User Behavior, CTR & NavBoost Patents Reference
111 patents across 8 categories — the behavioral signals that re-rank search results in real time. Confirmed in the 2024 DOJ antitrust trial.
111 Patents Across 8 Categories
| Category | Patents | Key Focus |
|---|---|---|
| CTR / Click Data | 20 | NavBoost, click fraud detection |
| Engagement Metrics | 14 | Engagement scoring, social sharing |
| Dwell Time / Session | 10 | Time-based quality signals |
| Personalization | 25 | Browsing history, emotional state |
| Intent Modeling | 18 | BERT query understanding |
| Behavioral Ranking | 12 | Pogo-sticking, feedback signals |
| A/B Testing | 7 | Direct and indirect ranking effects |
| Satisfaction | 5 | Searcher satisfaction measurement |
CTR / Click Data (20 patents)
| Patent | Description |
|---|---|
| US10229166B1 | NavBoost — tracks click selections, post-click behavior, weights by user reliability, re-ranks results |
| US9684697B1 | Repeat clicks on resource groups |
| US7657626B1 | Click fraud detection — anomaly analysis, bot signatures, IP tracking |
| US7917491B1 | Click fraud prevention — cross-platform pattern comparison |
| US20150032533A1 | Real-time click validation |
Engagement Metrics (14 patents)
| Patent | Description |
|---|---|
| US10296642B1 | Engagement scoring for ranking |
| US8626823B2 | Social sharing as quality signal |
Dwell Time / Session (10 patents)
| Patent | Description |
|---|---|
| US9558233B1 | Selection quality score — time-based signal |
| US8255413B2 | Dwell time analysis |
| US8838587B1 | User dwell duration in session |
| US20170140049A1 | Session context for ranking — previous searches, pages visited |
Personalization (25 patents)
| Patent | Description |
|---|---|
| US10810270B2 | Search based on browsing history and emotional state |
Intent Modeling (18 patents)
| Patent | Description |
|---|---|
| US20230334045A1 | BERT integration for query understanding |
| EP3005168A1 | Featured snippet eligibility detection |
Behavioral Ranking (12 patents)
| Patent | Description |
|---|---|
| US8661029B1 | User feedback-based ranking |
| US9092510B1 | Pogo-sticking detection — return-to-SERP signal |
| US8117209B1 | Ranking based on user behavior signals |
A/B Testing (7 patents)
| Patent | Description |
|---|---|
| US11132700B1 | Direct + indirect effect identification for ranking tests |
Satisfaction (5 patents)
| Patent | Description |
|---|---|
| US8442984B1 | Website rating relationship |
| US20120130814A1 | Searcher satisfaction measurement (R metric) |
NavBoost Processing Flow (Confirmed — 2024 DOJ Antitrust Trial)
NavBoost key facts:
- Confirmed operational by Google engineer Pandu Nayak in 2024 DOJ trial
- Aggregates signals from MANY users — not individual clicks
- Weights signals by user reliability (not all users treated equally)
- Query-specific: NavBoost signals apply to specific query + result pairings
- Historical: Google keeps click data for up to 13 months
Anti-Manipulation Detection (From Patents)
Google detects and filters artificial click manipulation:
| Manipulation Type | Detection Method |
|---|---|
| Click farms | Abnormal volume, timing patterns, geography concentration |
| Bot traffic | Device fingerprint, behavior pattern analysis |
| Coordinated campaigns | Cross-reference click signatures across accounts |
| Proxy-based manipulation | IP analysis, browser fingerprinting |
| Purchased traffic | Engagement depth analysis (shallow = flag) |
Why individual manipulation fails: NavBoost requires statistically significant patterns across many REAL users. Artificial clicks from a small pool of IPs/devices are filtered by ML classifiers before they reach the ranking system.
The only reliable improvement: Create genuinely engaging content that satisfies search intent at a higher rate than competing pages.
CTR Benchmarks by Position
| Position | Expected CTR Range |
|---|---|
| 1 | 28-35% |
| 2 | 15-20% |
| 3 | 10-14% |
| 4-5 | 5-9% |
| 6-10 | 1-5% |
| Below 10 | <1% |
Rich snippets can significantly boost CTR — FAQPage, HowTo, and featured snippets increase real estate and visibility, improving CTR even at lower positions.
Pogo-Sticking (US9092510B1)
What it is: User clicks a result, quickly returns to the SERP, and clicks a different result.
What it signals: The clicked page did NOT satisfy the user's search intent.
What Google does: Negative signal applied to that result for that query. Competing pages that retained users get a positive signal.
How to prevent pogo-sticking:
- Answer the query immediately (above fold)
- Match the intent of the specific query exactly — not a related topic
- Load fast (slow loading = pre-click abandonment)
- Format content for scanning (headings, bullets, short paragraphs)
- Include a strong hook in the first 50 words
Dwell Time vs. Bounce Rate
| Metric | NavBoost Relevance |
|---|---|
| Long dwell + no return to SERP | Strong positive signal |
| Short dwell + return to SERP + click different result | Strong negative signal |
| Short dwell + close search (task complete) | Neutral (task completed elsewhere) |
| Long dwell + return to SERP | Ambiguous — content consumed but not fully satisfying |
Dwell time alone is not the signal — it's the combination of time AND behavior after leaving the page.
Related Learning Modules
- Module 13: User Behavior Signals — Full NavBoost deep dive
- Module 22: CTR & User Behavior Optimization — Practical CTR strategies
- Module 10: Modern Era Patents — DOJ trial confirmations