Adding a company

Search by name or enter a registration number directly. A registration number is unambiguous and a name is not, so where you have the number, use it — two companies can share a trading name, and only one of them is your customer.

Each company is verified against its national register the moment you add it. You start from a known state rather than from an assumption, which matters later: an audit trail that begins with “we do not know what was true on day one” is not evidence of anything.

The first check

The onboarding check returns identity and status, ownership where the country makes it available, and a screening against the published sanctions and PEP lists. This is the record of what was true on the day you started doing business — dated, sourced, and kept.

What comes back depends on the country. The coverage table is the authority on that, and the product will not show you a signal it cannot actually deliver for the country you are looking at.

Who ultimately owns the client

Where the client is held by another company, the file follows the chain upwards, company by company, until it reaches natural persons or runs out. Each company in the chain is identified by the registration number the register publishes for it, not by matching a name: several live companies can share a name, and a wrong parent in a client file is worse than no parent at all.

Holdings are written as the band the register publishes, such as 75 to 100 per cent. Companies House does not publish exact figures for this, so neither do we. The file also records why a party has control, which is not always shareholding: a right to appoint and remove directors is control too, and a supervisor will ask about it.

Where the chain stops, the file names the point and the reason. A parent registered abroad, a register we could not read, a company listed on a regulated market that files nothing because it is exempt. In each case the file says that nothing above that point has been checked. A branch we could not follow is not a branch that ends, and the two must never read the same.

This runs against the United Kingdom’s register of people with significant control. In the other countries we cover, the beneficial ownership register is closed to us by law rather than missing, and the file says so instead of leaving a blank. The coverage table is the authority.

When no beneficial owner can be established

Some clients have no identifiable beneficial owner. The rules anticipate that. Once you have exhausted every means of identifying one, you may treat the senior person responsible for managing the client as the beneficial owner instead. It is a treatment the rules require, not a finding that the person owns anything, and the file says so in those words.

The condition is the part that gets missed. You have to be able to show what you did before you got there. EntityWatch asks for it and will not let the file be approved without it: every step you took to identify the beneficial owner, which of the two situations you are in, who you are naming, how you verified them, and any difficulties you met. Leaving the difficulties empty is a real answer and is recorded as one. Never being asked is not.

The person you name is then screened against sanctions and politically-exposed-person sources, because from that moment the file treats them as the beneficial owner. If the client is itself listed on a regulated market, that is a different answer entirely and the file records it as the exemption it is. A listed company higher up the ownership chain does not answer the question about your client.

One case is refused outright. If the beneficial ownership register was never read, EntityWatch will not accept the fallback at all, and no form is offered. Not reading a register is the opposite of exhausting every means. A file built that way would read well and would not survive being asked about.

Ongoing monitoring

After that, the companies you monitor are re-checked on a schedule. How often depends on your plan; what is compared depends on the country. In the United Kingdom, the one country that publishes a filed change history we can read, we read the change itself — including changes of owner and of director. Everywhere else we compare the current record against the last one we stored and report the difference: name, status, address, industry, legal form. That comparison cannot see a change of owner or director, because those registers do not republish those fields for us to compare.

The distinction is not pedantic. A filed change tells you what happened and lets you point at the document. A snapshot difference tells you that something is not what it was. We label them differently so you never present the second as the first.

When something changes

You get the change, the date it was filed, the date you were told, and a link to the filing. Nothing is summarised away, and nothing is scored into a single number that hides what it was made of. If a company is dissolved, the alert says the company is dissolved.

Possible sanctions matches

Sanctions screening compares names, and names are not unique. A fuzzy resemblance between a company you monitor and an entry on a list is a question, not an answer.

So a possible match is routed to review with the underlying record attached, and it is presented as something to look at — never as a confirmed hit, and never in red. EntityWatch does not declare a company sanctioned. That determination is yours to make, and it is the one part of this workflow that should not be automated.

The audit trail

Every check and every change for a company forms one timeline: what was true when you onboarded them, what changed since, when each change was filed, and when you were notified. The export carries the source-document links with it — they are not stripped when it becomes a document.

This is the part that is worth paying for. Monitoring tells you something today; the audit trail is what answers the question an inspector asks two years from now.