Entity resolution is the process of working out whether different records refer to the same real-world person or company. In AML, it makes screening and customer checks more accurate by linking related records and separating genuinely different ones. Data about a customer often arrives in many forms, and entity resolution ties those threads together while keeping different people apart.

Entity resolution works by comparing records and judging how likely they are to describe the same entity. It matches names, addresses, and other details, uses fuzzy matching to allow for spelling variations and typos, scores how strongly records point to the same entity, and links those that belong together. Because real data is messy, it weighs many clues rather than seeking an exact match.

Entity resolution matters because AML depends on knowing who you are dealing with. Without it, a firm might treat one risky customer as several harmless-looking ones, missing the full picture, or waste time treating separate people as the same. Screening a name is only useful if the firm can tell whether a hit really matches its customer, which accurate resolution makes possible.

Screening produces huge numbers of false positives, often over 90 percent of alerts by industry estimates, partly because systems cannot tell whether a name match is really the same person. Better entity resolution lets the system rule out coincidental name matches and confirm genuine ones, cutting the noise so analysts spend their time on real risk rather than clearing confusion.

Fuzzy matching is a technique that allows for variation between records, such as different spellings, typos, transliterations, and formats, rather than requiring an exact match. It is essential in entity resolution because the same person’s details rarely appear identically across records. Fuzzy matching helps recognize that two slightly different records are likely the same entity, improving accuracy over rigid exact-match rules.

Machine learning helps entity resolution by weighing many pieces of information at once, learning which combinations reliably indicate a match, and handling the variety of real data better than rigid rules. It can recognize that two records with different spellings are almost certainly the same person, or that a close name match is different, making resolution more accurate and less manual, though human judgment still matters.

The challenges include name variations such as different spellings and transliterations, poor data quality from incomplete or inconsistent records, different people who genuinely share a common name, and deliberate obfuscation by criminals varying their details to avoid being linked. That last point matters in AML, since bad actors actively try to defeat entity resolution, so it must be resilient to more than honest error.

Entity resolution is used in screening, to tell whether a sanctions or PEP hit really matches a customer; in customer due diligence, to build a single accurate view of each customer; in monitoring, to see a customer’s full activity across accounts; and in network analysis, to link related parties and reveal hidden connections. Its uses share the need for an accurate view of who is who.

Screening checks names and details against watchlists such as sanctions and PEP lists to find potential matches. Entity resolution is the underlying capability that determines whether records, including any matches, refer to the same real entity. Screening finds possible hits; entity resolution helps confirm whether a hit truly relates to the firm’s customer, making the screening results reliable rather than noisy.

Yes. By linking related records and parties, entity resolution can reveal connections that would otherwise stay hidden, such as the same person appearing under slightly different details across accounts, or related parties in a network. This helps a firm see the full picture of a customer or a group, which is valuable for spotting complex money laundering that relies on fragmentation.

Data quality is important because entity resolution can only be as good as the records it works with. Incomplete, outdated, or inconsistent data makes it harder to tell whether records match, leading to missed links or false ones. Improving data quality, alongside capable matching tools, is one of the most effective ways to make entity resolution, and the checks that rely on it, more accurate.

No. While names are a major part, entity resolution weighs many attributes, including addresses, dates of birth, identifiers, and other details, to decide whether records match. Relying on names alone would produce poor results, since different people share names and the same person’s name varies across records. Using multiple attributes together is what makes resolution reliable.