Ondorse × CnovaPay · company data assessment

What seven registers actually return

A 405-company sample assessed on three questions: can each company be identified, what does the register return about the people who run it, and can its beneficial owners be established. Coverage turns out to be the easy part — the interesting variation is in how much detail each register carries about a person once it has named them.

Assessed 2026-09-10 Sample 405 rows · 404 distinct companies Registers CN · DE · ES · FR · NL · PL · RO Mode authoritative / verification

Companies found

398 of 404 companies matched a register entry · all six misses are listed below

Outcome mix

Every company falls into exactly one outcome. Bars are absolute counts, scaled to the largest register.

The six that could not be identified

Attribute depth

What each of the three datapoints actually carries, field by field. Measured per company — for people, a field counts as present if any returned person has it. The commentary below describes all seven registers.

Company

The profile of the entity itself.

Legal representatives

The people who can act for it.

Beneficial owners

The people who ultimately own it.

Guaranteed
Five company fields are universal; everything past them is optional

Legal name, registered address, legal form, status and registry identifiers came back for 100% of companies in all seven registers. Nothing else does. Activity code reaches 78%, share capital 66%, a tax or VAT number 65%, employee count 48% — and commercial name and website are effectively absent at 5% and 1%. Anything built on those last two will be empty most of the time.

People
A person is always named and never fully described

Name, role and whether the representative is a person or a company are universal. Every identifying field beyond that is register-specific rather than generally available: a date of birth reaches 68%, gender 41%, place of birth 38%, and nationality and residence address only 32% each. Romania supplies seven of the nine fields for every company; China and Spain supply two.

Counter-intuitive
Beneficial owners are better documented than directors

On the same sample, owner records are richer than representative records on every shared field: date of birth 88% against 68%, residence address 58% against 32%, place of birth 48% against 38%. Beneficial-ownership filings demand more identity evidence than director filings do, so the harder datapoint to obtain is the more complete one once you have it.

Divergence
Ownership is quantified in two incompatible ways

France states a stake as a percentage (100%, never a monetary amount). Poland states it as a capital contribution (96% carry an amount, 3% a percentage). China supplies both. A single “ownership %” column across these registers would be empty for almost every Polish company while the information is in fact present — as PLN 2,500 rather than a share of the total.

Consequence
No register is strong everywhere, and the gaps do not overlap

France is the only source giving voting rights (100%) and owner addresses; Poland is the only one giving owner nationality reliably (100%); Romania is the only one describing representatives fully; China quantifies ownership precisely but supplies no dates of birth at all. A consolidated risk view inherits a different evidence base per country, not a common one.

Legal representatives

398 identified companies · every one returned at least one representative
Finding
Representatives are the one thing every register always returns

All 398 identified companies came back with at least one representative, in all seven registers, each with a role attached. No other datapoint in this assessment is complete everywhere — company profiles vary in depth and beneficial ownership is unavailable in Germany entirely.

Structure
Nine companies in ten are run by a single person

457 representatives across 398 companies — a mean of 1.15, with 93% returning exactly one. Governance depth is not where the complexity sits in this sample; the variation is in how much each register says about the individual.

Outlier
One Spanish company returned 23 representatives

Against a sample mean of 1.15, a single Spanish entity accounts for 23 of Spain's 24 representatives. With only two Spanish companies in the sample this says more about that one company's governance than about the register, but it is the only board of any size here.

Beneficial ownership

262 companies were tested for beneficial ownership. An answer came back for 234 of them; 26 cannot be assessed without register credentials or a resolved company, and 2 are still running.
Finding
Every register carries a different pair of fields

Ownership percentages come back for 85% of companies with owners, but not evenly: France and China return them for 100%, Poland for 3% — one company out of 32. Dates of birth invert part of that picture: France 100%, Poland 78%, China none. Nationality is the scarcest field overall at 35%, and the only register that supplies it consistently is Poland, at 100%. No register returns all three.

Finding
China names people but never dates them

The same gap appears on both sides of the Chinese data: 0% of its representatives and 0% of its beneficial owners carry a date of birth, while ownership percentages are always present and nationality appears for 16%. Chinese ownership can therefore be quantified but its people cannot be screened on identity alone — the constraint is the same one seen in the representative data, not a separate UBO problem.

Caveat
Not every empty result means "no owner"

21 companies returned an authoritative empty. In Poland that is mostly structural — 5 of 7 are sole proprietorships with no share capital to attribute. In France it is not: 13 of 14 are SARLs and SASs, forms that should have a beneficial owner on file. Those are worth treating as register gaps rather than as confirmed absences.

Not available
German beneficial ownership is out of scope for this assessment

The German transparency register is not accessible under the credentials used here, so all 26 German companies flagged for UBO testing are unassessed. This is an access question, not a measurement of the register's content — it is the single largest gap left in the sample, and the only one that is not a property of the data.

Company by company

Filter, search and sort. 404 unique companies; the duplicated sample row is merged.
Country Registration nº Company Source Outcome Register status Reps UBO Failure detail