KYB vs KYC: Why Verifying a Company Is Nothing Like Verifying a Person

Ask most people what KYB is and you'll get some version of "KYC, but for companies."

Technically true, but it is like saying chess is checkers, but for adults.

Verifying a person is a bounded problem. One passport, one address, one date of birth, a screening hit or two to clear. Verifying a company is an unbounded one: a legal entity is a wrapper that can contain other wrappers, registered in jurisdictions with wildly different transparency, changing shape whenever a lawyer files a form. In the investment world - where the "customer" is routinely a Cayman feeder owned by a Luxembourg holdco owned by a trust - KYB is where compliance programs are truly tested.

The core difference: people end, structures don't

KYC on an individual terminates. You verify identity, screen the name, assess the risk, done.

KYB doesn't terminate until you decide it does. Every entity opens three new questions: who owns it, who controls it, and where does it sit? Each answer can be another entity, in another registry, in another language, under another disclosure regime. The work isn't verification so much as excavation - you keep digging until you hit natural persons, and the ground varies from glass (a UK company with a filed PSC register) to concrete (a nominee-held BVI entity).

The 25% beneficial ownership threshold gives you a potential stopping rule. Knowing when you're allowed to stop digging is not the same as knowing what you found.

Where standard KYC tooling breaks

Most KYC software was designed for the bounded problem: onboard a person, check a database, produce a score. Point it at a master-feeder structure and watch it struggle.

Registry data is fragmented and stale - a structure chart is accurate on the day it's drawn and decaying every day after. Documents don't standardise: a certificate of incumbency from one jurisdiction, an extrait K-bis from another, a trust deed that runs to ninety pages. Ownership percentages have to be multiplied through chains and cross-checked against control rights that don't follow the equity. And the entities themselves change - mergers, redomiciliations, share transfers - without anyone sending you a courtesy note.

So the actual KYB work in most firms happens outside the tools: analysts reading PDFs, drawing structure charts in PowerPoint, and chasing lawyers by email. The system of record is a shared drive. The institutional memory is whoever did the file last.

Why this matters commercially, not just theoretically

For funds and their administrators, entities aren't the edge case. Institutional subscriptions come wrapped in structures almost by definition.

That means KYB throughput is onboarding throughput. Every day an analyst spends untangling a holding chain is a day an allocation sits unexecuted and a client wonders why your competitor onboarded them faster. It also means KYB quality is your regulatory exposure: when enforcement actions cite CDD failures, the failure is rarely a missed passport. It's an ownership chain nobody followed to the end, a controller nobody identified, a structure that changed after onboarding and was never re-examined.

Doing KYB properly: the practical version

1. Map before you verify

Resolve the full structure first - every layer, every jurisdiction - before collecting documents. Firms that collect documents first end up with a pile of paper and no picture. The structure chart is the deliverable and documents are its evidence.

2. Follow control, not just ownership

Equity percentages are the easy half. Voting agreements, board control, general partner roles and trust arrangements can put effective control somewhere the cap table doesn't show. A UBO analysis that only multiplies shareholdings is half an analysis.

3. Treat every structure as perishable

An entity file has a shelf life measured in months, not years. Build refresh triggers on registry changes, filings and adverse media - not just calendar-based reviews that formalise staleness.

4. Let machines do the reading

The mechanical core of KYB - extracting data from registry filings, incumbency certificates and constitutional documents, reconciling it across sources, computing effective ownership through chains - is now squarely within AI's capabilities. This is exactly the work that eats analyst days and produces transcription errors. Automate the excavation; reserve humans for the judgement call at the bottom of the hole.

5. Evidence the trail, not just the conclusion

A regulator reviewing your KYB doesn't want to know that you concluded Mr. X owns 40%. They want to see how you got there. Every layer of the chain needs its source, its date, and its document attached.

Where Steward stands

We built Steward for the investment world specifically because this is where KYB is hardest and where generic tools give up. Layered structures, cross-border registries, ninety-page trust deeds - that's the normal Tuesday. An AI-first system that reads the documents, builds the chart, computes the chain and keeps it all evidenced turns KYB from an excavation project into a process.

To sum up

KYC verifies a person; KYB excavates a structure. If your KYB process is an analyst, a PDF and a PowerPoint chart, this isn't a process. At best, it is reliance on human capital. And people leave.

Interested to see how AI-first system can read documents, build the chart, compute the chain and keep it all evidenced? Book a demo with us