How sanctions screening actually works
A name search on Google is not sanctions screening. Here is exactly what happens between typing a name into Screen100 and getting a Clear, Possible match, or Match result — normalization, alias matching, transliteration, and the OFAC name-matching logic underneath the score.
Four stages, one deterministic score.
Every screen — free web search, API call, or MCP tool call — runs through the same four stages before you see a result.
Normalization
Before any comparison happens, both the query and every list entry are reduced to a canonical form. Accents and diacritics are stripped (Muñoz → Munoz), punctuation and extra whitespace are removed, casing is flattened, and common company suffixes — Ltd, LLC, GmbH, S.A., Co, Inc, Pte — are recognised and normalised so "Acme Trading Co." and "ACME TRADING COMPANY" collapse to the same base form. Word order is also handled: sanctions data frequently lists names in "Surname, Given name" order, or in a culturally different order (family name first for many Asian and some Slavic names), so the matcher checks both orderings rather than assuming Western given-name-first convention.
Alias & AKA expansion
OFAC and UN entries rarely have just one name. A single SDN entry can carry a primary name plus a dozen or more aliases: "a.k.a.", "f.k.a." (formerly known as), "n.k.a." (now known as), trade names, and low- and high-quality weak aliases the source itself flags as less reliable. Screen100 indexes every alias as its own searchable name tied back to the parent entity, so a query matches on any alias — not just the headline name — and the result always shows you which specific name (primary or alias) produced the hit.
Transliteration handling
Many designated individuals and entities originate in Arabic, Cyrillic, Chinese and other non-Latin scripts. The official lists publish Latin-script transliterations, but there is no single standard — the same Arabic name might appear as "Mohammed", "Muhammad" or "Mohamed" depending on which transliteration convention the source agency used. The matching engine accounts for common transliteration variants and phonetic equivalence classes (Al- / El- / Al prefixes, Yusuf / Youssef / Joseph-style variants, Kh/H and other consonant swaps) so a reasonable spelling variant still surfaces the right entry instead of silently missing it.
Fuzzy match scoring
After normalization, the query is scored against every primary name and alias using a blend of token-based and edit-distance algorithms tuned specifically for sanctions data — full names, not free text. The result is a single confidence score from 0 to 100 for every candidate, and only the highest-scoring hits per entity are surfaced. Every score is deterministic and repeatable: the same query against the same list snapshot always produces the same score, which matters when you need to defend a screening decision later.
What a match score means
Every candidate gets a 0–100 confidence score. The score falls into one of three bands, and Screen100 shows you the band, not just the number, so review decisions are consistent across your team.
88–100
A high-confidence match. The name, and often supporting attributes (nationality, date of birth, known aliases), align closely with a listed entry. Treat as a hit requiring review before proceeding — do not auto-clear.
72–87
A partial or ambiguous correspondence — common with widely shared names, partial name matches, or transliteration variants. Requires human review against supporting details (DOB, nationality, address, program) before you can clear or escalate it.
0–71
No meaningful correspondence was found against any list entry or alias at the time of the screen. Record the result and the list version searched — lists change weekly, so a clear result today isn't permanent.
The thresholds are fixed and disclosed rather than hidden inside a black-box model, so a screening decision made today can be explained and reproduced later — including in front of an auditor or regulator.
Three official lists, refreshed weekly
Screen100 does not compile its own watchlist. Every entry is sourced directly from the primary government publication, with the source list and program cited on every result.
OFAC SDN
US Treasury — Office of Foreign Assets Control
The Specially Designated Nationals and Blocked Persons list. Individuals, companies and vessels whose assets are blocked and with whom US persons are generally prohibited from dealing, published directly by OFAC.
OFAC Consolidated
US Treasury — Office of Foreign Assets Control
The non-SDN consolidated list — sectoral sanctions, the Foreign Sanctions Evaders list and other OFAC programs that carry restrictions short of a full asset block.
UN Security Council
United Nations Security Council
The UN Security Council Consolidated list — sanctions committee designations that UN member states are obligated to implement domestically, regardless of local jurisdiction.
Each source publishes updates on its own cadence — OFAC updates the SDN list as designations happen, sometimes multiple times a week; the UN Security Council updates on committee decisions. Screen100 pulls all three on a weekly refresh cycle at minimum, and every entity page shows the list version it was last updated from.
How to review a possible match
A Possible match is a prompt to look closer, not a verdict. Here is the review sequence a compliance analyst should run through before clearing or escalating one.
Read the matched name type
Was the hit against the primary name or against an alias? A match on a weak or low-quality alias carries less weight than a match on the primary designated name.
Compare supporting attributes
Sanctions entries often include date of birth, place of birth, nationality, passport or national ID numbers, and known addresses. If your subject's details contradict the listed entry's attributes (wrong birth year, different country), that materially lowers the likelihood of a true match.
Check the program and source
Every hit cites its source list (OFAC SDN, OFAC Consolidated, or UN Security Council) and, for OFAC hits, the sanctions program(s) it falls under (e.g. UKRAINE-EO13661, NPWMD, SDGT). This tells you the legal basis for the designation and which counterparty relationships it affects.
Document the decision
Whatever you conclude — true match, false positive, or inconclusive pending more information — record it with a timestamp and the list version searched. On Pro, this is generated automatically as a PDF audit certificate for every screen.
Cutting down false positives
Common names — Mohammed Ali, John Smith, Ahmed Hassan — will always generate some possible matches against a list of tens of thousands of entities. These five habits cut noise without cutting recall.
Use full names, not initials
"J. Smith" will fuzzy-match against far more entries than "John Robert Smith". Supplying a full legal name is the single biggest lever for reducing noise.
Add a date of birth or country when you have one
Screen100's API accepts optional DOB and nationality fields. When supplied, they're used to re-rank candidate matches — a name match against an entry with a contradicting DOB is demoted rather than surfaced as a top hit.
Distinguish individuals from entities
Set the subject type (individual vs organisation) on every screen. It's a cheap filter that eliminates an entire class of false positives — a person's name colliding with an unrelated company name, or vice versa.
Don't over-trust a single Clear result
A clear result reflects the list contents at the moment you searched. For any counterparty relationship that persists beyond a single transaction, a one-off check decays in value the moment the lists update — which is what ongoing monitoring is for.
Review, don't auto-reject, Possible matches
Treating every possible match as a hit generates alert fatigue and, eventually, a team that stops reading the alerts. Treating every possible match as noise risks missing a true positive. The 72–88 band exists specifically to route these to a human reviewer with the supporting evidence in front of them.
See the matching engine on a real name.
Free, no account required. Screen a name now and see the score, the band, and the cited source entry.