API Reference
API Reference

Enformion 3G Fraud Score

EEnformion's third-generation fraud score leverages advanced machine learning and a unique combination of entity resolution across 6,000+ data sources with consortium fraud intelligence from thousands of financial institutions. The model employs XGBoost and Random Forest algorithms to deliver superior fraud detection capabilities with proven performance improvements over competing solutions.

Key Differentiators

  1. Superior Performance: 273% improvement over off-the-shelf models, 259% improvement over consortium-based models
  2. Entity Resolution: Comprehensive identity verification across 6,000+ data sources
  3. Real-time Processing: API response times as fast as 200ms – 1000ms
  4. Continuous Learning: Monthly model updates capturing emerging fraud patterns
  5. Specialized Coverage: Enhanced data coverage for thin-file and underserved populations
  6. Comprehensive Features: 350+ fraud-specific indicators across six feature categories

Feature Categories

The 3G Fraud Score analyzes 80+ unique features across these categories:


1. Phone Intelligence2. Identity Elements3. Match Intelligence
Phone tenure stability and metrics

Carrier Change patterns and history

Usage behavior analysis

Deactivation/port behavior tracking

Device type assessment

Behavior-based-risk anomalies		Cross-reference consistency validation

Historical identity stability assessment

Document verification scores

Identity element associations

Change velocity patter analysis

Network connection Mapping		 Multi-source verification results

Blacklist and watchlist matching

Consortium data alignment

Source quality assessment

Cross-Validation Scoring

Authority source verification
4.Location Intelligence 5. Account/Credit Analysis5. Environmental Context
Address tenure and stability analysis

Movement and migration patterns

Geographic risk assessment

Property type validation

Address change velocity tracking

Location based behavioral patterns		Recent inquiry pattern analysis

Account velocity metrics

Credit inquiry timing assessment

Profile change frequency

Application behavior patterns

Account relationship mapping		Income level indicators.

Population density metrics

Geographic risk factors

Economic stability measures

Demographic pattern analysis

Regional behavior Baselines

API Integration

The Third Generation Fraud Score is available as an include parameter in the eIDV product:

Request Parameters

Standard eIDV parameters apply. The more identity elements provided, the more comprehensive the fraud assessment:

Required Elements

  • Name
  • Phone Number
  • Address
  • Email Address
  • Date Of Birth

Optional Elements For Enhanced Accuracy

  • Social Security Number (last 4 digits)
  • Driver's License information
  • Additional contact information
  • Transaction context data

Response Structure


{
    "3GFraudScore": {
        "outputScore": 0.75,
        "riskCategory": 4,
        "primaryReasonCode": "PHONE_VELOCITY_HIGH",
        "secondaryReasonCode": "ADDRESS_INCONSISTENCY", 
        "tertiaryReasonCode": "IDENTITY_NETWORK_RISK",
        "type": "3GFraudScore",
        "scoreVersion": "3.0",
        "processingTime": 185
    }
}

Response Fields

outputScore

  • Type: Float (0.0-1.0)
  • Description Primary fraud risk score, where higher values indicate higher fraud risk
  • Interpretation
    • 0.0 - 0.2: Very Low Risk
    • 0.2 - 0.4: Low Risk
    • 0.4 - 0.6: Intermediate Risk
    • 0.6 - 0.8: High Risk
    • 0.8 - 1.0: Very High Risk

riskCategory

  • Type: Integer (1-5)
  • Description: Simplified risk classification for quick decision-making
  • Values
    • 1: Very Low Risk
    • 2: Low Risk
    • 3: Intermediate Risk
    • 4: High Risk
    • 5: Very High Risk

primaryReasonCode

  • Type: String
  • Description: The most significant factor contributing to the fraud score
  • Purpose: Provides explainability for regulatory compliance and decision justification.

secondaryReasonCode/tertiaryReasonCode

  • Type: String
  • Value: Additional contributing factors to the fraud assessment
  • Purpose: Enables comprehensive understanding of risk factors for manual review cases

type

  • Type: String
  • Value: "3GFraudScore"
  • Description: Identifies this specific score type when multiple scores are returned

scoreVersion

  • Type: String
  • Description: API processing time for performance monitoring

Phone Intelligence Codes

  • PHONE_VELOCITY_HIGH: Unusual phone number activity patterns
  • PHONE_CARRIER_RISK: High-risk carrier or service type
  • PHONE_TENURE_LOW: Recently activated phone number
  • PHONE_DISCONNECT_PATTERN: History of frequent disconnections

Location Intelligence Codes

  • ADDRESS_VELOCITY_HIGH: Frequent address changes
  • GEOGRAPHIC_INCONSISTENCY: Unusual geographic patterns
  • HIGH_RISK_LOCATION: Location associated with fraud activity
  • ADDRESS_VALIDATION_FAIL: Address verification issues

Match Intelligence Codes

  • CONSORTIUM_MATCH: Matches known fraud patterns in consortium data
  • BLACKLIST_MATCH: Matches negative databases
  • DATA_QUALITY_LOW: Poor data verification across sources
  • VERIFICATION_INCONSISTENCY: Conflicting verification results

Performance Benchmarks

Based on extensive testing across multiple business lines:

Fraud Detection Rates (Top Decile Performance)

  • Check Cashing: 16.3% fraud capture rate
  • Retail Verification: 29.1% fraud capture rate
  • Card Not Present ACH: 25.3% fraud capture rate
  • Buy Now Pay Later: 24.2% fraud capture rate
  • Gaming (CNP): 25.8% fraud capture rate
  • Gaming (Brick & Mortar) 55.3% fraud capture rate

Model Performance Metric

  • Training Dataset: 480,000+ records with 2.5% fraud rate
  • Validation: 100,000+ out-of-time records
  • Top 5% Capture Rate: 44.2% of all fraud cases
  • Comparative Performance: 273% better than off-the-shelf solutions

Implementation Best Practices

Integration Guidelines

  1. Minimum Data Requirements: Provide at least name and phone number for basic scoring
  2. Enhanced Accuracy: Include address, email, and DOB for comprehensive assessment
  3. Real-time Processing: Integrate at account opening or high-risk transaction points
  4. Threshold Management: Establish risk thresholds based on your risk tolerance and business model

Recommended Decision Logic

IF riskCategory <= 2: AUTO_APPROVE

ELIF riskCategory == 3: MANUAL_REVIEW

ELIF riskCategory >= 4: AUTO_DECLINE or ENHANCED_VERIFICATION

Performance Optimization

  • Cache results for repeated queries within short timeframes
  • Implement fallback logic for API timeouts
  • Monitor processing times and adjust integration as needed
  • Use reason codes for automated decision documentation

Support and Compliance

Security & Compliance

  • SOC 2 Type II compliant infrastructure
  • End-to-end encryption for all data transmission
  • Regular security audits and updates

Technical Support

  • 24/7 technical support available
  • Dedicated integration engineering team
  • Regular performance reviews and optimization

Model Updates

  • Monthly model refreshes to capture emerging fraud patterns
  • Automatic deployment of model improvements
  • Performance tracking and reporting
  • Custom model tuning available for enterprise clients

Error Handling

Common Error Responses


{
    "error": {
        "code": "INSUFFICIENT_DATA",
        "message": "Minimum required identity elements not provided",
        "details": "At least name and phone number required for scoring"
    }
}

Error Codes

  • INSUFFICIENT_DATA: Missing required input parameters
  • INVALID_FORMAT: Data format validation failed
  • RATE_LIMIT_EXCEEDED: API rate limits exceeded
  • SERVICE_UNAVAILABLE: Temporary service unavailability
  • AUTHENTICATION_FAILED: Invalid API credentials