A sanctions screening program with a 95% false positive rate is not a compliance program. It is a workload generator. Every alert that resolves to “clear” consumes compliance capacity that should be directed at genuine risk. When false positive volume is high enough, teams start clearing alerts on instinct rather than evidence — which is precisely when real matches slip through.
Tuning match logic is not a one-time configuration exercise. It is a continuous calibration that requires data from your adjudication history, not just a vendor’s default confidence threshold.
Why default thresholds fail most programs
Most screening vendors ship with a conservative default match threshold — often 70% or 80% fuzzy string similarity — designed to minimize false negatives across all customers. The result is a catch-all that surfaces every phonetically similar name in your portfolio as a potential SDN match.
Your program is not “all customers.” It has a specific entity profile: certain nationalities, industries, transaction types, and geographies. A threshold calibrated to your actual population will produce fewer false positives without meaningfully increasing false negative risk — if the calibration is done rigorously.
Three calibration levers worth adjusting
- Match score threshold: raise cautiously, document the rationale and validate against your last 90 days of adjudications.
- Entity type weighting: individual vs. corporate matches have different false positive profiles — weight them separately.
- Jurisdiction exclusions: low-risk jurisdictions with common names may warrant a higher threshold or secondary confirmation step.