What is Fuzzy Matching and How Does It Work in Identity Checks?

**Entrust** formerly known as Onfido , Zinc's Identity, Financial, Politically Exposed Persons, Adverse Media, Sanctions check provider utilises **fuzzy matching** on the name and address fields during record searches. This is because names and addresses can be provided in many different ways, and errors in data submission or collection can often occur. Fuzzy matching allows a more accurate confirmation of a person's identity during a record check by capturing data variations. For example, accepted data variations could include: - A candidate named "**Catherine Stanley**" enters their name as "**Catherine Stamley**" or "**Kate Stanley**". - An address is entered as "**1/26 Rowland Close**" or "**Rowland Close, 26a**". Fuzzy comparison allows for greater flexibility during comparison, catering for potential discrepancies (for example, when an applicant uses their middle or spouse name, or there has been an extraction error). Exact comparison, by contrast, requires a precise match for all fields and could increase false rejection rates. Entrust therefore recommends using a fuzzy comparison configuration. ## What is fuzzy matching Fuzzy matching is a technique used to compare names and other identity data, even when there are minor differences due to typos, misspellings, or formatting inconsistencies. This helps ensure legitimate users aren’t blocked by small errors, while still detecting potential fraud or watchlist matches. ## How fuzzy matching works - **Customised sensitivity for Zinc:** The algorithm allows for a certain degree of difference between names. By default for Zinc, it permits a single character change (insertion, omission, or replacement) within a word, which helps reduce false positives from minor typos. We also treat first names and middle names as interchangeable to help handle multi-word names. - **Minimum word length:** Fuzziness is only applied to words above a certain length (default for Zinc is 13 characters). Shorter words must match exactly, as even a single character change in a short name can lead to incorrect matches. For example, '**Leederheimer**' and '**Lexderheimer**' may be considered a match, but '**Lee**' and '**Lex**' would not. - **Normalisation:** The system ignores differences in case, punctuation, and extra spaces, and can handle some international characters and transliterations. At Zinc, we use all the **default** Entrust **settings** however we do treat **First Names and Middle Names as interchangeable to help handle multi word names.**