The subscription pack is the most misunderstood document in fund compliance. It is a contract. It records what the investor commits, what they represent and what they agree to. It is signed, filed, and frequently treated as though it were the due diligence.
It is not. A representation is something the investor asserts. Due diligence is something the firm establishes. The regulations ask the firm to identify and verify on the basis of documents or information from a reliable, independent source — and the investor is not an independent source about themselves.
That distinction is where most onboarding processes are weakest, and it gets more expensive the further the investor sits from being a natural person.
Five kinds of investor, five different problems
The individual. The easy case, and the one every process is designed around. Identity, address, tax residence, and — where risk warrants it — where the money came from. Documents verified against an independent source rather than accepted at face value.
The corporate subscriber. Now you have two jobs, not one. You must identify the company itself: its legal form, the law it is subject to, its constitution, its registered address, its directors and the senior persons responsible for its operations. And you must get through it to the natural persons who own or control it. The company is the customer; the humans behind it are the beneficial owners, and both sides have to be in the file.
The trust. A trust is not a company and does not behave like one. There is no shareholder register to read. What has to be established is the cast: the settlor who put the assets in, the trustees who control them, the beneficiaries entitled to them, and anyone else — a protector, an appointer — with real power over the arrangement. Discretionary trusts are harder still, because the class of beneficiaries may be defined rather than named.
The nominee. The name on the subscription is not the investor. Someone is holding for someone else, and the entire point of the arrangement is that the second name is not on the paperwork. The question is not whether you can identify the nominee — you can, easily, and it tells you almost nothing. The question is whether you have established who is behind it.
The feeder or fund-of-funds. You are onboarding an entity whose own investors you will never see. What matters here is not the underlying investor list, which you are not going to get, but whether the feeder itself is a regulated entity in a jurisdiction with equivalent standards, who manages it, who administers it, and what its own AML programme looks like. You are relying on someone else’s work, so the file needs to show why that reliance is reasonable.
Where the process actually breaks
Almost never at the first layer. Firms are generally competent at collecting a certificate of incorporation and a register extract. It breaks one or two layers up, in four recognisable ways.
The structure does not resolve. The subscriber is owned by a holding company, which is owned by another company in a jurisdiction with a thin registry, which is owned by a trust. Each step is legal. The chain simply runs out of public information before it reaches a person.
The name does not match itself. The director in the register, the signatory on the subscription and the person on the passport are three spellings of one name, or two people with the same name, or one person using a middle name in one place and not the other. Name matching is where most false positives and most false negatives are born.
Nobody re-checks. The file was complete at onboarding in 2023. The beneficial owner changed in 2024, was charged with something in 2025, and the fund found out from a journalist. Ongoing monitoring is an explicit requirement, not a courtesy, and periodic review is the thinnest part of most small firms’ programmes.
The negative result is over-claimed. The analyst searched, found nothing, and wrote “clear”. Two years later a supervisor asks what was searched, when, and on what spelling — and there is no answer, because “clear” recorded a conclusion instead of a method.
What the file has to be able to survive
Assume the investor becomes a problem three years from now, and that the person who onboarded them has left. The file has to answer, on its own:
- Who was identified, and what was verified against what independent source.
- How the ownership and control structure was understood, and where it stopped resolving — and why stopping there was reasonable.
- What was searched, on which spellings, on what date, and what came back.
- Where nothing was found, that nothing was found in those sources — not that the investor was clean.
- Who reviewed the material and made the decision.
None of that is exotic. All of it is hard to reconstruct after the fact, which is why it has to be captured as the work is done rather than written up afterwards.
The part that does not scale
Identity verification scales. There are competent vendors, the checks are cheap, and a small firm can run them at volume without difficulty.
The research does not scale. Establishing what a corporate subscriber actually is, reading filed accounts, finding the litigation, checking the trustee’s other appointments, working out whether the adverse media hit is the same person — that is analyst hours, and a two-person compliance function runs out of them long before it runs out of investors. It is also the part that a supervisor is most likely to test, because it is the part that requires judgement.
This is the specific problem EyesClear Investigations addresses: it screens a company or a person against public sources and returns a report in which every finding carries the quote, the source and the date it came from, so the research arrives already evidenced rather than needing to be reconstructed. Where nothing is identified, it says so in those terms. The methodology is published in full, including the limits — what a screening cannot establish is as important to a defensible file as what it can.
The judgement stays with your team. The tool assembles the evidence; a person reads it, weighs it and signs. That division is deliberate, and it is the same one we hold to across both of our tools.
Related
The regulatory frame for all of this is set out in AML requirements for UK asset managers. When the structure will not resolve to a natural person, that specific problem is covered in identifying beneficial owners through fund and trust structures. And when the question is not who the investor is but where the money came from, see source of wealth vs source of funds.