Ownership evidence a sanctions reviewer accepts
A reviewer asked who owns your counterparty, and the file came back. This page is about what actually closes that question: which document answers which layer of an ownership chain, why a date on the source matters more than the search that produced it, and the three places files usually fail.
It does not interpret any sanctions regime. Ownership rules belong to the authority that wrote them — OFAC's own answer on indirect ownership through layered structures is FAQ 401, and it is short. What follows is the file-preparation side: the evidence a reviewer can check, assembled so the person who has to decide can decide.
Not legal, financial, compliance, sanctions, investment, or insurance advice. Nothing here determines ownership, control, or sanctions status, and no page can: those determinations belong to your bank, counsel, or compliance function.
The question behind the question
A return notice rarely says what it needs. It says the ownership structure is unclear, or asks you to confirm the ultimate beneficial owner, or simply repeats a name and a question mark. Read literally, that looks like a request for a name. It almost never is.
What the reviewer has to satisfy is narrower and more mechanical: for each entity standing between your counterparty and the person who benefits, can someone outside your company verify who holds it, in what proportion, and on what date? A name supplied by the counterparty does not answer that. A record does.
The practical consequence is that the same file gets returned twice for what looks like the same question. The first answer supplies a name. The second supplies the name again, more emphatically. Neither adds a checkable record for the layer the reviewer could not see.
A statement is not a record
The distinction that decides most returned files: some documents assert a fact, and some evidence it. Both belong in a pack. Only one closes a question.
UBO declaration
A statement by the counterparty. It tells the reviewer what to verify. It does not evidence any layer beneath the signatory, and on its own it moves nothing.
Company registry extract
A record, and the strongest ordinary evidence for one layer. It carries an issuing authority and an issue date. It settles that layer only — the entity named as shareholder is now the next question.
Articles, charter, constitutive documents
Evidence of how control is exercised: voting rights, board appointment, veto rights, classes of share. This is what answers control where a percentage does not.
Shareholder agreement, trust deed, nominee arrangement
Often the document that explains why the registry and the economic reality disagree. Frequently the missing one.
Screenshot of a list search
Evidence of an act, not of a status: it shows a search was run. Without the list version and the date it was run against, it cannot be repeated by anyone else, which is what the reviewer needs.
End-user certificate
A statement about intended use, signed by the party with the incentive. It is required in many files and it is still a statement — it belongs beside the route and delivery records, not instead of them.
A pack of statements can be complete and still answer nothing. That is the usual shape of a file that comes back a second time.
Layer by layer, and stop at the first gap
An ownership chain is answered one layer at a time, and a chain is only as evidenced as its weakest link. Working from your counterparty outward, each layer needs three things: who holds it, in what proportion or under what control mechanism, and as of when.
Two failures are common enough to name. The first is the jump: a file evidences the counterparty and the ultimate owner, and asserts everything in between. The reviewer sees a gap in the middle and returns the whole pack. The second is the percentage that does not settle the question — two holders at 40 and 35 percent, with control living in a shareholder agreement nobody attached, or a holding company whose own shareholder is another company in a jurisdiction whose registry is not public.
When a registry genuinely cannot be obtained, that is a finding, not a dead end. A file that says plainly which layer cannot be evidenced, what was attempted, and what the counterparty was asked, is in a better position than one that papers the layer over with a declaration. Reviewers deal in known gaps. They cannot deal with gaps they have to discover themselves.
The date is the evidence
Sanctions lists change. A search result without the date it was run and the list it was run against cannot be reproduced, and a reviewer who cannot reproduce it has to run it again — which is the same as not having it.
Each authority publishes its list in a machine-readable form with a publication date attached, and those files, not a screenshot of a search box, are what makes a check repeatable:
- OFAC Sanctions List Search — the official search, and the SDN list as XML plus the consolidated non-SDN list
- EU Sanctions Map, over the EU consolidated financial sanctions list
- The UK Sanctions List, published by the FCDO
A name match against any of them is a possible string match and nothing more. It is not identity verification, not clearance, not an ownership finding: the entity behind a similar name is frequently unrelated, and transliteration from Cyrillic or Chinese produces near-matches routinely. What a match earns is a question, addressed to a named party, with the list version and date attached.
Where files usually fail
The middle of the chain
Both ends evidenced, the layers between them asserted. The fix is a registry extract or equivalent record per layer, and an explicit note where one cannot be obtained.
Control without a percentage
No holder reaches a threshold, and nothing in the pack shows who appoints the board or holds a veto. The answer is in the articles or the shareholder agreement, which is usually the document nobody sent.
Undated checks
Searches were run, and the pack cannot show against which list version or when. Repeating them costs an afternoon; not repeating them costs another round trip.
Questions
Is a UBO declaration enough?
Only for what it is: a statement by the counterparty telling the reviewer what to verify. It evidences no layer beneath the signatory. A file built on declarations alone tends to return with the same question asked more firmly.
Is a name match on a sanctions list a determination?
No. It is a possible string match — not identity verification, not clearance, not an ownership finding. Similar names are frequently unrelated entities, and transliteration produces near-matches routinely. The determination belongs to the bank, counsel, or compliance function that asked for it.
What if a registry is not public in that jurisdiction?
Say so in the file, name the layer, and record what was attempted and what the counterparty was asked. A stated gap is workable for a reviewer. A gap they have to find themselves is what sends the pack back.
Why does a list search need a date?
Because lists change, and a search nobody can reproduce is a search nobody can rely on. The list version and the date the check was run are what make it repeatable by the person deciding.
If a file has already come back
This page is the reasoning behind a service: preparing one returned trade file so the next human review has something to work with — a missing-evidence register, the questions to put to each party, and the structure to resubmit in. One fixed fee per file, agreed before work starts.