What is the Default Fraud Configuration?
Veryfi provides a default fraud detection configuration for all new CPG-type accounts that balances protection with operational efficiency.
This configuration activates key fraud detection capabilities from the Veryfi Fraud Detection Suite and Artifact Analysis while ensuring smooth integration for most use cases.
The default settings are designed based on industry best practices and can be fully customized using Veryfi's Business Rules engine to match your specific requirements, risk tolerance, and operational patterns.
Why Use Default Configuration?
Immediate Protection: Start fraud detection on day one without configuration overhead
Industry-Tested Settings: Based on analysis of fraud patterns across thousands of implementations
Balanced Approach: Optimized to catch fraud while minimizing false positives
Easy Customization: Serves as a starting point for fine-tuning based on your specific needs
Default Configuration Details
What's Enabled by Default
Document Classification & Analysis:
LCD Photo Detection: Identifies screen photography
Digital Background Analysis: Detects non-paper backgrounds
Not a Document Detection: Filters non-receipt submissions
AI-Generated Detection: Identifies artificially created documents
Artifact Analysis Features:
Handwriting Detection: Detects manual alterations
Digital Tampering Detection: Identifies software-based editing
Uniqueness Protection:
Duplicate Detection: Prevents resubmission of identical documents
Similar Documents Analysis: Identifies near-duplicate submissions
What's Disabled by Default
Device-Dependent Features (Require Veryfi Lens):
High Velocity Detection: Requires device info
Critical Velocity Detection: Requires device info
History Check: Requires device info
Profile Fraud Detection: Requires device info
Work-in-Progress Features:
Fraudulent PDF Analysis: Currently under development
Fraud Detection Implementation Approaches
Two Ways to Handle Fraud Detection Results
Users have complete flexibility in how they implement fraud detection logic using Veryfi's fraud signal results. You can choose between two primary approaches based on your operational requirements and complexity needs.
Option 1: Fraud Colors Concept
How it Works: Use Veryfi's built-in color-coded system that automatically assigns risk levels based on detected fraud signals:
{
"meta": {
"fraud": {
"color": "yellow",
"score": 0.65,
"types": ["handwritten_characters", "similar_documents"]
}
}
}Implementation:
Green: Process automatically, no review needed
Yellow: Flag for manual review or additional verification
Red: Block processing requires supervisor approval
Best For:
Quick implementation and poc
Gathering the initial users and behavioral patterns
Standard fraud detection workflows
Teams with limited fraud analysis expertise
Option 2: Custom Validation Logic Using Individual Fraud Flags
How it Works: Build your own business logic using the granular fraud detection fields and signals returned by the fraud suite:
{
"meta": {
"handwritten_fields": ["total", "line_items.total"],
"fraud": {
"types": ["handwritten_characters", "digital_tampering"],
"score": 0.72
},
"similar_documents": [
{
"similarity_score": 0.85,
"document_id": "123456786"
}
]
}
}Custom Logic Examples:
// CPG Campaign Possible Tagging Logic
if (handwritten_fields.includes("total")) {
tag = "reject";
} else if (digital_tampering.fields.includes("total")) {
tag = "manual_review";
} else if (similarity_score > 0.9) {
tag = "reject";
} else if (handwritten_fields.includes("line_items.total")) {
tag = "suspicious_modification";
} else if (ai_generated_detected) {
tag = "reject";
} else if (lcd_photo_detected) {
tag = "screen_capture";
} else {
tag = "approved";
}
Best For:
Cases when you only need few fraud signals to incorporate
Organizations with specific fraud patterns or requirements
Complex multi-stage approval workflows
Integration with existing compliance systems
Advanced fraud analytics and reporting needs
Which Approach Should I Choose?
Start with Fraud Colors if:
You're new to fraud detection
You want immediate protection with minimal configuration
Your use case fits standard fraud patterns
You prefer proven, industry-tested thresholds
Build Custom Logic with tags if:
You have unique fraud patterns or business requirements
You require granular control over each fraud signal
You have specific compliance or audit requirements
Tags might be easier and more flexible
Field-Level Configuration
Which Fields Are Monitored for Handwriting?
Default Handwriting Detection Fields:
"handwriting_enabled_fields": "total", "subtotal", "date", "line_items.total", "line_items.price"
These fields represent the most commonly manipulated areas in fraudulent submissions across CPG and expense management use cases.
Which Fields Are Monitored for Digital Tampering?
Default Digital Tampering Detection Fields:
"digital_tampering_enabled_fields": "total", "subtotal", "date", "line_items.total", "line_items.price"
The same critical financial fields are monitored for pixel-level manipulation and digital editing artifacts.
Why These Specific Fields?
Total & Subtotal: Most direct financial impact, highest fraud motivation
Date: Commonly altered to fit expense reporting periods or campaign timeframes
Line Items: A Frequent target for adding products or inflating quantities
Price: Individual item cost manipulation to reach thresholds or increase value
Similarity Detection Configuration
How Does Similar Documents Detection Work?
The default similarity detection uses configurable thresholds to identify similar submissions:
"similar_documents": "yellow_threshold": 0.8, "red_threshold": 0.95, "size": 10
What Do These Thresholds Mean?
Yellow Threshold (0.8): Documents with 80% similarity trigger moderate risk flagging
Red Threshold (0.95): Documents with 95% similarity trigger high risk flagging
Size (10): System compares against the last 10 submitted documents
Are These Universal Settings?
No - these are average guidelines that work well for most use cases. Different industries and campaign types may benefit from adjusted thresholds:
High-Volume Campaigns: May need higher thresholds to reduce false positives
Premium Programs: May use lower thresholds for stricter duplicate detection
Seasonal Campaigns: May adjust comparison window size based on submission patterns
Device-Dependent Features
Why Are Velocity and History Features Disabled?
Technical Requirements: These features require device fingerprinting and tracking capabilities that are only available through Veryfi Lens SDK implementation.
Device Information Needed:
Unique device identifiers
App installation tracking
Submission timing patterns
User behavior analytics
What Are These Features?
High Velocity Detection: Identifies unusually high submission rates from specific devices
Critical Velocity Detection: Flags extremely high submission volumes exceeding normal thresholds
History Check: Analyzes past fraud patterns from the same device
Profile Fraud Detection: Identifies multiple user profiles operating from single devices
How to Enable Device-Dependent Features?
Implement Veryfi Lens SDK in your application
Configure device tracking parameters
Contact your account manager to enable velocity thresholds
Set appropriate limits based on your expected user behavior
Customization Options
How Can I Customize My Fraud Configuration?
Business Rules Engine: Create sophisticated conditional logic for fraud detection
Field Customization: Add or remove fields from handwriting and tampering detection
Threshold Adjustment: Modify similarity and AI detection sensitivity
Industry-Specific Settings: Configure rules based on your vertical requirements
CPG Campaign Customization:
IF campaign_value > $1000 AND handwritten_fields contains "total" THEN fraud_color = "red" IF vendor_category = "grocery_store" AND similarity_score > 0.7 THEN flag_for_review = true
Expense Management Customization:
IF expense_amount > $500 AND digital_tampering_detected = true THEN require_supervisor_approval = true IF vendor_category = "restaurant" AND handwritten_fields = ["total"] THEN fraud_color = "green"
When Should I Customize?
Immediate Customization Scenarios:
Industry-specific receipt patterns differ from defaults
False positive rates exceed acceptable levels
Fraud patterns unique to your use case emerge
Monitor First Approach:
Run with defaults for 2-4 weeks
Analyze fraud detection patterns and false positive rates
Adjust based on actual data and operational feedback
Implementation Best Practices
Getting Started with Default Configuration
Phase 1: Baseline Operation (Weeks 1-2)
Deploy with default settings
Monitor fraud detection rates and accuracy
Document false positives and patterns
Gather users feedback & operational impact
Phase 2: Initial Optimization (Weeks 3-4)
Adjust similarity thresholds based on observed patterns
Fine-tune field-level detection if needed
Configure any obvious industry-specific requirements
Implement basic Business Rules for edge cases
Phase 3: Advanced Customization (Month 2+)
Develop sophisticated Business Rules workflows
Enable device-dependent features if using Veryfi Lens
Create custom fraud handling workflows
Implement reporting and analytics for continuous improvement
Monitoring and Optimization
Key Metrics to Track:
Fraud Detection Rate: Percentage of submissions flagged
False Positive Rate: Legitimate submissions incorrectly flagged
Processing Impact: Effect on submission processing times
Manual Review Volume: Operational overhead from fraud detection
Optimization Triggers:
False positive rate exceeds 5-10%, depending on use case
Fraud detection rate is significantly below industry benchmarks
Manual review volume overwhelms operations team
Support and Guidance
Available Resources:
Account manager consultation for advanced configuration ->request with support@veryfi.com
Frequently Asked Questions
Can I change my configuration after going live?
Yes, fraud configuration is fully adjustable at any time through your account manager or Business Rules engine. Changes can be implemented gradually to avoid operational disruption.
How often should I review my fraud configuration?
Bi-weekly & Monthly reviews are recommended for active optimization. Quarterly reviews are sufficient for stable, mature configurations. Major campaign changes or fraud pattern shifts may require immediate adjustments.
What if I need features not in the default configuration?
Contact your Veryfi account manager to discuss:
Enabling device-dependent features
Beta access to new fraud detection capabilities
Custom integrations for enterprise requirements
Advanced Business Rules development support
Contact Information
Technical Support: support@veryfi.com
Account Management: Contact your assigned account manager