Ownership mapping
At the first level, how company ownership works is straightforward. Shares carry economic rights and voting rights, and whoever holds them is recorded on the shareholder register or in the memorandum of association. Ask who owns a company, and that register answers the question.
The answer stops being useful the moment a shareholder is itself a company. A register that lists a holding company at 40 percent has told you where to look next, not who benefits. Repeat that two or three times across two or three jurisdictions and the paper trail still exists, but it no longer points at a person. Reconstructing the chain is the work of following each corporate shareholder upward until only natural persons remain, and then proving that the picture you have drawn is accurate.
This walkthrough covers the sequence itself, what makes each jurisdiction behave differently, how ownership gets verified rather than assumed, and the signals that tell you a structure deserves a closer look.
Why it matters here
A trading company licensed in a Dubai free zone can be held by a Cyprus intermediate, which is held by a Luxembourg vehicle, which is held partly by a trust. Each of those jurisdictions publishes a different amount, in a different language, on a different schedule. The structure is not unusual and it is not, on its own, a problem. It simply means the register in front of you covers one link of a chain that runs through four legal systems.
The name on the register, the number of shares, the class attached to them, and the date the holding was recorded. This is a matter of record and it is where every reconstruction begins.
The natural person who ultimately owns or controls the entity, whether or not their name appears anywhere in the filing. Indirect holdings are calculated by multiplying percentages down the chain.
The right to appoint or dismiss the majority of directors, veto rights, powers of attorney, or golden-share arrangements. A person holding nothing on the register can still direct the company.
Exact legal name, licence or registration number, jurisdiction of incorporation, registered address, incorporation date. Most failed reconstructions fail here, because a name-only search picked up a similarly named entity in another emirate or another country and every subsequent layer was built on the wrong foundation. The registration number is the anchor, not the name.
Retrieve the shareholder register, the memorandum of association or the equivalent constitutional filing. Record every holder, the exact percentage, the share class, and the date the holding took effect. Percentages that do not total 100 are a signal that the record is incomplete or out of date, not a rounding artefact to be ignored.
Natural persons are candidate endpoints and move to identity verification. Every corporate shareholder is an open branch. Trusts, foundations, partnerships and special purpose vehicles are also open branches, and each needs its own treatment: for a trust you are looking for the settlor, the trustees and the beneficiaries, not a shareholding.
Treat every corporate shareholder as a new target and repeat moves 1 through 3 against it. Multiply the percentages as you descend to calculate indirect holdings: a person holding 20 percent of a company that holds 100 percent of a company that holds 40 percent of the target holds 8 percent of the target. Keep climbing until a branch resolves to natural persons or to a documented dead end. Structured ultimate beneficial owner screening is what turns this from a manual crawl into a repeatable process across hundreds of files, with an analyst reviewing what each pass returns.
Shareholder agreements, side letters, voting arrangements, nominee declarations, trust deeds, board appointment rights and powers of attorney all confer influence that percentages will never reveal. A holder at 12 percent with the right to appoint the majority of the board matters more than a passive holder at 30 percent.
Verify the identity of each natural person against government-issued documents, record which source produced each link and on what date it was retrieved, and state plainly which branches remain unresolved and why. A reconstruction without its sources attached is an assertion, not a finding.
Worked example
The entities below are invented to illustrate the arithmetic. Meridian Gulf Trading FZ-LLC is the target, licensed in a Dubai free zone. Its register shows three direct shareholders, two of which are companies.
Running the arithmetic down each branch: L. Farrow holds 20 percent of Castellan, which holds 100 percent of Harbourline, which holds 40 percent of the target, giving 8 percent indirectly. Farrow also sits behind the nominee holding at Al Nadir, which holds 35 percent directly. The two paths combine to 43 percent, comfortably above the 25 percent threshold. R. Kassim qualifies on a direct 25 percent holding. The family trust accounts for a further 32 percent, which means the settlor, trustees and beneficiaries have to be recorded rather than treated as an endpoint.
| Layer | Entity or person | Jurisdiction | Record used | What it resolved |
|---|---|---|---|---|
| Target | Meridian Gulf Trading FZ-LLC | Dubai free zone | Licence extract, shareholder register | Three direct holders, two of them corporate |
| Layer 1 | Harbourline Holdings Ltd | Cyprus | Registry filing, annual return | A single corporate parent at 100 percent |
| Layer 1 | Al Nadir Investments Ltd | British Virgin Islands | Registered agent record, nominee declaration | Shares held on behalf of one natural person |
| Layer 2 | Castellan Partners SARL | Luxembourg | Registry extract, constitutional documents | Trust at 80 percent, one individual at 20 percent |
| Outcome | L. Farrow | Natural person | Cross-checked across three registers | 43 percent indirect, above the threshold |
Friction points
The method above is the same everywhere. What changes between one layer and the next is how much the local system will actually hand you.
Verification
Drawing a structure and standing behind it are two different exercises. A diagram assembled from summaries is a hypothesis. It becomes an intelligence finding when each link traces back to a record you can name, retrieved on a date you can state, and reviewed by a person who can explain the reasoning.
The material that carries that weight comes from sources anyone can identify: official government registers, public records and public filings, registered agent records where the register itself is thin, open-source intelligence, and lawful third-party providers. There is no privileged back channel in this work, and any provider suggesting otherwise is describing something you should not rely on.
Ownership work also sits alongside screening rather than replacing it. Once the natural persons are named, PEP and sanctions screening surfaces the risk indicators attached to each of them. Both outputs are compliance intelligence prepared to support a decision: specialists review the findings, and the institution reading the file makes the call.
The signals below, commonly called red flags in compliance practice, are not proof of wrongdoing on their own. Layered and offshore structures exist for tax, succession, joint venture and privacy reasons that are entirely lawful. What these indicators do is raise the standard of evidence an analyst should require before accepting the chain as complete.
A holder who appears on the filings of dozens of unrelated companies is acting professionally on behalf of others. The record is accurate and the person named is not the owner.
No employees, no premises beyond a registered agent address, no filed activity. One in a chain is routine. Several in sequence deserve an explanation.
Four intermediates across three secrecy-friendly jurisdictions to hold a single operating company, with no tax, financing or succession rationale that anyone can articulate.
Transfers clustered just before a tender, a financing round, a listing or an enforcement action, particularly where the consideration is nominal or undisclosed.
Missing annual returns, percentages that do not reconcile, a shareholder agreement referenced but never produced, identity documents that cannot be matched to a filing.
Entities holding each other in a loop, so the chain never terminates in a natural person. The loop has to be broken by looking at control rights rather than percentages.
United Arab Emirates
Reconstruction is an investigative exercise. In the UAE it is also a standing obligation on the company itself, which is useful: it means a compliant counterparty should already hold the answer.
This section describes the framework in general terms as regulatory intelligence, not as legal advice on any particular structure. For onboarding teams it is where ownership reconstruction and AML and KYC compliance checks converge: the register the counterparty is obliged to maintain is also the document your file needs, and any divergence between the two is itself worth examining.
The practical trigger for escalation is repetition. One foreign corporate shareholder is a manual task. Three branches across four jurisdictions, refreshed quarterly across a portfolio of counterparties, is not: it needs an intelligence platform built for cross-border reconstruction, drawing on more than 400 million entities and 1.5 billion traceable ownership links assembled from official registers, public filings and lawful third-party providers. The platform assembles and organises the material. The specialist reading it decides what the structure means, and so does the institution acting on the file. Escalate when the chain crosses a jurisdiction you cannot read, when a nominee or trust interrupts it, when the counterparty carries high exposure, or when the decision the file supports is large enough that a gap would be difficult to defend later.
In short
Understanding how company ownership works past the first register comes down to discipline rather than technique: anchor the target on a registration number, take one layer at a time, multiply the percentages as you go, test for control that sits outside the shareholding, and record the source behind every link. Where the chain runs past what public records will resolve, the honest answer in the file is which branch is open and why, not a blank space.
How we work
The same principles govern every reconstruction described on this page, whether it is run by an in-house team or delivered as an intelligence assessment.
Findings are built from official government registers, public records and public filings, open-source intelligence and lawful third-party providers. Nothing rests on privileged or undisclosed access.
Every intelligence assessment passes through an analyst who checks the sources, reconciles conflicts and states the limits of what the records support.
The platform organises information and surfaces connections. It does not decide who is risky, approve or reject a counterparty, or take any action on your behalf.
Findings are presented as risk indicators and risk factors with the evidence behind them. DeepSearch Intelligence™ does not issue scores, grades or ratings for any company or individual.
DeepSearch Intelligence™ is not a law firm. Its output supports legal, regulatory and compliance intelligence work, and it does not constitute legal advice, a legal opinion or a certified legal assessment.
Each finding carries its source and retrieval date, with a full audit trail and role-based access, so any conclusion can be re-examined later by whoever needs to.
A shareholder is whoever appears on the register as holding shares, which can be a natural person, a company, a trustee or a nominee. An ultimate beneficial owner is always a natural person: the individual who ultimately owns or controls the entity, whether directly, through a chain of companies, or through control rights that do not appear as a shareholding at all. In a single-layer company the two are the same. In a layered structure they usually are not.
There is no fixed number. You trace each branch until it resolves to natural persons, or until you reach a documented point where no further record is obtainable. A three-layer chain may resolve in two steps on one branch and six on another. What matters is that every branch is either closed with a named person or explicitly marked as unresolved, with the reason recorded.
Under Cabinet Decision No. 109 of 2023, a beneficial owner is a natural person who directly or indirectly owns or controls 25 percent or more of a legal person's capital, or holds 25 percent or more of the voting rights, or has the right to appoint or dismiss the majority of its directors. Ownership held through a chain of entities counts toward the threshold. If no one meets it, the test moves to whoever exercises control by other means, and then to the senior managing official. This is a general description of the framework rather than legal advice on a specific structure.
No. Nominee arrangements are used lawfully for privacy, administration and commercial convenience in many jurisdictions, and in the UAE nominee directors and managers are declared on their own register. What a nominee does mean is that the name on the filing is not the answer to your question, so the file has to go further: the nominee declaration, the underlying agreement or the trust deed is what identifies the person behind the holding.
Official government registers, licence extracts, memoranda and articles of association, annual returns, registered agent records, trust deeds and shareholder agreements are primary. Public reporting and open-source intelligence are useful for direction and for surfacing links you would otherwise miss, but a link should be confirmed against a primary record before it goes into the chain. Every record carries a date, and the date belongs in the file alongside the finding.
It is documented rather than dropped. Record the last entity reached, the jurisdiction, the record you attempted to obtain, why it was not available, and the effect the gap has on the overall picture. An unresolved branch that represents 4 percent of the target carries different weight from one representing 45 percent, and stating which it is lets the decision-maker weigh the position rather than assume it away.
When the chain crosses a jurisdiction whose records you cannot obtain or read, when a nominee, trust or foundation interrupts it, when the counterparty carries high regulatory or reputational exposure, or when the same reconstruction has to be repeated and refreshed across a portfolio. Deeper corporate intelligence adds entity resolution across languages and spellings, control mapping beyond shareholdings, analyst review of what the records actually support, and a documented audit trail that a manual search cannot produce at scale.
No. DeepSearch Intelligence™ is an intelligence platform: it gathers information from official government registers, public records, open-source intelligence and lawful third-party providers, organises it into a traceable structure, and presents risk indicators with the evidence behind them. It issues no scores, ratings or automated assessments. Specialists review the findings, and the institution reading the report makes the decision.
Neither. A reconstruction is compliance and regulatory intelligence: structured, traceable, source-cited information reviewed by specialists to support a decision. DeepSearch is not a law firm, and its output is not legal advice, a legal opinion, a certified legal assessment or a legally binding determination about any company or person. Where a finding needs to carry legal weight, it is the starting point for counsel and formal procedure, not a substitute for either.
If you want to see what a finished reconstruction looks like on paper, the reporting levels page sets out four depths of intelligence assessment, each one reviewed by our specialists, from basic screening through to multi-jurisdictional due diligence.
DeepSearch Intelligence™ is a brand of
INSIDE RISK MANAGEMENT FZCO
Level 2 Central 1 Building
Dubai World Trade Center
