Finds the name, however it’s typed.
- Word order and punctuation
- “Victor Grin” finds “GRIN, Victor”. Surname-first, natural order and list order are all the same name.
- Accents and case
- A name typed without accents, in lower case, still matches the accented list entry.
- Typos
- A single slip, like two neighbouring letters swapped, doesn’t hide a listed person.
- Transliteration
- Common differences in how names are written in Latin script, such as Iuri and Yury, are matched.
- Aliases, in other scripts too
- Every listed alias is screened, including aliases written in Cyrillic or Arabic.
- Dates of birth rule out namesakes
- A PEP whose listed date of birth is more than a year from your customer’s is ruled out, so ordinary people who share a name aren’t flagged.
In practice
Typed one way, listed another. Matched, with the reason shown.
typed
listed Yury BELOV
transliteration
typed
listed Anna SORENSEN
word order
typed
listed Юрий Белов
Cyrillic alias
Benchmark
99% of listed people found. 100% with a date of birth.
| How the name was written | Found |
|---|---|
| As the list writes it | 100% |
| Natural order | 100% |
| No accents, lower case | 100% |
| Surname first | 100% |
| Middle names dropped | 99% |
| One typo | 100% |
| Transliteration difference | 93% |
| Listed alias | 100% |
| Alias in another script | 89% |
| All variants | 99% |
Ordinary names flagged for review
| Name only | 12% |
| With a date of birth | 6.5% |
Most are genuine namesakes of PEPs, such as an ordinary David Taylor sharing a name with an MP. The PEP flags that remain with a date of birth are office-holders with no date of birth on record, which only a person can rule out.
Known limits
- Names in Arabic, Persian or Chinese script are matched only through the lists’ own Latin aliases.
- Some transliterations of Arabic and Persian names are missed when screened without a date of birth.
- No list or matching method is complete.
Cite this data
FirmScreen, “FirmScreen name matching benchmark”, firmscreen.io/solutions/fuzzy-name-matching (updated 9 Oct 2026). Free to reuse with credit under CC BY 4.0.
No. Matching is deterministic fuzzy matching. The same input always produces the same result, which makes it possible to explain and reproduce any decision.
Most false alarms are people who share a name with someone listed. A date of birth separates them: in our benchmark, the share of ordinary names flagged fell from 12% to 6.5% when a date of birth was given, and every listed person was found.
300 real individuals from the OFAC, UK, EU and UN sanctions lists and PEP sources, each written nine realistic ways, and 400 combinations of common UK names, screened against the full live lists using the same pre-filter as production screening. A test fails the build if any figure drops below the floor set from these results.