How Cross-referencing Builds a Usable Brief

Cross-reference is the editorial part of the audit. It is the practice that distinguishes a useful brief from an unsorted feed. A short account, written in plainer terms than the methodology paper, of what cross-reference actually looks like in practice.

The audit reads each source individually. A reading of Companies House gives one set of facts about the client; a reading of the FCA Register gives another; a reading of the Land Registry gives a third. Each reading is, on its own, partial.

Cross-reference is the practice of placing each reading next to the others and noticing what they together describe.

A worked example. A client appears on Companies House as a director of a UK company at a particular registered office. The FCA Register shows the same individual as an approved person, with a different home address recorded against their FCA application from three years prior. The Land Registry shows a property at the FCA-recorded address, owned by an Isle of Man company. The Isle of Man Companies Registry shows the owning company has the original Companies House registered office as its director's contact.

Each entry, read singly, is unremarkable. Read together, they describe a structure: an individual who has moved their personal residence, retained the previous address for a UK directorship, holds the previous-address property through an Isle of Man company, and uses the new address as the contact for the offshore company. The structure is not, in itself, irregular; it is what cross-reference would surface.

The audit's reading would identify the structure and present it as a finding. Whether the structure is correctly maintained, whether the various entries should be reconciled to a single current address, whether the offshore-company arrangement is currently appropriate: these are questions the audit raises but does not answer. The questions belong to the client's counsel and tax advisers.

What cross-reference does not do is fabricate connections that the records do not support. The audit's confidence readings express how strongly each connection is supported. A connection at 95 percent confidence is one that the records explicitly identify; a connection at 60 percent confidence is one that the records strongly suggest but do not directly assert. Findings below 50 percent confidence do not appear in the delivered report.