Definition
Combined economic cost of missed fraud and false-positive customer friction, expressed in basis points of transaction value, measuring whether detection thresholds sit at the economic optimum.
The Measurement Gap
Measurement Gap and Decision
What FTOS Measures: FTOS combines the economic cost of fraud losses from missed detections with the customer friction costs from false positives at the current threshold settings, expressed as a single optimization target in basis points of total transaction value.
Decision Supported:
Is the fraud threshold set at the economic optimum, or is it minimizing one cost (fraud losses or customer friction) while ignoring the other? Without FTOS, fraud teams optimize for detection rate (tighter thresholds) while CX teams push for fewer false positives (looser thresholds), and no one computes the combined economic impact of the current setting.
Why This Is Unmeasured:
Fraud management and customer experience operate as separate functions with separate KPIs. Fraud teams report fraud loss and detection rates. CX teams report customer complaints about false blocks. Nobody computes the single number that represents the total economic cost of the current threshold position. This means the threshold is set by organizational politics (whichever team has more influence) rather than by economic optimization. FTOS makes the trade-off explicit and computable: every threshold position has a total cost, and the optimal position minimizes the sum (TM Forum, "Fraud Management Best Practices: Threshold Optimization," 2024; Journal of Financial Crime, "Economic Optimization of Fraud Detection Thresholds in Telecommunications," 2023).
Formula
Plain Language: Add up the cost of all fraud that passed the detection threshold undetected. Add the cost of all false positives: blocked legitimate transactions plus subscriber churn attributed to false blocks multiplied by lifetime value. Normalize by total transaction value for comparability. The threshold is optimal when this combined cost is minimized.
Notation:
FTOS = (Fraud_loss_from_misses + Customer_friction_cost_from_FP) / Total_transaction_value × 10000
Variable Definitions:
- Fraud_loss_from_misses = confirmed fraud value that passed the detection threshold undetected during the measurement period (dollars)
- Customer_friction_cost_from_FP = (false_positive_blocks × average_revenue_per_blocked_transaction) + (FP_attributed_churn × subscriber_lifetime_value) (dollars)
- false_positive_blocks = legitimate transactions incorrectly blocked by the fraud detection system (count)
- average_revenue_per_blocked_transaction = average revenue lost per blocked legitimate transaction (dollars)
- FP_attributed_churn = subscribers who experienced a false positive block AND churned within 90 days, adjusted for baseline churn rate (count)
- subscriber_lifetime_value = average remaining lifetime value per subscriber (dollars)
- Total_transaction_value = total CDR or charging value in the measurement period (dollars)
Units:
Basis points of total transaction value. Lower is better. The multiplier of 10,000 converts the ratio to basis points for readability.
Sign Convention:
Lower is better. The optimal threshold setting minimizes FTOS.
Edge Cases:
- FP-attributed churn is hard to isolate precisely. Use a proxy: subscribers who experienced a false positive block AND churned within 90 days, minus the expected churn from the baseline rate for similar subscribers.
- Some false positives are caught by manual review before the customer experiences the block. Include only customer-experienced false positives.
- Minimum: requires 30 or more confirmed fraud events and 30 or more confirmed false positives per measurement period for meaningful optimization analysis.
Proposed formula, pending validation pilots.