What changed overall from Rev. 4 to Rev. 5?

The overall change from Rev. 4 to ISIC Rev. 5, UN Statistics Division, was a substantial restructuring rather than a light update, with new sections, new divisions and many revised definitions. The headline markers include 22 sections, 87 divisions, 258 groups and 463 classes in Rev. 5, a new treatment of information activities, and 18 new intermediation classes. A reader moving from Rev. 4 data to Rev. 5 data should therefore expect breaks in series, not a smooth continuation.

What does substantial restructuring mean in practice?

The opening paragraph chooses its words carefully: a substantial restructuring rather than a light update. A light update would tidy titles, correct notes, and add a few categories at the margins, leaving most series intact. Rev. 5 did more. It added sections, added divisions, revised many definitions, reworked the treatment of information activities, and created new classes, including 18 new intermediation classes. The headline totals, 22 sections, 87 divisions, 258 groups and 463 classes in Rev. 5, UN Statistics Division, summarise the scale of the result, but the word substantial describes the process: a revision that reached into every level of the hierarchy and changed how large fields are organised.

How did the treatment of information activities change?

Information activities receive special mention in the opening paragraph because they illustrate why a restructuring was needed at all. Under Rev. 4, activities concerned with producing, distributing, and handling information sat across headings shaped by older distinctions, such as the medium through which content travelled. As production and distribution converged, those distinctions became harder to apply consistently, and tables built on them became harder to interpret. Rev. 5 gives information activities a new treatment, organising them around the work involved rather than around the older dividing lines.

The lesson for readers is broader than one field. When a revision singles out a field for new treatment, it is usually because the old treatment was generating borderline cases faster than notes and guidance could resolve them. A new treatment does not settle every borderline case, and it creates fresh ones at its own edges, but it resets the framework so that coders and readers share a current logic. Anyone comparing information series across the revision therefore needs the correspondence material close at hand, because this is exactly the field where title matching across revisions is least reliable.

Where do breaks in series actually bite?

Breaks bite in three places, and recognising them helps a reader plan around them. First, they bite in long charts, where a jump at the revision point looks like sudden growth or decline. Second, they bite in growth rates, where a rate computed across the break mixes classification change with economic change. Third, they bite in rankings and shares, where a field that gained scope rises in a league table for definitional rather than economic reasons. In each case the error is the same: reading a classification event as an economic event.

The defences are also the same in each case. Footnote the break where it occurs. Compute growth rates within each revision segment, and describe cross-break movement in words rather than as a single rate. For shares and rankings, state whether the scope behind a share changed, and if it did, present the pre-break and post-break rankings separately. These defences do not restore a continuous series. They stop a discontinuous series from pretending to be one, which is the achievable and honest goal.

How should a reader pace their move from Rev. 4 to Rev. 5?

Moving from Rev. 4 to Rev. 5 is best treated as a project with stages, not as a relabelling exercise done in an afternoon. Begin with the overall structure: the new totals, the new sections, and the fields given new treatment. Then move to the levels a given piece of work actually uses, tracing correspondence rows for those categories only, rather than attempting to master all 463 classes at once. Then update methods notes, table headings, and chart footnotes to carry the Rev. 5 label and any breaks. Finally, check national implementation dates for the datasets in hand, since the international revision and national adoption do not arrive together.

This staged approach also explains why different readers experience the revision differently. A reader of broad section tables meets the restructuring mainly as new letters and a new section count. A reader of detailed class tables meets it as dozens of changed rows. Both are reading the same revision, at different depths, and both need documentation pitched at their depth. A single paragraph claiming the revision was minor, or major, without saying at which level, will mislead one of them.

This guide offers general statistical information and does not determine any reader’s own code, principal activity, or registration status. Where a particular record needs settling, the national statistical institute or business register holding that record should be consulted.