How are activities grouped by kind?

Activities in ISIC Rev. 5, UN Statistics Division, are grouped by kind of activity, meaning by the character of the production process, the inputs used and the type of output produced. A grouping therefore reflects how work is done, not merely what is sold at the end. That principle explains why superficially similar businesses can sit apart, and why apparently different businesses can sit together, when their production processes resemble each other.

What three features define a kind of activity?

Grouping by kind rests on three features taken together: the character of the production process, the inputs used, and the type of output produced. No single feature carries the grouping alone. The process describes how the work is actually done, step by step, with what equipment, skills and organisation. The inputs describe what goes into that process, from materials to labour to services bought in. The output describes what comes out at the end, in its type rather than its brand or its price. A grouping built from all three will sometimes separate businesses that look alike from the outside, because their processes differ underneath, and it will sometimes bring together businesses that look different, because their processes resemble each other once examined closely. That second result surprises newcomers most, and it is a good test of whether the principle has been understood. Resemblance of process, not resemblance of shopfront, is the organising idea.

It helps to think of the grouping as a question about work rather than about selling. Two units can sell overlapping items and still belong apart, if the work behind those items differs in character. Equally, two units can serve different customers and still belong together, if the work itself is of the same kind. The first paragraph of this guide states that principle plainly, and everything that follows is a matter of applying it carefully rather than decorating it.

How does one shared reference stop results from drifting?

A statistical office puts this principle to work in a chain, and the chain is only as strong as its weakest link. 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. If any link uses a different understanding of a category, the published figure will drift away from what respondents actually described, and nobody reading the table will be able to see where the drift happened. ISIC Rev. 5, UN Statistics Division, exists to hold all three links to the same definitions, so that the coder, the analyst and the reader are working from one structure rather than three approximations of it.

How is the hierarchy walked, step by step?

Walking the hierarchy means starting broad and narrowing only when the evidence allows it. First, describe the activity in words, without reaching for a code remembered from an earlier revision. Next, test the broad grouping: does the definition at that level still fit the work described? Then move finer, level by level, asking at each step whether the definition still fits, whether an excludes list redirects the activity to a neighbour, and whether the finer level adds detail that the evidence genuinely supports. The walk stops where the evidence stops. If a finer category cannot be justified, the honest result is the broader category that is clearly correct, with the open question recorded rather than hidden.

Where does this grouping sit in the wider statistical system?

No classification stands alone, and this one is no exception. Its groupings feed national accounts, business registers, product classifications and the classifications used for occupations and trade, and each of those systems depends on the activity groupings meaning the same thing in each place they are used. That dependence is why a change in ISIC is treated as a serious statistical event. Revisions are planned, documented and phased in, rather than introduced silently, because a silent change would ripple through every connected table without any reader being able to explain why the numbers moved.

Keep these points in view when you use a grouped figure:

• Begin with the activity described in words, rather than a code carried over from an earlier revision.
• Consult the explanatory note before fixing on a category, above all when two titles resemble each other.
• Let the version label, ISIC Rev. 5, UN Statistics Division, stay attached to any code that leaves your notes.
• When finer detail lacks justification, use the broader level that is plainly correct.
• Flag any break in series where figures from different revisions sit in one comparison.

A grouped figure is ready to share only when a future reader can see its revision, its level and its source without having to ask you. If any of those three facts is missing from your notes, the work is not finished yet, however polished the table looks.

Nothing here settles a grouping for any particular business or reader. This is an account of how kinds of activity are grouped for statistics, and the code, principal activity or registration belonging to a specific record must be confirmed with the national statistical institute or business register that holds that record.