Content Quality & Panda Patents Reference
153+ patents across 17 quality dimensions — the foundation of Google's site quality scoring system.
Site Quality Scoring (12 patents)
| Patent | Description |
|---|---|
| US9031929B1 | Site Quality Score — Navneet Panda, inventor |
| US9135307B2 | Panda Algorithm — generates alternative queries when top results are low-quality; applies scaling factor (R0+1)/R1 |
| US8442984B1 | Website quality signal generation |
| US8843477B1 | Onsite and offsite search ranking |
Panda Formula:
Site Quality Score = (R0 + 1) / R1
Where:
R0 = relevance score of results WITHOUT Panda adjustment
R1 = relevance score WITH Panda adjustment (lower R1 = worse quality)Content Scoring (8 patents)
| Patent | Description |
|---|---|
| US8707459B2 | Content originality — original-to-copied ratio |
| US8898296B2 | Boilerplate detection — DOM tree shape + text ratio |
| US20070067294A1 | Readability level matching to audience |
| US9767157B2 | N-gram quality prediction |
| US9959315B1 | Passage quality scoring for snippets |
Freshness (6 patents)
| Patent | Description |
|---|---|
| US8549014B2 | Content freshness — update frequency, temporal changes |
| US8521749B2 | Page age signals |
| US7346839B2 | Historical data patterns, link velocity |
Freshness scoring factors:
- Time since last substantive update
- Frequency of updates over time
- Type of content (news = high freshness requirement; evergreen = lower)
- Query freshness demand (QDF — query-dependent freshness)
Duplicate Detection (13 patents)
| Patent | Description |
|---|---|
| US7734627B1 | Document fingerprinting — term relationships |
| Multiple | Continuation patents for near-duplicate and cross-domain detection |
Fingerprinting uses term co-occurrence patterns rather than exact string matching — catches paraphrased and AI-rewritten duplicates.
Author / Publisher Quality (6 patents)
| Patent | Description |
|---|---|
| US8150842B2 | Author reputation scoring |
| US8126882B2 | Author credibility from online content |
| US8645396B2 | Reputation scoring over time |
| US8606792B1 | Scoring authors of posts |
| US11275895B1 | Author Vectors — writing style fingerprint |
| US7565358B2 | Agent Rank — author credibility via signed links |
Author Vectors (US11275895B1): Neural network trained on author's writing style creates a fingerprint. Consistent style attribution signals genuine human expertise.
Content Classification (35 patents)
| Patent | Description |
|---|---|
| Website Representation Vector (2018) | Expert/Apprentice/Layperson scoring for YMYL content |
| US10108694 | Content clustering for topical authority |
| US8224827B2 | Document classification-based ranking |
YMYL Classification (Website Representation Vector):
- Expert level: MD, JD, CPA credentials visible; institution-backed content
- Apprentice level: Experience-based writing, practitioner viewpoint
- Layperson level: General public perspective — penalized for health/finance/legal topics
Spam Detection (14 patents)
| Patent | Description |
|---|---|
| US7533092B2 | Link-based spam detection |
| US7953763B2 | Detecting link spam in hyperlinked databases |
| US7603345 | Spam pattern identification |
Trust Signals (6 patents)
| Patent | Description |
|---|---|
| US7603350B1 | TrustRank — seed site trust propagation |
| US8352467B1 | Search result ranking based on trust |
| US8818995B1 | Trust-based ranking (continuation) |
| US10268641B1 | Trust ranking (2019 continuation) |
| US8554601B1 | Managing content based on reputation |
TrustRank mechanism:
- Start with manually verified trusted seed sites
- Propagate trust through outbound links (good sites link to good sites)
- Sites far from seed sites receive lower trust scores
- Spam sites cannot easily receive high trust (seed propagation decay)
UGC Quality (7 patents)
| Patent | Description |
|---|---|
| US8965883B2 | User credential scoring |
| US9792330B1 | Local expert identification |
Ranking / Relevance (24 patents)
| Patent | Description |
|---|---|
| US9218397B1 | Query-dependent ranking factors |
| EP3005168A1 | Featured snippet eligibility |
| WO2020081082 | Information Gain Scoring |
Information Gain (WO2020081082): Measures how much unique, new information a page provides compared to existing content on the same topic. Pages that add genuinely new information rank higher than those restating existing facts.
Panda Quality Audit Checklist
Apply these checks to every page, each grounded in a specific patent:
[ ] Originality >80% (US8707459B2)
[ ] Boilerplate <40% of DOM (US8898296B2)
[ ] Readability grade 8-10 general / professional for B2B (US20070067294A1)
[ ] Zero spelling/grammar errors (US6424983B1)
[ ] Substantive update within appropriate interval (US8549014B2)
[ ] Full query intent coverage (US9031929B1)
[ ] All relevant entities mentioned with context (US8594996B2)
[ ] Each H2 section contains standalone answer passage (US9959315B1)
[ ] Author identified with credentials (US8150842B2)
[ ] Statistics cited with primary sources (US9684871B2)Related Learning Modules
- Module 14: Content Quality & Panda — Full Panda framework
- Module 20: E-E-A-T & Citations — Author authority patents
- Panda Quality Score Audit — Scoring methodology