Most teams verify beneficial ownership once, at onboarding, by typing a percentage into a form. That gets you through the check and not much else. If you want the definition of UBO verification, start there; this article is about building it properly — modelling ownership as a graph, resolving it to real people, and keeping it current as the structure changes. It sits on top of KYC/KYB verification software.
Why a spreadsheet fails at beneficial ownership
A beneficial-ownership register held in a spreadsheet looks fine until it is tested. Ownership is not flat: companies own companies that own companies, individuals sit on multiple entities, and a single controller can appear in five places under three spellings. A row-per-owner sheet cannot express who owns what through whom, so it cannot answer the only question regulators actually ask — who ultimately controls this entity, and how do you know?
The failure compounds over time. Ownership changes; the sheet does not. When a controller sells their stake or a new shareholder is added, nothing in a flat record tells you the decision made on the old structure is now stale.
Model entities, people and the edges between them
The fix is to model ownership the way it actually exists: as a directed graph. Nodes are legal entities and natural persons. Edges are ownership or control relationships, each carrying a percentage, a role and a source. A beneficial owner is then not a field — it is the result of walking that graph from an entity down to the natural persons who meet your ownership or control thresholds.
This model earns its keep in three ways. It represents layered and shared ownership without special cases. It makes recalculation trivial when anything changes — re-walk the graph, re-apply thresholds. And it gives you something to show an auditor: not a number, but the path that produced it.
Resolve ownership to a person, not a name
Registry data is messy. The same individual appears as "J. Whitfield", "James A. Whitfield" and "James Whitfield" across filings, with different addresses and dates of birth. If you treat those as three people you over-report directors; if you collapse two genuinely different people you miss a controller entirely.
Entity resolution is the discipline of deciding when two records are the same real-world person or company. In practice it is a weighted match across name, date of birth, nationality and address, producing a confidence score rather than a yes-or-no. The pragmatic design keeps the source records, links them to a resolved identity, and keeps a human in the loop for the ambiguous middle instead of guessing silently.
Apply thresholds consistently — and record why
Most regimes define a beneficial owner as someone holding above a threshold (commonly 25% of shares or voting rights) or exercising control by other means. The trap is treating the threshold as the whole rule. Owning 20% while chairing the board and signing off the accounts is control, and control is what the regulation targets.
Encode the policy explicitly: the ownership threshold, the control tests, and the fallback to a senior managing official when no natural person qualifies. Then version it. When the threshold changes or you add a jurisdiction, you need to know which past decisions were made under which rule — both to re-run them and to defend them.
Re-verify on change, not on a calendar
Periodic review is the floor, not the strategy. The structures that matter change on their own schedule: acquisitions, share sales, new directors, court-ordered ownership. Waiting twelve months to notice is how a sanctions-linked beneficial owner slips through between reviews.
The better trigger set is event-driven — a registry filing change, a new shareholder, an upstream screening hit, a change of control — with a periodic sweep as a backstop. Every re-verification re-walks the graph, re-applies the current policy, and stores the new decision alongside the old one so the history stays intact.
Make every decision auditable
For each beneficial-ownership decision, an auditor will ask what you knew, when, from which source, and under which policy. If the answer is a set of registry calls and a policy document someone edited last week, the trail does not hold.
Immutable evidence is the goal: the source responses, the resolved ownership graph, the policy version, and the resulting decision, all bound to a timestamp and not editable after the fact. That record is what turns UBO verification from a checkbox into something you can stand behind.
Key takeaways
- Beneficial ownership is a graph of entities, people and ownership edges — not a flat list of names and percentages.
- Resolve messy registry data to real people with confidence scoring, and keep a human in the loop for ambiguous matches.
- Trigger re-verification on structure changes, and store an immutable, versioned decision record.
Founder & CEO
Yashwant DanielYashwant Daniel writes on compliance and trust engineering, custom software delivery and applied AI at Solvstrat — covering how regulated and mid-market teams build systems that hold up under audit, scale and scrutiny.
LinkedIn