Sanctions and PEP screening has a paradox: the safer you make it, the more useless it gets. Loosen the matching and you miss real hits; tighten it and analysts drown in false positives. The answer isn’t a stricter switch — it’s smarter matching and a workflow built for review.
Why exact matching fails
Names don’t arrive clean. Transliterations differ, aliases abound, dates of birth are partial, and the same person appears a dozen ways across lists. Exact string matching misses the variations that matter and flags the ones that don’t.
Fuzzy matching, with the right signals
KeplerAI’s matching engine scores similarity across several signals at once:
- Name similarity with alias and transliteration awareness.
- Date of birth, including partial and fuzzy matches.
- Country and nationality to disambiguate common names.
Each candidate gets a match score, so you can triage by confidence instead of treating every hit equally.
Watchlists you control
Screening is only as current as its lists. KeplerAI ships an embedded seed and ingests OpenSanctions open data, with room for premium feeds. Lists live where you need them, and screening can run near your data.
The review queue is the product
Matching produces candidates; people make decisions. A first-class review queue lets analysts mark each hit true positive, false positive or escalate, with the reasoning captured for audit. Over time, those decisions are the record that proves your program works.
Screen at onboarding and continuously
New customers should be screened at onboarding; existing ones re-screened as lists change. Both belong in the same engine.
Good screening isn’t about catching everything and everyone — it’s about surfacing the right hits and clearing the rest fast. See the screening solution.