Every due diligence procedure contains a sentence to the effect that the firm will identify the beneficial owner. It is a short sentence describing a task that, for a fund investor, is frequently the hardest thing in the file — and occasionally impossible in the form the procedure imagines.
The threshold, and the change coming to it
In the UK, a beneficial owner of a body corporate is an individual who ultimately owns or controls, directly or indirectly, more than 25% of the shares or voting rights, or who otherwise exercises ultimate control over its management. For a partnership other than an LLP, the equivalent test is entitlement to or control of more than 25% of the capital, profits or voting rights.
In the EU, that is about to read differently. Under the AML Regulation — Regulation (EU) 2024/1624, which entered into force in July 2024 and applies from 10 July 2027 — an ownership interest means direct or indirect ownership of 25% or more of the shares, voting rights or other ownership interest, including rights to a share of profits or the liquidation balance.
The distance between “more than 25%” and “25% or more” is one shareholder holding exactly a quarter. That is not a hypothetical position — a four-way split is a common way to hold a vehicle, and it is precisely the arrangement that sits outside the UK test and inside the EU one. A firm operating either side of that line, or both, needs to know which test it is applying.
The EU threshold may go lower in specific sectors. The Commission may set a reduced figure by delegated act following an assessment due by 10 July 2029, capped at 15%. It is not in force, it is not sector-wide, and it should not be planned around yet — but it is the direction.
Neither threshold is the whole test. Control by other means — a shareholders’ agreement, a golden share, a veto, the right to appoint the board, or simple practical dominance — makes someone a beneficial owner regardless of percentage, and it is the limb most often missed, because it does not show up in a shareholding table.
Why the chain stops
A corporate subscriber is rarely one company. It is a company owned by a holding company owned by another vehicle, and each hop loses information.
Registries differ enormously. Some jurisdictions publish shareholders, directors, accounts and charges. Others publish a name and a date. The chain does not stop because someone is hiding something; it stops because the next jurisdiction does not require the filing.
Registers are self-declared. Beneficial ownership registers, including the UK’s, are populated by the entities themselves and are not verified as a matter of course. They are a starting point and a cross-check, and treating a register extract as verification is a common and identifiable weakness — UK firms also carry a duty to report discrepancies between what they find and what the register says, which only makes sense on the premise that registers are wrong often enough to matter.
Nominees and bearer arrangements are designed to stop it. The registered holder is the answer to a different question than the one you are asking.
Layers may be entirely legitimate. Tax treaty access, investor confidentiality, ring-fencing, regulatory requirements in a fund’s home jurisdiction. Complexity is a risk indicator, not a finding. The question is whether there is a commercial explanation that fits, and whether anything corroborates it.
Trusts do not have shareholders
A trust cannot be resolved with the corporate method, because the thing the method looks for does not exist. What has to be identified instead is the set of roles:
- The settlor, who put the assets in — the question of whose wealth this actually is.
- The trustees, who hold and control the assets. Often a professional corporate trustee, which starts a second identification exercise of its own.
- The beneficiaries, who are entitled to benefit. Where a discretionary trust defines a class rather than naming individuals, the class itself has to be described.
- Anyone else with real power — a protector or appointer who can remove trustees or veto distributions holds control that no ownership percentage records.
A file that names a corporate trustee and stops has identified an administrator, not the people whose money and interests are actually involved.
When it genuinely will not resolve
Sometimes there is no natural person to find at the end of the chain. The regulations anticipate this: where no beneficial owner can be identified after all reasonable measures have been exhausted, the senior managing official may be treated as the beneficial owner.
That provision is a fallback and is treated as one. Using it well means the file shows the measures that were actually taken, why each line of enquiry ended where it did, and that the conclusion was reached after the work rather than instead of it. Using it badly — reaching for it at the first opaque layer — converts a difficult case into a documented failure, because the file now records that the firm stopped early.
The same applies to the risk assessment. A structure that cannot be resolved is a structure that is riskier than one that can, and the file should say so and respond accordingly, rather than closing the point neutrally.
What the work actually consists of
Reading a corporate chain is not a lookup. It is registry filings in several jurisdictions, accounts that name a parent the register does not, director histories that connect two entities nobody said were connected, litigation and insolvency records, regulatory registers, sanctions and PEP lists, and adverse media — followed by the reconciliation problem, where the same person appears under three spellings and two dates of birth and you have to decide whether it is one individual.
That reconciliation is where errors go in both directions. Miss the match and a real finding is lost; force it and an unrelated person is attached to your investor’s file, which carries its own consequences.
This is the problem EyesClear Investigations was built around. It screens a company or a person against public sources and returns every finding with the verbatim quote, the source and the date it came from, so the chain you were able to establish is evidenced rather than asserted — and where the record runs out, the report says what was searched and that nothing further was identified, which is the honest form of that answer. The methodology is published in full, including what a screening cannot establish: it cannot prove an ownership chain ends where the public record ends, and it does not claim to.
Your analyst decides what the structure means and signs the conclusion. The tool does the assembling. That division — the AI as maker, the analyst as checker — is the same one we hold across both of our tools, and it is the reason a finding can be put in front of an examiner.
Related
The regulatory frame is in AML requirements for UK asset managers, the process this fits inside is in investor onboarding due diligence for fund managers, and once you know who the owner is, the next question is usually source of wealth vs source of funds.