What stayed the same in Rev. 5?

Much stayed the same in ISIC Rev. 5, UN Statistics Division, even amid the restructuring: the four level hierarchy, the activity basis of the classification, and the principle of grouping by kind of activity all carried over from Rev. 4. Continuity of that kind matters because it lets compilers and readers transfer much of their understanding from one revision to the next. What changed was mainly the content and arrangement of categories, not the basic logic of the system.

Why does continuity matter as much as change?

Revision guides naturally dwell on what changed, and this cluster does so at length. Continuity deserves equal attention, because it is what makes a revision usable. If every principle changed at once, no compiler could recode a register, no analyst could interpret a back series, and no reader could transfer any understanding from Rev. 4 to Rev. 5. The opening paragraph names three continuities that prevented that outcome: the four level hierarchy, the activity basis of the classification, and the principle of grouping by kind of activity. Together they mean that Rev. 5 is a new arrangement of a familiar system, not a new system altogether.

What does the four level hierarchy preserve?

Four levels, from broad section down through division and group to fine class, preserve a way of thinking as much as a way of counting. A reader can start broad, where the picture is simple and the totals are robust, and move finer only as far as the question and the data justify. A compiler can code at a fine level and aggregate upward with confidence, because the levels nest cleanly. A table can show several levels at once, letting different readers take what they need from the same publication.

Preserving the number of levels also preserves comparability of method, even where category contents changed. A Rev. 4 division total and a Rev. 5 division total are both division totals, produced by the same kind of reasoning, even when their scopes differ. That shared method is what lets a methods note explain a break in scope without also having to explain a break in logic. The reader must still check scope, category by category, but does not need to relearn what a level means.

What does grouping by kind of activity still require?

Grouping by kind of activity sounds abstract until it is put to work. It requires the classifier to ask what kind of work an organisation does, rather than what it sells, what its customers do, or what its premises look like. It requires that organisations doing similar kinds of work sit together, even when their outputs differ in appearance, and that organisations doing different kinds of work sit apart, even when their outputs look alike. Those requirements did not change in Rev. 5, and they remain the discipline that holds the whole structure together.

Because the principle carried over, so did its familiar difficulties. Mixed organisations still need a principled way of deciding which kind of work dominates their classification. Activities that straddle kinds still need notes and excludes lists to place them. New methods of production still blur old boundaries. Rev. 5 addresses some of those difficulties with new and revised categories, but it does not abolish them, and a reader who expects a revision to end borderline cases will be disappointed by every revision. The principle persists precisely because borderline cases persist, and a stable principle is what lets each new case be argued consistently.

How does continuity help a reader transfer understanding?

Transfer works best when it is explicit. A reader moving from Rev. 4 to Rev. 5 should carry over three things with confidence: the habit of reading from broad to fine, the habit of describing an activity in words before choosing a category, and the habit of checking explanatory notes and excludes lists. Those habits are revision-proof, because they belong to the system’s logic rather than to any single edition’s contents. A reader who carries them over will make fewer errors in Rev. 5 than a reader who memorised Rev. 4 categories and tries to reuse them unchanged.

What should not transfer unchanged is any specific mapping from a title to a category. Titles persist, scopes shift, and the continuity of logic makes that trap more tempting, not less: because the system feels familiar, the reader assumes its contents are familiar too. The safeguard is to let logic transfer and contents be rechecked, every time, category by category. Familiar logic plus rechecked contents is the combination that continuity is meant to enable.

This guide explains general classification principles only. It makes no finding about any reader’s code, principal activity, or registration, and for a specific record the confirming body is the national statistical institute or business register that holds it.