What is ISIC?

ISIC is the International Standard Industrial Classification of All Economic Activities, and ISIC Rev. 5, published by the UN Statistics Division, is its current revision. It gives statisticians a shared language for grouping economic activities so that output, employment and value added can be counted in comparable ways. A reader who meets ISIC in a dataset is therefore meeting a statistical tool, not a licence, a tax status or a business registration category.

What does the full name promise, and what does it leave out?

The full name does a lot of quiet work before a reader reaches any table. International signals that the structure is meant to travel, standard signals that its categories are fixed by agreement rather than improvised per dataset, and all economic activities signals breadth rather than a focus on one sector. Put together, those words describe a shared language for statisticians, and the first paragraph of this guide already names what that language counts: output, employment and value added, each in ways that can be compared. A shared language matters most when the people producing a figure and the people reading it never meet. The producer needs a category definition that will still mean the same thing when the table lands in another office, another year or another country, and the reader needs to trust that the label in the row is not a local nickname.

That promise has a boundary built into it. A statistical tool sorts activity for counting. It does not grant permission to trade, it does not settle a tax question, and it does not register a business. Those jobs belong to other bodies with other rules, and a reader who treats a row label as though it carried those powers will draw conclusions the table never offered. Keeping the promise and the boundary in mind at the same time is the whole discipline of reading ISIC well: use the label for what it was built to do, and do not load it with decisions it was never asked to make.

Why do familiar words behave differently inside a table?

Everyday language is the first trap for a new reader of industry statistics. A word that feels obvious in conversation can carry a narrower, technical definition inside ISIC Rev. 5, UN Statistics Division, and the definition governs the count, not the everyday sense. A label that looks like a job title may in fact describe a kind of activity, and a label that looks like a brand may in fact describe a scope of work. The difference only shows up when a figure is challenged. Someone asks what was included, what was left out, and why a neighbouring category was not used, and the answers have to come from the category scope, its place in the hierarchy and its history across revisions, not from what the word usually means at a dinner table.

How do level, period and source change a single figure?

Four details decide whether a figure can be trusted in a new context, and all four deserve a check before quotation. First, the version label: ISIC Rev. 5, UN Statistics Division, must be named where Rev. 5 is meant, because a bare number can belong to more than one revision. Second, the level: a division total and a class figure answer different questions, even when their titles share a word, so the level must travel with the number. Third, the reference period: revisions and national adoption dates can fall inside a series, which means a long run may straddle a change in definition. Fourth, the source: the classification behind the figure should be named, not guessed from the table title.

A short walkthrough keeps those checks in order:

1. Name the activity in words, as work being done, rather than starting from a remembered code or a job title.
2. Find the section and division in the UNSD Rev. 5 structure and test whether that broad grouping still fits.
3. Read the group and class notes, watching for an excludes list that redirects the activity elsewhere.
4. Record ISIC Rev. 5, UN Statistics Division, beside the result, so a later reader can retrace the same path.

What should stay attached to any figure you carry forward?

Carry the version label, the level and the source with the figure, as part of the figure, not as an afterthought in a footnote nobody reads. When a category looks borderline, go back to the UNSD structure rather than relying on memory, because memory preserves titles well and definitions badly. A number that can be retraced to a definition, a level and a named revision is useful to a stranger. A number without those attachments may look exact and still be impossible to check, and in statistics that difference matters more than the number of decimal places on display.

This guide explains a statistical classification and nothing beyond it. It settles no reader’s own code, no principal activity and no registration, and anyone needing a code for a particular record should ask the national statistical institute or business register that holds that record, since only the holder of the record can settle it.