Key Concepts
Definitions of the core patent concepts used across the 16 audit frameworks.
A
Agent Rank Google patent system for scoring author authority across topic domains and propagating that authority to documents authored by that person. The agent (author) accumulates trust through their publishing history, and that trust flows to new documents they create.
Anchor Text Distribution The mix of anchor text types in a site's backlink profile: branded, naked URL, generic, partial-match, exact-match. Healthy profiles have diversified distribution. Over-concentration on exact-match keyword anchors is a manipulation flag in the Historical Data patent.
B
Boilerplate Detection Patent US8898296B2 describes how Google's DOM tree analysis separates repeated template elements (navigation, footer, sidebar) from unique main content. Only the main content body receives full quality evaluation — boilerplate is weighted much less.
Bounce Pad A page that exists to rank in Google and pass users through to another destination without providing independent value. Explicitly identified as a negative quality signal in Panda patent analysis.
C
Co-Citation When your brand or entity is mentioned alongside other established entities in the same context, on an authoritative source. Co-citation strengthens your entity's association with the topical category of those entities.
Content Freshness Decay The gradual degradation of a page's freshness signal as it ages relative to its topic's expected update frequency. Defined in patent US 8,549,014 B2.
Crawl Budget The number of pages Googlebot will crawl on a site within a given timeframe. Determined by crawl rate limit (server capacity) and crawl demand (content freshness and popularity). A finite resource that should be allocated to important pages.
D
Disambiguation The process of resolving which specific meaning is intended for a polysemous (multi-meaning) term. Critical for entities like "Mercury" (planet, element, car brand, god) and for ensuring Google associates content with the correct word sense and queries.
Document Fingerprinting Patent US7734627B1 — Google generates a unique hash of page content to detect exact duplicates. Near-duplicates are detected via SimHash (70%+ similarity threshold).
Dwell Time The duration between a user clicking a search result and returning to the SERP. Long dwell time signals satisfaction; very short dwell time signals a pogo-stick (dissatisfaction). Formalized in patent US9558233B1.
E
E-E-A-T Experience, Expertise, Authoritativeness, Trustworthiness. Google's quality rater evaluation framework for page quality. Particularly important for YMYL content. Algorithmic proxies include author credentials, site authority, and third-party citations.
Entity Any uniquely identifiable "thing" in Google's knowledge model: Person, Organization, Place, Concept, Product, Event, Creative Work. Entities have attributes and relationships to other entities stored in the Knowledge Graph.
G
Geo-Semantic Index Google's multimodal geographic relevance system that combines text signals, image EXIF data, visual content analysis, and geographic metadata into a unified location relevance model.
H
Historical Data Patent US 7,346,839 — "Information retrieval based on historical data." Filed by Matt Cutts and Paul Haahr. Describes how Google builds temporal profiles for pages and domains, detects deviations from historical patterns, and uses those deviations to identify manipulation.
K
Knowledge Graph Google's structured database of entities and their relationships, containing 500+ billion facts about 5+ billion entities. Used to provide direct answers in search results and to understand the meaning behind queries.
L
Link Velocity The rate of new referring domains acquired per month. Natural link velocity follows gradual organic growth patterns with variance correlated to content publication events. Spikes without correlating events are a manipulation flag in the Historical Data patent.
Locally Prominent Semantic Features Geographic context signals identified in local search patents — specific neighborhood names, local landmarks, cultural references, area-specific business context — that distinguish genuinely local content from keyword-stuffed location pages.
N
Named Entity Recognition (NER) Natural language processing system that identifies and classifies entities in text. Google's NER identifies Person, Organization, Location, and other entity types without requiring explicit schema markup.
NAP Consistency Name, Address, Phone — the three core business identity attributes. Consistency across all web sources is the primary signal Google uses to resolve geographic entity ambiguity and build trust in business data.
P
Panda Quality Score The algorithmic quality classifier trained on human quality rater judgments. Operates at both page level and site level. Uses a sequential decision tree where content uniqueness is evaluated first. Described in patents US9135307B2 and US9031929B1.
Paraphrase Equivalence Google's knowledge that multiple different phrasings map to the same underlying concept. Content using only one phrasing for a concept misses the paraphrase-based indexing benefit; content using 3+ natural phrasings of the same idea gets indexed under all of them.
Phrase Co-occurrence The tendency for specific phrases to appear together in authoritative documents about the same topic. Identified and operationalized in Anna Patterson's Phrase-Based Indexing patent (US7536408). Missing co-occurring phrases = shallow topic coverage signal.
Pogo-Stick User behavior pattern where a user clicks a search result and immediately returns to the SERP (typically within 8-30 seconds) to click a different result. The strongest negative behavioral quality signal in Google's click feedback systems.
Q
QDF (Query Deserves Freshness) The algorithmic identification of queries where search results should prioritize recent content. Applied to breaking news, trending topics, seasonal queries, and time-sensitive information. Content targeting QDF queries must maintain freshness signals.
Quality Rater Guidelines (QRG) Google's internal guidance document for human quality raters who evaluate search results. The algorithmic quality systems are trained to replicate human quality rater judgments. Publicly available — reading it informs all E-E-A-T optimization.
R
Reasonable Surfer The model in Google's Reasonable Surfer patent that assigns click probability to each link based on position, visual prominence, anchor text quality, and context relevance. Links with high click probability pass more PageRank.
Referral-to-Page Ratio The ratio of pages receiving traffic signals to total crawled pages on a site. A Panda patent signal: sites with too many pages relative to their traffic signals have a poor ratio that suppresses site-level quality scores.
S
Selection Quality Score Google's per-result quality score derived from user click behavior — including CTR relative to position, dwell time, and whether users returned to SERP to click other results. Formalized in patent US9558233B1.
SimHash Near-duplicate detection algorithm used by Google to identify pages with ~70%+ content similarity even without exact text matches. Used in duplicate content risk detection.
Site-Level Quality Score The Panda patent's mechanism for propagating quality signals across an entire domain. Calculated as a ratio: (pages with positive quality signals) / (total crawled pages). Poor ratio suppresses rankings for all pages on the domain.
Social TrustRank The application of PageRank's authority-propagation model to social graphs. Trust flows from high-authority social entities to content they link, share, or mention — weighted by the quality of engagement from high-trust social accounts, not just raw volume.
T
Topical Depth The extent to which content extends beyond the surface definition of a topic into subtopics, related concepts, prerequisites, downstream applications, and edge cases. A deep topical signal distinguishes shallow content from genuine expert content in Google's phrase-based evaluation.
W
Web Decay The gradual degradation of a site's quality signals through neglect — broken links, 404 errors, outdated content, abandoned pages. The Historical Data patent's temporal model detects sites where maintenance patterns indicate low editorial investment.
Word Sense Disambiguation Determining which specific meaning of an ambiguous word is intended in context. Critical for polysemous terms that appear in your content — surrounding context words are the primary disambiguation signal Google uses.
YMYL (Your Money or Your Life) Google's category for content where low quality could directly harm users' health, financial wellbeing, safety, or society. YMYL content is held to the highest E-E-A-T standards. Categories include medical advice, legal guidance, financial advice, safety information, and civic content.