Fuzzy Name Matching
A technique used in screening to identify potential matches between names or addresses even when there are minor variations, misspellings, or phonetic differences. This is critical for catching sanctions evaders who deliberately or incidentally spell a name differently from the version a designating government published.
Exact matching only catches the customer who filled in a form using the same transliteration the designating government chose. The gap comes from five sources: transliteration (Mohammed / Muhammad / Mohamad are all the same underlying Arabic name), name order and structure (patronymics, compound surnames, particles like "van" or "al-" that move across cultures), partial and informal forms (dropped middle names, initials, married names), data-entry errors, and deliberate obfuscation by someone actively trying to evade the exact list entry.
Screening engines close that gap with layered techniques rather than any single one: normalised comparison, token-level matching, phonetic encoding such as double metaphone, and edit-distance measures — each catching what the previous layer misses, and each adding some false positives in exchange. "We do fuzzy matching" is not a specification; the honest question is which techniques run against which fields, and at what measured cost in alert volume. No technique solves deliberate obfuscation through an entirely unlisted intermediary — that is what beneficial-ownership screening and transaction monitoring are for.