MODULE 2 · LESSON 4 / 5
Intelligence Grade™ — signal reliability
⏱ 12 min
Pro content
This module is available from 200€/mois. The content below is a full preview — progress will sync with your account once your plan is active.
Why the Intelligence Grade™ comes first
Before reading a score, a verdict, or a recommendation — check the Intelligence Grade™. It determines the reliability of the entire analysis.
Statistical analogy: A result of 75% with a ±30% confidence interval is practically unusable. Likewise, an INVEST verdict with a LOW Intelligence Grade™ (0–44) is worthless without a strong disclaimer. The grade is your confidence interval.
The 3 factors that determine the grade
The Intelligence Grade™ is computed automatically by the analytical pipeline. Three main factors influence the final result.
Patent volume
≥500 patents → +5 confidence points · <200 → −8 points · <100 → −15 points
Minimum threshold for INVEST/MONITOR: 50 patents
Metadata quality
Complete Filing Date + patent owner (Assignee) column → better computed grade
A non-normalized Assignee column (duplicates) reduces the reliability of Actor Intelligence
Source used
Tier 1 Pro (Derwent Innovation, PatSnap…) → HIGH grade (80–95%)
Hierarchy: Derwent Innovation > PatSnap > Questel Orbit > …
The two display rules
The pipeline automatically applies the following rules based on the computed grade.
Grade < 60Add qualification disclaimer when grade < 60 (limited signal)
Grade < 40Add strong disclaimer: treat findings as indicative only
Blocking guards tied to the grade
intelligence_grade < 40→INVEST → EXPLOREpatent_count < 50→INVEST/MONITOR → EXPLOREHow to improve your grade
A LOW Intelligence Grade™ is not a dead end — it is an indication of how to improve your dataset before drawing conclusions.
Aim for ≥200 patents
Ideally ≥500 for HIGH
Use Derwent Innovation
Tier 1 source — grade 80-95%
Filing Date + patent owner (Assignee)
Complete columns required
Combine 2+ sources
+5 bonus confidence points
Concrete case — improving a LOW dataset
Here is an example of real progression of an Intelligence Grade™ after improving the dataset.
BEFORE
- Expansion to 200+ patents via Espacenet (Tier 2)
- Added the complete Filing Date column
- Normalization of patent owners (assignees) — deduplication
AFTER
The Grade feeds Layer C — Tier gate
The categorical label of the Intelligence Grade™ directly drives the Layer C tier gate (Phase 5.3). A Grade of HIGH opens TIER_HIGH (categorical verdict preserved); MODERATE opens TIER_MODERATE (categorical verdict preserved + a confirm-first caveat, verdict_conditional stays false); LOW opens TIER_LOW (directional mapping: INVEST → OPPORTUNITY_SIGNAL · MONITOR → MIXED_SIGNAL · EXPLORE → WEAK_SIGNAL · AVOID → NEGATIVE_SIGNAL, rendered “— Conditional”); finally a Grade of VERY_LOW → INSUFFICIENT_DATA.
Key takeaways from this lesson
- 3 labels: HIGH(70–100) ●●● · MED(45–69) ●●○ · LOW(0–44) ●○○
- Grade < 40 → strong disclaimer + INVEST → EXPLORE guard
- Dataset < 50 patents → INVEST/MONITOR overridden to EXPLORE
- Recommended source: Derwent Innovation (Tier 1 — grade 80–95%)
- The Grade drives the Layer C tier gate: HIGH → TIER_HIGH (categorical verdict), MODERATE → TIER_MODERATE (categorical verdict + confirm-first caveat), LOW → TIER_LOW (directional signal “— Conditional”), VERY_LOW → INSUFFICIENT_DATA
QUIZ DE VALIDATION
1. An Intelligence Grade™ of 38 triggers which rule?
2. Which source improves the Intelligence Grade™ the most?
3. A dataset of 45 patents produces a verdict INVEST. What happens?