What does a code shown without its note prove?

A code shown without its explanatory note proves only that a code string was shown. It does not prove what the category covers, which version the code belongs to, or what entry a register holds. The explanatory note states what a category includes and excludes, and it is the record of meaning. The version states which edition of the classification the code belongs to. The holder states which register or dataset the code came from.

A code without those three companions is a label without its definition, and it should be treated as a lead for a check, not as a finding.

What is missing when the note is missing?

Three things are missing when the note is missing. First, the inclusions and exclusions are missing, so the boundary of the category is unknown, and a reader cannot tell whether an activity sits inside or outside it. Second, the borderline cases are missing, so the cases that sit near the boundary, and that the note resolves explicitly, are left to guesswork. Third, the version context is missing, because the same code string can have a different note in a different version, and a note from one version cannot be used to interpret a code from another. Eurostat states that its roll out work included publishing introductory guidelines that describe the overall logic of classification, and explanatory notes that give guidance for specific categories. The notes are therefore part of the classification, not an optional commentary on it, and a code quoted without its note is quoted incompletely.

How does a bare code mislead?

A bare code misleads in three common ways. It invites a reader to infer the meaning from the title alone, and titles are short, general, and sometimes similar across categories that differ in their notes. It invites a reader to assume the current version, when the code may belong to an earlier version, such as NACE Rev. 2 in a dataset for the period 2008 to 2024, which Eurostat states uses NACE Rev. 2, while statistics from 2025 onwards (FIXED, as stated by Eurostat) use NACE Rev. 2.1. It invites a reader to treat a code in a listing as a register entry, when the listing may have copied the code from an older record, an aggregate, or a different unit. Each of those inferences feels natural, and each is a check that was skipped. The check that prevents all three is the same: open the note for the category, in the version stated, and record the holder from which the code was taken.

What is the checking sequence for a bare code?

The checking sequence has four steps, and it should be followed in order. First, record the code exactly as shown, including any letter, dot, or national digit, and record where it was shown. Second, identify the version, from the source page, the dataset metadata, or the register record, and record it with the code. Third, open the explanatory note for the category in that version, and read the inclusions, exclusions and borderline cases. Fourth, open the correspondence table, where the code belongs to a version that has been revised, and record the mapping type for the category. The table below shows the sequence as a record, with the fourth step included only where a revision is involved, so that a reader does not add a correspondence check where none is needed, or omit one where it is.

Step	Action	Record
1	Copy the code exactly as shown	Code string and source
2	Identify the version	NACE Rev. 2, NACE Rev. 2.1, or the national version
3	Open the explanatory note	Inclusions, exclusions, borderline cases
4	Where revised, open the correspondence table	Mapping type for the category

The table is a working sequence, not a guarantee. A code that passes all four steps is a checked code, with a meaning, a version, and a mapping. A code that passes only the first step is still a bare code, and it should be described as such in any quotation.

What does a bare code not prove about a business?

A bare code does not prove that a named business carries on the activity shown, that it holds a licence, that it has a particular tax position, or that it is eligible for funding or a collective agreement. Only statistical use of the classification is covered by the legislation, the EUR-Lex summary states. A code in a listing, a dataset, or a report is a statistical or register label, and its meaning is in the note, while the entry for a specific business is in the register that holds the record. A bare code quoted as proof of activity, licence or status overstates both the code and the source, and it should be corrected by adding the note, the version, and the holder, or by withdrawing the claim the bare code was used to support.

Where is a bare code completed?

A bare code is completed in the Eurostat explanatory notes, for the version in use, and in the correspondence tables, where a revision is involved. The entry for a specific business is completed in the register that holds the record, with the national statistical institute or the business register named. The dataset version is completed in the dataset metadata. A code that cannot be completed in those places, because the version is not stated and the holder is not named, should be recorded as uncheckable, with the date of the check, rather than being completed from a finder tool, a vendor listing, or a guess about the version.

Copy the code, identify the version, open the note, and confirm a business entry with the institute or register that holds it.