceSpread, meaning uncertainty, of the chardScore. A high value means an uncertain judgment.chard - B
contentEffortCreation effort and depth of a page, as estimated by an LLM (large language model).contentEffort - B/C
OriginalContentScoreDegree of originality and first-hand material. It is the measure that comes closest to Experience for a machine (a proxy).OCS - B/C
tofu / keto / RhubarbCodenames for refinement and delta signals at the subchunk level.tofu-Familie - B
siteEmbeddingsThe site's topic space, expressed as a vector (a list of numbers).Topic-Embeddings - B
siteFocusScoreHow sharply outlined the site's topic field is.Topic-Embeddings - B
siteRadiusHow widely the site spreads across topic space. A small value means focused.Topic-Embeddings - A
IS-ScoreInformation Satisfaction, taken from the judgments of human raters. It serves as the calibration norm.IS-Kalibrierung - B/C
siteQualityStddevStandard deviation of quality across a site. It measures consistency.IS-Kalibrierung