Four boxes, three jurisdictions, and a lawyer being helpful
The onboarding meeting is going smoothly. Across the table is the client's lawyer — pleasant, well-prepared, and a great deal more fluent in this structure than the analyst asking the questions.
She has brought a chart. Four entities, a holding company, a trust arrangement and a management vehicle. She talks through it clearly and answers every question accurately.
Forty minutes in, the analyst has a complete structure diagram and cannot say, in a sentence, who actually decides what happens to this money.
Nobody has obstructed anything. Every answer was true. And the interview has produced almost nothing, because the analyst asked about the structure and the structure is not what he needed to understand.
The obstacle is competence, not resistance
This is the inverse of most due diligence scenarios. There's no offence, no evasion, no reluctance — and that's precisely what makes an ownership and control interview fail quietly.
Four features define it.
The counterparty is a professional. A lawyer, an accountant, a corporate services provider. They are cooperative, they answer what's asked, and they have no obligation to answer what wasn't.
Accurate answers can be unhelpful. "The shares are held by the holding company, which is owned by the trust" is correct and tells you nothing about who exercises influence. Technically complete, substantively empty.
Pressing looks like incomprehension. The social cost of asking again is that you appear not to have followed the chart — so people stop asking, and the file records a structure nobody in the firm can actually explain.
And the question you need answered isn't a legal one. Who owns it is a documentary fact. Who decides, who benefits, and who could stop a decision are practical questions, and the documents frequently don't answer them.
Say what you're trying to establish
Before any question about entities, name the objective in ordinary language.
"Before we go through the chart — what I need to be able to do at the end of this is write a short paragraph explaining who controls this and who benefits from it, in plain English, for someone who won't see the diagram. Can we work toward that?"
This reframes the whole meeting. It tells a competent intermediary what standard is being applied, which most of them will happily meet, and it makes later plain-language questions expected rather than naive. Intermediaries generally don't obstruct — they answer the question they were asked, and this changes the question.
Ask about people and decisions
"Who decides?" Not who owns. If a significant transaction were proposed next month, whose agreement is needed in practice?
The stop question. "If the directors wanted to sell the business tomorrow, is there anyone who could prevent it?" The single most productive question in this interview. It surfaces influence that sits outside shareholdings — a protector, a lender, a family member, an arrangement that isn't in the chart.
"Who benefits, in practice?" Distinct from who is entitled to. Both matter and they frequently differ.
"Why is it structured this way?" Almost always answerable and almost never asked. The answer is usually mundane and legitimate — tax, succession, a joint venture, a historical acquisition — and having it in the file is worth more than another entity name. Where there is no coherent answer, that is itself the most useful thing the meeting produced.
"Has any of this changed in the last couple of years, and is anything changing now?" Structures in motion are a different object from static ones.
Don't accept a document as an answer
"It's all in the trust deed" is not a response to a question about who exercises control. Documents are evidence; the explanation is what the interview is for, and a file containing only documents leaves the firm's understanding sitting in an unread attachment.
The correct move is unembarrassed: "I'll read it — but I'd rather have your explanation in your own words as well, because I need to be able to summarise it."
And say "I don't follow that"
Plainly, without apology, as many times as required.
This is the hardest part of the interview and it is entirely a confidence problem. A junior analyst facing a senior lawyer will nod through an explanation they haven't understood, because asking a third time feels like an admission.
It isn't. Nobody who understands a structure well minds explaining it again, and the people who do mind are telling you something. An analyst who leaves with an explanation they can repeat has done the job; one who leaves with a chart and a vague sense of having been told has not.
Four ways it goes wrong
The chart-accepter, who treats the diagram as the deliverable and never asks who decides.
The embarrassed nodder, who stops asking in order not to look lost.
The entity-lister, whose interview produces names and jurisdictions and no explanation of how any of it works in practice.
The intermediary-deferrer, who directs everything through the adviser and never hears anything from the people actually involved.
Why this isn't trained
Training teaches the framework, not the interview. Staff learn definitions, categories and documentary requirements. Asking a professional intermediary a question they haven't volunteered an answer to is a different skill with no curriculum.
The asymmetry is never acknowledged. A twenty-six-year-old analyst interviewing a partner at a law firm is a status problem as much as a technical one, and nobody says so, so nobody is given permission to be persistent.
Completeness is confused with understanding. Checklists are completed, documents are collected, and a file that satisfies the process can leave nobody in the firm able to explain the client in a sentence.
And peer role play produces a helpful explainer. A colleague playing the intermediary volunteers the useful information, because they want the exercise to work. The real counterpart is cooperative and answers exactly what was asked — a register that requires the interviewer to ask better, which a helpful colleague quietly removes.
What complex-structure interview training can rehearse
A simulation can hold a courteous, expert intermediary who answers accurately and narrowly, so the interviewer has to find the questions that produce a usable explanation — and score whether the analyst left with something they could summarise in plain English. Foretell AI supplies the counterparty configuration, transcripts and rubric-based scoring; the firm's definitions, thresholds, documentary standards and all regulatory determinations stay with the firm.
Four to build:
- The accurate narrow answerer, who never says anything untrue and never volunteers anything.
- The document-pointer, who responds to every question with a reference to an instrument.
- The senior intermediary, whose seniority is the obstacle — testing whether a junior interviewer keeps asking.
- The one with no good answer to “why this structure”, where the exercise is to notice the absence and record it without accusation.
Design caution. Definitions of ownership and control, documentary requirements and risk thresholds vary by jurisdiction and are set by the firm; nothing here defines them, and scenario libraries should not either. No typologies or risk indicators appear in this material by design. The exercise rehearses interview quality only and confers no compliance assurance.
Designing the module
Pass one — the framing. Score whether the interviewer stated the plain-English objective before asking about entities.
Pass two — the questions. Score how many concerned people and decisions rather than entities and percentages, and whether the stop question was asked.
Pass three — the summary. At the end, ask the interviewer to explain the structure in four sentences without notes. That is the actual output and it's the only pass that matters.
Rubric on observable behavior: Was the objective stated? Ratio of people-and-decision questions to entity questions. Was the stop question asked? Was "why is it structured this way" asked? How many times did the interviewer say they didn't follow? Could they summarise it afterwards in plain English?
The four-sentence summary is the measure worth building the whole module around. It's binary in practice, it's the thing the file is supposed to contain, and it exposes a gap that checklists actively conceal.
The operator case
Test your own files with the paragraph test. Pick ten onboarded structures and ask whether anyone in the building can explain each in plain English without the diagram. The result is usually uncomfortable and it's the fastest available argument for doing something about this.
Checklist completion and understanding are different outputs, and most quality assurance measures the first. A file can be fully compliant and leave the firm unable to say who controls the client.
Junior staff need explicit permission to be persistent with senior intermediaries. This is a status problem, and it's solved by someone senior stating plainly that asking three times is expected behaviour — not by more technical training.
And "why is it structured this way" belongs in the standard question set. It's cheap, it's usually answered well, and its absence is informative in a way no checklist captures.
For compliance and financial crime programmes, this is a useful distinction between two things usually taught as one: collecting the required information, and understanding the client. The first is a process and the second is an interview.
Frequently asked questions
How do you interview a client's lawyer about a complex ownership structure? State the plain-English objective first, ask about people and decisions rather than entities and percentages, ask why the structure exists, and don't accept a document reference in place of an explanation.
What's the best question to ask about control? Who could prevent a major decision. It surfaces influence that doesn't appear on a shareholding chart — a protector, a lender, a family member, an informal arrangement.
What if you don't understand the structure being explained? Say so, plainly, as many times as needed. Nodding through an explanation you can't repeat produces a file nobody in the firm can use, and competent intermediaries don't mind explaining twice.
How do you know if a due diligence interview worked? Try to summarise the structure in four sentences afterwards without notes. If you can't, the interview didn't produce what it was for, however complete the checklist is.
The short version
The lawyer was helpful, the chart was accurate, and forty minutes later nobody can say who decides.
Tell her what you need to end up with. Ask who decides, who benefits in practice, and who could stop a sale tomorrow. Ask why the structure looks like this. Don't take a deed instead of an answer. Say "I don't follow that" as often as it takes.
Then try to explain it in four sentences. If you can't, you haven't finished — and the checklist will tell you that you have.
Foretell AI lets banks and financial firms build conversational simulations — including complex-structure onboarding, intermediary interviews and control questioning like the one above — with configurable counterparties, transcripts, recordings, and rubric-based evaluation. If your files are complete and nobody can explain the client in a paragraph, we're happy to walk through how other firms have structured it.