How Risk Thresholds Work in NGD

A risk threshold (also called a relevance threshold) is the minimum match score a profile must reach for that risk type to be surfaced to your team. It is how you translate the match score into a screening decision.

How thresholds filter results

Thresholds are set per risk type within each risk collection. When a profile is returned, the platform checks its match score against your configured threshold for each risk type. Only risk types where the score meets or exceeds your threshold will appear on the profile.

For example: a profile for "John Smith" has a match score of 70. The profile has both fraud and sanctions risk indicators. If your configuration is set to threshold 20 for sanctions and 80 for fraud, you will only see the sanctions risk on John Smith's profile. The fraud risk is filtered out because the match score of 70 did not reach the fraud threshold of 80.

This is by design: The same match score can pass the threshold for a lower-sensitivity risk type (e.g. sanctions at 20) while being filtered for a higher-sensitivity risk type (e.g. adverse media at 90). The threshold is your risk appetite translated into a number.

How thresholds and match scores work together

The match score and the threshold serve two different functions that compound each other. The match score ranks how likely a profile is to be the same entity as the one being screened. The threshold is your organisation's policy on how certain you need to be before acting on that risk type. Together, they replace the complex multi-parameter tuning that legacy screening systems required.

How to set the right threshold

Run a data test on your customer population to understand the expected hit rate at different threshold settings, then calibrate. See the companion article on recommended starting points for guidance on where to begin. Contact your Customer Success Manager to arrange a dedicated configuration session.