> ## Documentation Index
> Fetch the complete documentation index at: https://docs.planeconnection.com/llms.txt
> Use this file to discover all available pages before exploring further.
# SmartScore Methodology
> Technical reference for the SmartScore composite safety metric -- pilot-level scoring, operator-level Insurer SmartScore methodology, pillar definitions, temporal smoothing, freshness tiers, and data firewall.
SmartScore is PlaneConnection's proprietary composite safety and operational quality metric. It operates at two levels: a **Pilot SmartScore** (0--100) for individual crew members and an **Operator SmartScore** (250--1000) for the organization. This reference covers both, with detailed methodology for the Insurer SmartScore derivative used in insurance underwriting.
SmartScore is a proprietary system. Factor categories and general methodology are published for
transparency; specific algorithm weights and scoring formulas are not disclosed.
## Pilot SmartScore (0--100)
The Pilot SmartScore evaluates individual crew members across three broad areas and combines them into a single overall score using a proprietary weighting model.
| Component | What It Measures |
| ----------- | ------------------------------------------------------------------ |
| Experience | Depth and recency of flight experience. |
| Proficiency | Currency of training, certifications, and demonstrated competency. |
| Risk | Safety event history and risk exposure over time. |
Each component produces a sub-score. The three sub-scores are blended into the Overall Score through a proprietary algorithm that accounts for the relative importance of each dimension.
### Score Ranges
| Band | Range | Interpretation | Action Required |
| --------------- | :-----: | -------------------------------------------------------------------- | -------------------------------------------------------- |
| Excellent | 90--100 | Pilot demonstrates strong performance across all scoring dimensions. | None -- maintain current performance. |
| Good | 75--89 | Pilot meets all standards with minor areas for improvement. | Routine monitoring. |
| Fair | 60--74 | Pilot has notable gaps in one or more scoring areas. | Review contributing factors; consider targeted training. |
| Needs Attention | 0--59 | Pilot has significant deficiencies requiring intervention. | Mandatory management review; action plan required. |
### Fleet-Level Aggregation
The fleet SmartScore provides an aggregate view of pilot readiness across the organization.
| Metric | Description |
| ----------------- | -------------------------------------------------- |
| Fleet average | Mean SmartScore across all scored pilots. |
| Minimum score | Lowest individual SmartScore in the fleet. |
| Band distribution | Count and percentage of pilots in each score band. |
| Quarterly trend | Score comparison against the previous quarter. |
Fleet SmartScore appears on the safety dashboard and is included in quarterly management review data per 14 CFR 5.73.
### Pilot Detail View
| Element | Description |
| ------------------- | ---------------------------------------------------------------------- |
| Overall Score | Composite score with band indicator. |
| Component Breakdown | Visual representation of Experience, Proficiency, and Risk sub-scores. |
| Trend History | Score trajectory over the last 12 months. |
| Actionable Items | Expired or approaching-expiry items with links to relevant records. |
### How Scores Update
SmartScore recalculates automatically in response to changes in the underlying data -- flight
activity, training completions, safety events, and other operational inputs. A nightly batch
ensures all scores remain current even when no individual triggers occur.
***
## Operator SmartScore (250--1000)
The Operator SmartScore assesses an organization's overall safety and operational quality. It exists in two forms:
| Aspect | Internal SmartScore | Insurer SmartScore |
| -------------------- | ------------------------------------------------------ | ---------------------------------- |
| **Purpose** | Safety improvement coaching | Insurance underwriting support |
| **Audience** | Operator only (never shared externally) | Shared at operator's discretion |
| **Data sources** | Zone A operational metrics + safety culture indicators | Zone A operational metrics only |
| **Granularity** | 15 dimensions, full detail | 4 pillars + composite (250--1000) |
| **Update frequency** | Real-time | Quarterly (90-day rolling average) |
| **Recommendations** | Specific improvement actions | Not included |
### Score Bands (Operator-Level)
| Band | Score Range | Interpretation |
| --------------------- | ----------- | -------------------------------------------------- |
| **Excellent** | 850--1000 | Consistently strong performance across all pillars |
| **Good** | 700--849 | Solid performance with minor areas for improvement |
| **Adequate** | 550--699 | Meets baseline standards with notable gaps |
| **Below Average** | 400--549 | Significant areas requiring attention |
| **Needs Improvement** | 250--399 | Material deficiencies across multiple pillars |
***
## Insurer SmartScore Methodology
The Insurer SmartScore is a **4-pillar derivative** of the full 15-dimension Internal SmartScore, designed for insurance underwriting.
### Four Pillars
#### 1. Organizational Foundation
Structural elements that support safe operations:
* Years in operation and organizational stability
* Fleet size, composition, and aircraft age
* External certifications (IS-BAO stage, ARGUS rating)
* Certificate status and type
#### 2. Operational Excellence
Day-to-day operational quality:
* Maintenance compliance rates and overdue item counts
* AD compliance rate
* MEL usage patterns
* Dispatch quality and flight following coverage
* Route complexity management
#### 3. Safety Risk Management
Maturity of safety risk management processes:
* Reporting rate threshold (binary -- see below)
* Training completion rates
* Duty/rest compliance
* Presence of systematic safety processes
#### 4. Human Capital & Resilience
Crew qualifications, experience, and organizational capacity:
* ATP certificate percentage
* Average total flight hours
* Recurrent training completion rates
* Simulator training hours
* Crew currency rates
### Pillar Weighting
Each pillar contributes to the composite score with proprietary weighting. Weights are calibrated against historical loss data and adjusted through actuarial validation. The composite is not a simple average -- pillars interact, and certain deficiencies carry more weight than others.
***
## Temporal Smoothing
Insurer-facing scores use a **90-day rolling average** to prevent insurers from inferring safety events through sudden changes in operational metrics. Scores are delivered quarterly, not in real time.
This means:
* A single bad week does not drastically change your Insurer SmartScore
* Sustained improvements are reflected over the following quarter
* Short-term fluctuations are smoothed out
The Internal SmartScore updates in real time and is not subject to temporal smoothing.
***
## Reporting Rate Threshold
The reporting rate metric is a **binary threshold**, not a linear scale:
* **Meets healthy minimum:** >= 1 report per 10 employees per month
* **Below minimum:** fewer than this threshold
There is no additional scoring benefit for higher reporting rates. This prevents gaming (submitting low-quality reports to inflate scores).
### Quality Filters
Reports must meet quality criteria to count toward the threshold:
* Progressed to assessment or corrective action (not immediately closed)
* Assessment contains substantive content (minimum 50 words)
* Time-to-close is within normal range for the report category (not rubber-stamped)
Anomaly detection flags both sudden spikes and systematic low-quality inflation patterns.
***
## Score Freshness
Every score delivery includes freshness metadata:
| Age | Status | Insurer Guidance |
| ------------ | ----------- | -------------------------------------------------- |
| 0--30 days | **Current** | Acceptable for all underwriting decisions |
| 31--90 days | **Valid** | Acceptable; recommended refresh for large policies |
| 91--180 days | **Stale** | Warning; refresh before use |
| 180+ days | **Expired** | Must not be used; require new report |
Every response includes: `generated_at`, `valid_until`, `data_as_of`, and `freshness` enum.
***
## Granularity Tiers
Different audiences receive different levels of detail:
| Audience | What They See |
| ---------------------------- | -------------------------------------------------------------------------------- |
| **Operator** (internal) | Full 15 dimensions, sub-scores, all reason codes, improvement recommendations |
| **Insurer** (standard) | Composite score, 4 pillar scores, top 5 reason codes, confidence band, trend |
| **Insurer** (summary) | Composite score + trend direction only |
| **Benchmarking** (aggregate) | Anonymized percentile band ("top quartile"), minimum bucket size of 15 operators |
***
## Reason Codes
Reason codes explain the top factors influencing a score, modeled after FICO credit score factors:
* **Top 5 positive** -- strongest contributing factors (e.g., "Fleet maintenance compliance rate exceeds 98%")
* **Top 5 negative** -- weakest contributing factors (e.g., "Three overdue maintenance items in trailing 90 days")
Reason codes are generated from the underlying dimension scores and use plain-language descriptions. They are included in standard-tier insurer deliveries and in all operator views.
***
## Confidence Assessment
Confidence reflects data completeness, not score quality:
| Completeness | Label | Confidence Interval |
| ------------ | ----------------------- | ------------------- |
| 80--100% | **High confidence** | Narrow |
| 60--79% | **Moderate confidence** | Moderate |
| 40--59% | **Limited confidence** | Wide |
| Below 40% | **Low confidence** | Very wide |
A score with low completeness is **not penalized** -- the point estimate is unbiased. Low completeness widens the confidence interval, meaning the true score could be higher or lower than reported.
***
## Data Firewall (Zone A / Zone B)
The scoring engine accesses only Zone A data. Zone B data is physically separated:
| Category | Zone | Score-Eligible | Examples |
| ---------------------- | ---- | -------------- | --------------------------------- |
| Fleet composition | A | Yes | Aircraft count, types, age |
| Flight activity | A | Yes | Total hours, route complexity |
| Maintenance compliance | A | Yes | AD compliance, overdue items |
| Crew qualifications | A | Yes | ATP %, average hours |
| Training completion | A | Yes | Recurrent rates, sim hours |
| Dispatch quality | A | Yes | Briefing rate, flight following |
| Duty/rest compliance | A | Yes | Violations, rest compliance |
| Certifications | A | Yes | IS-BAO, ARGUS, certificate status |
| Hazard reports | B | **No** | Safety reports, near-misses |
| Investigations | B | **No** | Findings, evidence |
| Corrective actions | B | **No** | Remediation details |
| NTSB notifications | B | **No** | Reportable events |
| Confidential reports | B | **No** | Identity-protected submissions |
| Risk assessments | B | **No** | Internal risk analysis |
Zone A and Zone B use separate databases with separate encryption keys. The firewall is enforced through code-level import restrictions, automated testing, and lint rules.
***
## Related
Overview of the two-score architecture, data stewardship, and PBC commitment.
How to read scores, reason codes, and confidence indicators.
Tutorial for reading and acting on safety performance data.
How SmartScore fits into the broader safety assurance framework.
Safety Performance Indicators that complement SmartScore.
Role definitions and qualification requirements for scored crew members.