Sanctions screening is a legal requirement and a genuine protection. It is also, in practice, one of the largest sources of friction in cross-border payments. The problem is not screening itself but its calibration. When systems over-match, legitimate payments are held, investigated and sometimes refused. The cost of that friction is real, even when no rule has been broken.

Why over-blocking happens

  • Name matching on thin data. Common names, transliterations and partial matches produce alerts that are not matches.
  • Risk aversion. For an individual analyst, releasing a payment carries more personal risk than holding it. The rational local choice produces an irrational aggregate outcome.
  • Duplicated checks. Screening repeated at each hop multiplies alerts without adding assurance.
  • Poor data quality. Missing or malformed fields force manual review where automation should decide.

What over-blocking costs

Delayed payments damage commercial relationships, and for smaller operators a held payment can disrupt cash flow entirely. The operational cost is also significant: exception queues consume skilled staff, and each alert resolved by hand is time not spent on the alerts that matter. Excessive friction also has a broader effect — it is one of the pressures behind de-risking, as institutions withdraw from corridors whose compliance burden outweighs the business.

Where better tuning helps

Improving precision is a data and design problem more than a policy one. Cleaner reference data reduces false matches. Concentrating screening in a single, well-tuned function reduces duplication, since most alerts are repeated versions of the same question. And distinguishing a rejection from a held payment matters: an operation that exceeds a limit can become a human approval request rather than a failure, which preserves the control while avoiding an unnecessary block.

In short

  • The issue is calibration, not the existence of screening.
  • Over-blocking delays legitimate business and consumes skilled staff.
  • Duplicated screening multiplies alerts without adding assurance.
  • Better reference data and centralised screening improve precision.
  • Held-for-approval is often better than outright rejection.

This content is general information, not legal, tax or financial advice.

For how a single validation function fits into the stack, see /institutions/.