Entity resolution for private company data

The simplest version of an entity resolution problem is this: a company appears in one source under its legal name, in a second source under the trading name on its website, in a third as a subsidiary of its parent group, and in a fourth under a name it held five years ago before a rebranding. All four entries describe the same business. None of the names match.

Before any screening, valuation or AI-assisted research can proceed reliably, these records need to be collapsed to a single, stable identity. That is what entity resolution means for private company data, and it is the step most commonly skipped when a file is assembled from multiple sources.

The registration number is the anchor

A company name is a marketing choice. A registration number is a legal fact. Every jurisdiction that requires incorporation issues a unique number at the point of registration. In the United Kingdom it is an eight-character company number. In Singapore it is a UEN. In Australia it is an ACN distinct from the ABN used for tax purposes. In Hong Kong it is a company number issued by the Companies Registry.

That number does not change when the company changes its name, rebrands, or adds a trading name. It changes only in specific circumstances — a re-registration from one entity type to another, a jurisdiction change — and those circumstances create a paper trail precisely because the identifier is considered the authoritative anchor.

Matching records from different sources on the company number rather than the name is the foundation of reliable entity resolution. Without it, name-based matching produces a list where the same business appears more than once under different names, and where different businesses appear to be the same one because they share a generic word.

The three layers of name ambiguity

Names cause identity failures in three distinct ways, and each requires a separate resolution step.

The legal name is the name the company is registered under. It may be nothing like the brand the company trades under — a long-established holding vehicle often carries a founder’s name or a number that bears no relation to the operating business beneath it. A record that carries only the legal name is not wrong, but it is incomplete.

The trading name is what the business calls itself publicly. It may be registered as a business name in some jurisdictions (ACRA in Singapore and ASIC in Australia maintain separate business-name registers), or it may be entirely informal and appear only on the company’s own website. A screen run against a database of trading names will miss legal entities whose trading name has not been separately recorded. A screen run against a database of legal names will miss any company whose public identity is built around its trading name rather than its legal one. Both screens run against the same underlying population.

The group name is the brand associated with the parent, which may appear on subsidiary companies’ websites, email footers and correspondence even when the subsidiary is the actual legal counterparty. Approaching the group entity when the trading subsidiary is the relevant one — or the reverse — is one of the two most common outreach failures in target identification. The legal name versus trading name distinction and the group resolution question are related but separate steps.

Why address fields do not resolve identity

A registered office address is a legal address for service. For many small companies it is their accountant’s address. Two separate companies using the same accountant will share a registered office address that looks, in a database, like a deduplication signal when it is nothing of the kind.

A trading address is more specific to the business, but a business with multiple sites will have multiple trading addresses, and a business that has moved will have a historical address that still appears in older records. Address is a useful field for outreach routing, but it is unreliable as an identity anchor.

The practical implication is that neither address field should be used to collapse or merge records. Records should be merged only on the registration number. Address fields confirm where to direct an approach once the entity is resolved, not whether two entries are the same entity.

Group resolution as a separate step

A single trading business commonly sits beneath a holding company, an operating subsidiary and perhaps a property entity, all separately incorporated and all appearing as distinct companies in a registry export. The operating subsidiary is the entity that employs staff, files revenue-bearing accounts and holds the commercial relationships. The holding company is usually the one approached by a buyer or investor. The property entity is neither.

Resolving which entity is which requires reading the filed accounts, not just the registry entry. An operating subsidiary files accounts that describe a real business: payroll costs, trade debtors, a description of the principal activity that matches what the company actually does. A holding entity files at group level or shows only investment income.

This step cannot be automated without a risk of misclassification that propagates into every downstream use of the record. It is the difference between a registry export and a research file — and it is what the evidence checks in a well-maintained company record are designed to capture.

Cross-border work and the LEI

When research spans multiple jurisdictions, national registration numbers create a coordination problem: a UK company number and a Singapore UEN are both reliable within their jurisdictions, but there is no common format that links them.

The Legal Entity Identifier — a 20-character ISO standard code issued by GLEIF-accredited Local Operating Units — provides a cross-jurisdictional stable identifier. For entities that hold one, an LEI record links the legal name, registered address, entity status and ultimate parent, using a consistent data schema across participating jurisdictions.

The practical limitation is coverage. LEI adoption among private companies is concentrated in financial services and regulated entities. A manufacturing subsidiary in a mid-market group typically has no LEI, not because it is obscure, but because no regulatory requirement has made registering one worthwhile. Where an LEI exists, it supplements the national registration number. Where it does not, the national number remains the only reliable anchor.

What a resolved record contains

A record that has been through a proper identity resolution process holds at minimum:

the national registration number with its jurisdiction labelled;
the current legal name as it appears in the registry today;
the trading name if different, with a note of its source;
the identified operating entity within the group, if the company is part of a group;
registry status and the date of the most recent status check;
registered office address, clearly labelled as registered office rather than trading address;
trading address if identified from the company’s own sources, with the source noted; and
a flag for any field where the source date is more than twelve months old.

What a resolved record does not contain is any field filled by estimation. Where trading address, ownership structure, financial data or contact details are absent from public sources, the field stays blank. A blank is an honest record of what the sources established. An estimate passed off as a fact is a resolution error waiting to propagate.

The coverage and verification standard for a well-maintained private company directory should describe how it handles each of these fields and what population the resolution process was applied to. Any file that does not draw a clear line between registered data, research data and modelled data has not resolved its records — it has assembled them.

To understand how a resolved record fits the research your work requires, see the coverage detail and register interest.

Frequently asked questions

What is the most reliable identifier for a private company?

The registration number issued by the relevant national registry — a UK company number, a Singapore UEN, an Australian ACN or a Hong Kong company number. Unlike a name, a registration number is stable, unique within its jurisdiction and does not change when the company is renamed. Matching on name alone produces false positives (two unrelated companies sharing a common word) and false negatives (the same company appearing under different names in different sources).

Why can't you match private companies by name?

A company name is neither unique nor stable. Two companies can share a name in different jurisdictions. A single company can change its registered name, operate under a different trading name, and have a parent with a third name — all at the same time. Name-matching is reliable enough when a dataset has a single source and a narrow geography; it fails at exactly the scale that acquisition and market-mapping research requires, where records arrive from multiple registries across multiple points in time.

What does a resolved private company record mean in practice?

A resolved record links a unique registry identifier — company number or equivalent — to its current legal name, its operating entity within any group structure, its registry status and a source date for each field. Resolved does not mean complete; fields such as trading address and financial data may remain blank where public sources do not supply them. Resolved means the record can be matched to one and only one legal entity without depending on the company name remaining consistent across sources.

How do you identify the operating entity within a group?

Check the filed accounts. An operating subsidiary typically files accounts showing real business activity — payroll, trade creditors, a named activity description. A holding company typically files at the consolidated group level or shows only investment income. A dormant shell files nil transactions. When the accounts are not conclusive, the company's own website and named trading address should direct research to the relevant entity. The goal is to record which entity trades, which owns, and which one an approach would actually reach.

When does a Legal Entity Identifier help with entity resolution?

An LEI — a 20-character ISO identifier issued to a legal entity by a GLEIF-accredited body — provides a cross-jurisdictional stable identifier and, for entities that carry one, links the legal name, registered address and ultimate parent. Not all private companies hold an LEI; adoption is most consistent in financial services and regulated entities. Where an LEI exists it supplements the national registry number; where it does not, the national registration number remains the anchor. An LEI's absence is not evidence that the entity is obscure — most private companies outside regulated sectors simply have no reason to register one.

Sources

Introducing the Legal Entity Identifier (LEI) — Global Legal Entity Identifier Foundation; accessed September 22, 2026
Companies House — company information service — Government of the United Kingdom; accessed September 22, 2026
Accounting and Corporate Regulatory Authority — Government of Singapore; accessed September 22, 2026
Australian Securities and Investments Commission — Government of Australia; accessed September 22, 2026