The customer you cannot meet
Onboarding a person is hard. Onboarding a business is harder, and for a reason worth stating plainly: a company is a legal fiction. There is nobody to look at. What exists is a registration, some documents, and a set of humans who say they speak for it.
That produces three entirely separate questions, which are collapsed into one in most onboarding processes and should not be:
- Does this business exist, and is it what it claims to be?
- Is the person in front of me permitted to act for it?
- Who ultimately owns and controls it?
Each can be answered yes while another is no. A real, properly registered company can be represented by somebody with no authority. A company with impeccable paperwork can be owned through four layers ending in a jurisdiction that publishes nothing. A business that genuinely trades can have an owner who is sanctioned.
The standard treats them as a set. The FATF's Recommendation 10 requires identifying and verifying the customer, identifying the beneficial owner and taking reasonable measures to verify that identity, understanding the ownership and control structure, understanding the purpose and intended nature of the relationship, and conducting ongoing due diligence. For a business customer, "understanding the ownership and control structure" is a specific piece of work, not a restatement of the first measure.
One framing to carry through the rest of the course. You are not trying to produce a diagram. You are trying to be able to answer a question later: if this company's payments turn out to be funding something criminal, will the file show that you knew who was behind it and what it said it was going to do? A beautiful structure chart that nobody interrogated fails that test. Three sentences that explain the chain, with evidence for each, passes it.
It is worth being clear about why this is harder than it sounds, because the difficulty is structural rather than a matter of effort. Everything you learn about a business customer arrives through the customer. The documents are supplied by them, the structure is described by them, and even the registry entry usually records what they filed. With a personal customer you can at least compare a face to a document; with a business you are mostly assessing the internal consistency of a story told by the person who benefits from your believing it.
Which is why the two levers that matter most are independent corroboration and writing down what you were told. Corroboration tests the story against something the customer did not author. Writing it down means that when the story changes — and in the cases that matter it does change — you can see that it changed.
Existence is not the same as registration
The first check is usually a registry lookup, and the commonest mistake in business onboarding is treating a successful lookup as the answer.
A registration proves that somebody filed a form. In most jurisdictions, incorporating a company requires a name, an address, at least one officer and a fee. None of that is evidence that the business trades, has premises, employs anybody, or does the thing its application says it does. The register confirms the legal person exists. Whether the business exists is a different question.
What separates a registered entity from a trading one:
- Age and filing history. A company incorporated three weeks ago, applying for a facility sized for an established firm, is not necessarily fraudulent — but the mismatch is the thing to ask about, and the answer should be on the file.
- Independent evidence of activity. A website that predates the application, a verifiable telephone number, a VAT or tax registration, premises, named staff, customers who can be confirmed, an industry licence where the sector requires one.
- Consistency across sources. The address on the application, the address on the register, the address in the bank mandate and the address where goods are delivered should be explicable as a set. They frequently are not, and the explanation matters more than the discrepancy.
Two specific patterns to recognise. A shelf company is incorporated and left dormant so that it can later be presented as having history; the giveaway is a long registered life with no filings of substance and a recent change of everything. A repurposed company is a real, previously trading business taken over for its history and reputation; the giveaway is an abrupt change of officers, ownership, address and sector at once.
None of this means refusing the customer. It means the file records what the business is, on what evidence, so that later activity can be measured against something.
One more distinction worth holding, because it decides how much weight to put on the registry at all: some jurisdictions verify at the point of incorporation and most do not. Where a register is a filing system, the sequence is that a form is submitted, accepted and published, and the publication then looks like confirmation. Nothing in that sequence checked anything. Knowing which kind of register you are reading is part of knowing what your own check is worth, and it is the single most useful fact to establish about any country you onboard from.
Authority is a separate question from identity
You have verified that the person in front of you is Amina Patel. That tells you nothing about whether Amina Patel can open an account for this company, and the two checks are routinely merged into one step with one outcome.
Identity asks: is this person who they claim to be? Authority asks: is this person permitted to bind this entity, and to what extent? The evidence is different, and so is the failure mode. A failure of identity verification lets in an impostor; a failure of authority verification lets in a real person doing something they are not entitled to do — a junior employee opening a facility, a departed director still operating a mandate, one shareholder acting against the others.
Where authority actually comes from:
- The constitution of the entity — its articles or equivalent, which say who may do what.
- A board resolution or equivalent decision, naming the people authorised and the limits on them.
- A mandate the entity gives you, listing authorised signatories and what each may do alone or jointly.
- A power of attorney, where somebody acts for the entity without holding an office in it.
Three practical rules. Record the limits, not just the names — "authorised" is not a useful record if the resolution said two signatures above a threshold. Date everything and tie it to the officer record, because an authority granted by a board that has since changed may no longer stand. And have a defined process for removing authority, since the person who leaves is a larger risk than the person who joins: a live mandate held by a former employee is a standing hole, and it is one of the most common findings in a real review of business accounts.
Two authority patterns deserve particular attention because they are where real losses occur. The first is the sole director who is also the sole signatory and the sole beneficial owner. Nothing is wrong with it, and it is extremely common in small businesses, but it means no internal control exists at all — every protection has to come from outside the company, which raises what your own monitoring is worth. The second is authority that outruns the business reason for it: a mandate that permits international transfers for a company whose stated model is entirely domestic. That is not a fraud by itself. It is a capability nobody needed, and in a compromise it is the capability that gets used.
Who owns it, and what the standard actually asks
Beneficial ownership is where business onboarding becomes genuinely difficult, and where the vocabulary matters.
The FATF's defined term is beneficial owner — the natural person who ultimately owns or controls the customer. The standard is about reaching a person, not a percentage, and not an intermediate company.
Recommendation 24 sets what countries must provide for legal persons. They should assess the risks of misuse, and ensure there is "adequate, accurate and up-to-date information on the beneficial ownership and control of legal persons that can be obtained or accessed rapidly and efficiently by competent authorities, through either a register of beneficial ownership or an alternative mechanism". It also says countries "should not permit legal persons to issue new bearer shares or bearer share warrants", should prevent misuse of existing ones, and should ensure nominee shareholders and directors are not misused. And it asks countries to consider facilitating access to that information by financial institutions carrying out their own Recommendation 10 duties.
Note the two things that follow for you. There may be a beneficial ownership register where you operate, or there may be an "alternative mechanism" — so do not assume a register exists, and do not assume one that exists is open to you rather than only to authorities.
The percentage is a national choice, not a FATF rule. The FATF standards set no percentage threshold at all. Countries pick one, and they differ: the United Kingdom's register of people with significant control uses more than 25%, while Kenya's regulations use at least 10%. The same shareholding is therefore reportable in one country and not in another — which is why a group operating in both cannot run one global threshold and call it compliance.
And the threshold is a floor for looking, never a ceiling on thinking. Somebody holding 12% who appoints most of the board controls the company. Control through other means is explicitly part of the concept, and a process that stops at a percentage will miss precisely the structures built to be missed.
Reading an ownership chain, and knowing where it stops
A simple structure is a person holding shares. Real structures are chains, and reading one is a skill.
Work upwards and multiply. If your customer is owned 60% by Company B, and Company B is owned 50% by a person, that person's indirect holding is 30%. Do this for every branch, and remember that a person may appear on several branches — small holdings that are individually below a threshold can add up to a controlling one, and a process that evaluates each branch separately will miss that entirely.
Then look for the things that break the arithmetic:
- Control without ownership. Rights to appoint or remove directors, veto rights, golden shares, shareholder agreements. A person with 5% and the right to appoint the board is a beneficial owner and the percentages will never show it.
- Circular and cross-holdings, where A owns B, B owns C and C owns A. The arithmetic does not terminate and the structure has to be described rather than computed.
- Layers across jurisdictions, chosen so that at least one link publishes nothing.
- Entities with no natural owner, such as a foundation, a cooperative or a widely held listed company. These are not failures to resolve; they are different answers, and the file should say which.
Know what to do when the chain stops being readable. At that point you stop computing and start asking: a written ownership declaration from the customer, signed, naming the beneficial owners and the basis of their control. That is a statement by the customer, not verification — but it converts an unanswered question into a documented assertion you can test against behaviour, and it moves the consequence of a lie onto the person who told it.
A note on effort, because this work is open-ended and somebody has to decide where to stop. The standard asks for reasonable measures to verify the beneficial owner's identity, and reasonable is a function of risk — a low-risk domestic customer with a two-line structure does not warrant the work a four-layer cross-border structure does. What is not acceptable is letting the effort be decided by how busy the team was that week. Write the expected depth down by risk band, so that a thin file on a high-risk structure is visibly a departure rather than an accident.
If after all of that you cannot identify the beneficial owner and you are not satisfied, the standard's answer to incomplete customer due diligence (CDD, which much of the industry calls KYC) applies: do not open the relationship, and consider whether a suspicious transaction report (STR) is warranted — the same filing the United States and the United Kingdom call a suspicious activity report (SAR). Proceeding with an unresolved question recorded as resolved is the single worst outcome available.
Nominees, bearer instruments and trusts in the chain
Three devices exist largely to make the last lesson's work fail, and the standard addresses each.
Nominees. A nominee shareholder or director holds the position for somebody else. The arrangement is lawful and ordinary in many markets, and it is also the simplest way to put a different name on a register. Recommendation 24 requires countries to take effective measures to ensure nominee shareholders and directors are not misused. Operationally, the questions are: is this person a nominee, for whom, and on what terms — and the answer belongs on the file, because a nominee who is not identified as one makes the register actively misleading rather than merely incomplete.
Bearer shares and bearer share warrants. Ownership travels with the physical instrument, so the holder is whoever has the certificate and there is no register to consult. Recommendation 24 says countries should not permit legal persons to issue new ones and should take measures to prevent the misuse of existing ones. If an entity in your chain has them, the ownership question cannot be answered from documents at all, and that fact should drive the decision rather than be footnoted.
Trusts and similar arrangements. Here the relevant standard is Recommendation 25, which asks countries to ensure there is "adequate, accurate and up-to-date information on express trusts and other similar legal arrangements including information on the settlor(s), trustee(s) and beneficiary(ies)", accessible to competent authorities "efficiently and in a timely manner".
That list is the question set. When a trust appears in a chain, you are not looking for a shareholder; you are looking for who put the assets in, who controls them now, and who benefits — plus anyone else with effective control, such as a protector. A trust is not an obstacle to be noted and passed over. It is a different shape of answer, and the file should contain that answer in those terms.
When the registry cannot be relied on
Registers differ enormously, and a course that assumes a good one is useless in most of the world.
Three independent properties, which people conflate:
- Coverage — which entities and which information it holds at all.
- Verification — whether anybody checked what was filed. Many registers are filing systems: they record what was submitted, accurately, without examining whether it is true.
- Currency — how quickly it reflects change. A register updated on an annual return is a year out of date for most of the year.
The FATF's wording is the useful benchmark here, because it asks for information that is "adequate, accurate and up-to-date". Those are three separate tests, and a register can pass one while failing the others.
So treat a registry extract as a claim with a known provenance, not as proof. What to do with a weak one:
- Triangulate. Agreement between two independent sources — a registry filing and a tax registration, an audited account and a bank statement — is worth more than either alone.
- Prefer documents with a cost of lying attached. A filing that carries a penalty for inaccuracy, or that a third party relied on, is better evidence than a self-declaration.
- Ask the customer to explain inconsistencies, and record the explanation. An explanation that turns out to be false is far more useful later than an unexplained discrepancy.
- Write down what you could not verify. A file that distinguishes "verified against the register" from "stated by the customer and not independently confirmed" is honest, reviewable, and the basis for setting the right level of ongoing monitoring.
That last point is the whole discipline in one line: a weak registry is not a reason to lower the standard of the file. It is a reason for the file to be explicit about which parts rest on evidence and which rest on assertion.
Verifying the activity, not just the entity
You can finish all of the above and still know nothing about what the business is going to do. Recommendation 10 requires understanding the purpose and intended nature of the relationship, and for a business customer that is the part most often reduced to a dropdown.
Ask for the business model in terms that can later be measured: what it sells, to whom, in which countries, how it gets paid, in what sizes, how often, and what a normal month looks like. Record the answer in numbers. That record is what makes monitoring possible, because every alert afterwards is a comparison against it, and a stated expectation of "general trading" cannot be compared with anything.
Mismatches worth pausing on at onboarding:
- Sector against payment pattern. A wholesaler whose receipts are hundreds of small consumer-sized credits is not necessarily wrong, but it needs an explanation.
- Scale against substance. Volumes that imply a workforce, premises and suppliers the business does not appear to have.
- Geography against business reason. Counterparties in countries with no evident link to what the business says it does.
- Immediate full-rate use. A new facility used to its limit from the first day, with no ramp, is a pattern worth a question.
The last point to make here is organisational. The understanding you record at onboarding is the baseline the monitoring system compares against, and in many institutions it is captured by a relationship team and never reaches the people who tune the rules. Where that happens, monitoring is comparing activity against an industry average rather than against what this customer actually said it would do, and the single cheapest improvement available is to connect those two records.
Keeping it current
Ownership changes quietly. Nobody writes to tell you that a 30% shareholder sold to somebody sanctioned, and the register may not show it for a year. A business file is accurate on the day it is made and decays from then on.
Trigger events are worth more than calendars. A periodic review finds a change on average half a review cycle late; a trigger finds it when it happens. Triggers worth wiring in: a change of officers or address in the registry, a change in the mandate or signatories, a significant change in payment behaviour, a new country appearing in the flows, adverse media, a sanctions list change touching any name in the chain, and the customer's own notification.
Then review by risk, not uniformly. A high-risk structure with three layers across two opaque jurisdictions does not get the same cycle as a sole trader. And "high risk" should mean the file says why, in a sentence somebody can act on, rather than a score with no explanation.
Re-screen the whole chain, not just the customer. Sanctions and politically exposed person screening applied only to the account holder misses the beneficial owner who was the reason to look. The chain you mapped is a list of names, and that list belongs in screening.
On records, the international floor is short and specific. The FATF's Recommendation 11 requires records of transactions, records obtained through customer due diligence, account files and business correspondence — including the results of any analysis undertaken — to be kept for at least five years after the relationship ends. Your ownership analysis, your unresolved questions and the customer's declaration are all results of an analysis. National law often requires longer, and where it does, the longer period governs.
The test of a business file, in the end, is the same as the test of any compliance record: a year from now, when somebody asks why this company was taken on and what it said it would do, can the file answer without anybody remembering?
Thirty questions
Answer all thirty questions, then check your answers. You need 27 out of 30 to pass and receive a certificate. If you score less, you will see which answers were right and wrong, and then go through the course again before you retake the check. Your answers, progress and times are kept only in this browser.
Your answers
Your certificate of completion
Enter your name as you want it to appear, then save the certificate as a PDF. In the print window, choose Save as PDF. A certificate is issued once per completion of the course.
Rate this course
How would you rate this course? Your rating is anonymous — just a number, with no name and no account.
Bring this to your team
Your colleagues can take this course too — it is free, needs no sign-up, and nothing they enter leaves their browser. If your institution wants the same ground covered as structured training for a team, with an instructor and your own cases, we run courses in a classroom or online for groups of ten or more.
Talk to us about training your team → See the paid courses →
Saolix does not record who takes this course, so it cannot verify these certificates. The certificate confirms completion of a free self-paced course and is not an accredited qualification.
Standards and frameworks covered
The published frameworks this course teaches toward. Saolix is not affiliated with, endorsed by or accredited by any of these bodies; naming a standard is a citation, not a claim about its author.
Sources
The official documents this course relies on. Laws and guidance change, so check the current version.
- The FATF Recommendations (Recommendations 10, 11, 24 and 25, updated October 2025) · Financial Action Task Force
- Guidance on Beneficial Ownership of Legal Persons (March 2023) · Financial Action Task Force
- Guidance on Beneficial Ownership and Transparency of Legal Arrangements (March 2024) · Financial Action Task Force
- Summary guidance for companies, register of people with significant control · Department for Business and Trade and Companies House, United Kingdom
- Companies Act, 2015 (section 93A, register of beneficial owners) · Kenya Law



