7 min read

Fraud Shield

Automatic fraud and error detection for your invoices

Fraud Shield is Flux Capture's built-in protection system that automatically detects potential fraud, duplicate payments, and data errors before they become costly mistakes.

Overview

Every document processed by Flux Capture runs through multiple validation checks. When issues are detected, they're flagged for your review before approval.

Detection Categories

Duplicate Detection

Catch the same invoice being submitted multiple times.

Duplicate Invoice Number

Flags when an invoice number has already been processed for the same vendor.

What it catches:

  • Same invoice submitted twice
  • Invoice resubmissions after rejection
  • Vendor billing errors

Example alert:

Invoice #INV-2024-001 from ACME Corp already exists in the system.

Duplicate Payment

Detects when the same vendor, amount, and date combination exists.

What it catches:

  • Different invoice numbers for same transaction
  • Split invoice fraud
  • Vendor duplicate billing

Example alert:

Another invoice from ACME Corp for $5,000.00 dated Jan 15, 2024 already exists.

Amount Validation

Verify invoice amounts are mathematically correct and reasonable.

Line Items Total

Checks that individual line items sum to the subtotal.

What it catches:

  • Extraction errors
  • Missing line items
  • Addition errors on invoice

Example alert:

Line items total $4,800.00 but subtotal shows $5,000.00 (difference: $200.00)

Subtotal + Tax Validation

Ensures subtotal plus tax equals the total amount.

What it catches:

  • Tax calculation errors
  • Transposed numbers
  • Hidden charges

Example alert:

Subtotal $4,000 + Tax $500 = $4,500, but total shows $5,000

Positive Amounts

Flags unexpected negative amounts.

What it catches:

  • Credits submitted as invoices
  • Sign errors
  • Data entry mistakes

Example alert:

Total amount is negative (-$500.00). This may be a credit memo.

Round Amount Detection

Optionally flags suspiciously round numbers.

What it catches:

  • Estimated rather than actual amounts
  • Potential fabricated invoices
  • Unusual pricing patterns

Example alert:

Total amount $10,000.00 is an unusually round number.

ℹ️ Info: This check is off by default as many legitimate invoices have round amounts.

Amount Outlier

Flags amounts significantly higher than vendor historical average.

What it catches:

  • Abnormally large invoices
  • Decimal point errors ($5000 vs $50.00)
  • Potential fraud

Example alert:

Amount $50,000.00 is 5x higher than ACME Corp's average invoice ($10,000.00)

Date Validation

Ensure invoice dates make sense.

Future Date Detection

Flags invoices dated in the future.

What it catches:

  • Premature invoicing
  • Data entry errors
  • Time zone issues

Example alert:

Invoice date March 15, 2025 is in the future.

Due Date Sequence

Ensures due date comes after invoice date.

What it catches:

  • Reversed dates
  • Data entry errors
  • Extraction mistakes

Example alert:

Due date Jan 1, 2024 is before invoice date Jan 15, 2024.

Stale Date Detection

Flags invoices more than 90 days old.

What it catches:

  • Forgotten invoices
  • Late submissions
  • Potential duplicate risk

Example alert:

Invoice date October 1, 2023 is more than 90 days old.

Unusual Payment Terms

Detects terms that differ from vendor's typical pattern.

What it catches:

  • Changed terms without notice
  • Potential manipulation
  • Data entry errors

Example alert:

Payment terms (Net 10) differ significantly from ACME Corp's typical terms (Net 30).

Vendor Validation

Verify vendor information is correct and consistent.

Vendor Not Found

Flags when no matching vendor exists in NetSuite.

What it catches:

  • New vendors
  • Misspelled vendor names
  • Potential fraudulent vendors

Example alert:

Vendor "ACME Corporation Ltd." not found in system. Did you mean "ACME Corp"?

Low Vendor Confidence

Flags uncertain vendor matches.

What it catches:

  • Similar vendor names
  • Ambiguous matches
  • Need for verification

Example alert:

Vendor match confidence is 65%. Please verify correct vendor selected.

Invoice Format Change

Detects when invoice number format differs from historical pattern.

What it catches:

  • Changed invoice systems
  • Potential forged invoices
  • Vendor acquisitions

Example alert:

Invoice number format "FC-2024-001" differs from ACME Corp's typical pattern "INV-NNNN".

Required Field Validation

Ensure critical information is present.

Invoice Number Required

Flags documents without an invoice number.

What it catches:

  • Incomplete invoices
  • Extraction failures
  • Informal billing

Example alert:

No invoice number found. This field is required.

Total Amount Required

Flags documents without a total amount.

What it catches:

  • Incomplete invoices
  • Extraction failures
  • Quote documents

Example alert:

No total amount found. This field is required.

Understanding Anomaly Severity

Anomalies are classified by severity:

Severity Color Meaning Action
High Red Critical issue Must review before approval
Medium Orange Important warning Should verify
Low Yellow Minor concern Review if time permits

High Severity Examples

  • Duplicate invoice detected
  • Amount is negative
  • Required field missing

Medium Severity Examples

  • Amount outlier detected
  • Vendor not found
  • Due date before invoice date

Low Severity Examples

  • Low vendor confidence
  • Invoice format changed
  • Round amount warning

Reviewing Flagged Documents

When Fraud Shield flags a document:

In the Queue

Documents with issues show:

  • Warning icon next to status
  • Anomaly count indicator
  • "Needs Review" status

In Review

The review interface displays:

  • List of all detected anomalies
  • Severity for each issue
  • Suggested actions
  • One-click dismiss options

Taking Action

For each anomaly, you can:

  1. Investigate - Check the original document
  2. Correct - Fix the underlying issue
  3. Dismiss - Acknowledge and proceed
  4. Reject - Decline the document

Configuring Fraud Shield

Customize which checks run in Settings:

Enable/Disable Checks

Toggle each validation on or off based on your needs.

Adjust Thresholds

Some checks have adjustable thresholds:

  • Amount outlier multiplier (default: 3x)
  • Stale date days (default: 90)

Per-Vendor Overrides

For vendors with known exceptions:

  • Contact support for custom configuration
  • Or dismiss recurring false positives

Best Practices

Start Conservative

Enable all checks initially, then adjust based on false positive rate.

Review High Severity Always

Never skip high severity anomalies—they catch the most costly errors.

Investigate Patterns

If the same anomaly appears repeatedly:

  • There may be a legitimate pattern
  • Or a systematic issue to address

Document Exceptions

When dismissing anomalies:

  • Add notes explaining why
  • Creates audit trail for later

How Calculations Work

Amount Outlier Detection

The system calculates:

  1. Historical average for vendor (last 50 invoices)
  2. Standard deviation of amounts
  3. Flags if: amount > average × 3 OR amount > average + 2σ

Benford's Law Analysis

For advanced fraud detection:

  • Analyzes first digit distribution
  • Compares to expected mathematical distribution
  • Flags significant deviations

ℹ️ Info: Benford's Law is a mathematical principle about digit frequency. Fabricated numbers often violate this pattern.

Audit Trail

Fraud Shield maintains a complete audit trail:

What's Logged

  • Every anomaly detected
  • User actions (investigate, dismiss, correct)
  • Timestamps and user IDs

Accessing Audit Data

  • Available in document detail view
  • Exportable for compliance reporting
  • Retained per your data retention policy

False Positives

Sometimes legitimate invoices trigger alerts:

Common False Positives

Alert Legitimate Cause
Round amount Subscription or retainer fees
Amount outlier Annual payment or large order
Format change Vendor updated their system
Stale date Delayed vendor submission

Handling False Positives

  1. Review the underlying document
  2. Verify it's legitimate
  3. Dismiss with appropriate note
  4. Consider adjusting settings if recurring

Reporting

Anomaly Dashboard

View aggregate fraud detection stats:

  • Most common anomaly types
  • Trends over time
  • High-risk vendors

Export Reports

Export anomaly data for:

  • Compliance reporting
  • Audit documentation
  • Trend analysis

Next Steps