Fraud Ring Pre-Detection Exposure (FRPDE)

GMI-INS-009InsuranceIntegration-Required

Definition

Total dollar amount paid on fraud ring claims between the first ring claim and the point at which the network analysis model identified the ring, measuring the economic cost of detection latency in coordinated fraud schemes.

The Measurement Gap

Fraud detection metrics report what the model catches: flagged claims, confirmed fraud rate, SIU referral volume, recovered dollars. None of them measure the cost of what the model caught late. A coordinated fraud ring does not begin costing money when the network analysis model flags it. It begins costing money at the first staged claim, often months earlier, and between those two points every ring-connected claim is paid as legitimate business. Loss ratio absorbs those payments invisibly, blended into the honest claim population. Detection metrics actually improve when a large ring is finally identified, at the precise moment the model's latency has already cost the most. No legacy metric attributes pre-detection payments to detection speed, so the organization cannot see whether its ring detection is getting faster or slower, or what that latency is worth in dollars.

The Decision:

FRPDE tells SIU and analytics leadership whether investment in earlier ring detection (graph features, cross-carrier data sharing, referral integration) is economically justified. If pre-detection exposure per ring is rising, detection latency is worsening and the model refresh cadence or feature set needs attention. If FRPDE concentrates in specific ring types, that concentration defines where detection investment pays back first.

Formula

Plain Language: For each fraud ring identified during the measurement period, sum all claims paid to ring-connected entities before the ring was flagged by the detection model. This total is the pre-detection exposure. Report per ring and in aggregate.

Formula:

FRPDE = Sum of (Claims_paid_pre_detection_r) for all identified rings r

FRPDE_per_ring = Claims_paid_pre_detection_r for a single ring r

Where Claims_paid_pre_detection_r = total dollars paid on claims connected to ring r before the ring detection alert date.

Variables:

  • Claims_paid_pre_detection_r: total claim payments to entities connected to fraud ring r, paid before the ring detection date (dollars)
  • Ring detection date: the date the network analysis or SIU first flagged the connected structure as a potential ring
  • r: an identified fraud ring during the measurement period

Units: Dollars. Report total FRPDE, average FRPDE per ring, and FRPDE as a percentage of total detected fraud savings.

Data Requirements:

Data Element Source System Classification
Fraud ring identification records with detection dates SIU case management / fraud analytics platform NATIVE
Claims linked to identified ring entities Claims management system cross-referenced with SIU ring records NATIVE
Claim payment dates and amounts for ring-linked claims Claims management system NATIVE
Network analysis model alert timestamps Fraud analytics platform NATIVE

Edge Cases:

  • Ring boundaries are judgment calls. Define ring membership consistently using the SIU's entity-linking methodology and apply uniformly.
  • Some ring members may have legitimate claims mixed with fraudulent ones. Include only claims that the SIU ultimately attributed to the ring's fraudulent activity.
  • Rings detected through tips or law enforcement referrals rather than the AI model should be tracked separately to avoid crediting or penalizing the model for external detection.

Proposed formula, pending validation pilots.

At a Glance

Unit
Currency
Trend Direction
Lower
Reporting Frequency
Quarterly
Category
Fraud Detection Calibration and Collateral Damage
← False Positive Attrition Cost (FPAC)SIU Queue Currency Index (SQCI) →

Assess Your Organization

The AI-GPS Assessment evaluates your AI governance maturity and recommends KPIs like this one for your implementation roadmap.

Take the Free AI-GPS Assessment