What is ISIC used for in statistics?

ISIC Rev. 5, UN Statistics Division, is used wherever statistics need to be broken down by industry: business surveys, employment counts, production indices and structural business statistics all lean on it. The classification lets compilers publish the same kind of table in different years and different countries and still have the rows mean comparable things. For a reader, that is the main practical value: an industry label in a table carries a defined meaning behind it.

Which statistical products actually lean on ISIC?

The opening of this guide names four uses, and each deserves a moment of its own, because they show how widely one classification reaches. Business surveys use ISIC to design questionnaires around kinds of activity, so that responses can be coded consistently. Employment counts use it to break a workforce down by industry, so that a jobs figure can say where work sits, not merely how much of it exists. Production indices use it to track output by industry over time, so that change can be located in particular kinds of work. Structural business statistics use it to describe the population of businesses by industry, giving a portrait of an economy’s composition rather than a single total. Different products, different questions, one shared set of category definitions holding them together.

That sharing is the practical value for a reader. Because the same classification sits behind all four products, an industry label encountered in one product carries a defined meaning that can be recognised in another. A reader can move from a survey table to an employment table and know that the row label, if the version and level match, refers to the same defined category in both places. Without a shared classification, each product would invent its own industry language, and cross-product reading would become guesswork.

Why does the same table shape matter across years and countries?

Compilers publish the same kind of table in different years and different countries, and the rows should mean comparable things each time. That comparability is earned, not automatic. It requires the same version, the same level and the same definitions, carried visibly with the figures. ISIC Rev. 5, UN Statistics Division, supplies the common structure, but the compiler and the reader must both do their part: the compiler by labelling version and level, the reader by checking those labels before comparing. A table from one year and a table from another year can look identical in layout while resting on different revisions, and layout similarity is no guarantee of comparability.

How does the classification travel from survey to published table?

Inside a statistical office, the journey runs through three stages that must agree. The survey questionnaire asks about activity in words. The coding manual turns those words into categories. The published table presents the results under category titles. ISIC Rev. 5, UN Statistics Division, provides the shared reference for all three stages, so the coder, the analyst and the reader work from the same structure. The system works best when the version label travels with the data at every stage, from the first coding decision to the final chart, because a lost label is how cross-revision errors enter an otherwise careful analysis. A reader sees only the final stage, but the final stage is trustworthy only because the earlier stages used the same definitions.

What does the analyst’s desk show about everyday use?

Picture an analyst preparing a table for readers who never collected the data. The analyst writes the activity in plain words first, then traces it through the UNSD Rev. 5 structure, reading each note, and only then records a category. A colleague later asks why a neighbouring category was not chosen, and the analyst can point to the excludes list that decided the matter. The table is published with its version label visible, so a reader in another country can check the same structure and reach the same understanding. That chain, from words to notes to a labelled result, is the whole method in miniature, and it repeats across every product named at the start of this guide: surveys, employment counts, production indices and structural business statistics all depend on the same chain holding.

Close any use of such a table by writing down, in one sentence, which version you used and at what level. Kept with the table or chart, that sentence is the simplest protection against the most common classification error, which is to compare figures whose labels matched while their definitions did not.

This guide covers statistical uses only. It assigns no code, principal activity or registration to any reader, and a code for a specific record is a matter for the national statistical institute or business register that holds the record.