What changed at group and class level?

At group and class level, ISIC Rev. 5, UN Statistics Division, contains 258 groups and 463 classes, with new classes for activities that had grown since Rev. 4, including 18 new intermediation classes. Group and class changes are where most day to day code work feels the revision, because detailed tables and datasets operate at those levels. A methods note that only mentions section changes will therefore miss the changes that affect most published detail.

Why do groups and classes carry the weight of the revision?

Sections and divisions make the headlines, but groups and classes do the daily work. Detailed tables, survey coding frames, and business register records mostly operate at these finer levels, because this is where distinctions become specific enough to act on. A reader who follows only the broad levels will understand the shape of Rev. 5 and miss almost everything that changes a published figure. The opening paragraph makes this point directly: a methods note that mentions sections alone misses the changes affecting most published detail.

What does a group change look like from a table’s point of view?

A group gathers classes that belong together, so a group change can mean several things: classes moved in or out, a group split into two, two groups merged, or a group renamed with its scope adjusted. From a table’s point of view, the visible symptom is usually a total that no longer reconciles with its predecessor, or a set of class rows that no longer add to the group row a reader expects. Both symptoms point to the same need, which is to check the group definition and the correspondence rows before interpreting the movement as economic change.

Group level reading also rewards attention to neighbours. A class that moved groups will appear as a fall in one group and a rise in another, and a reader who examines only one side of that move will tell half the story. Tracing a moved class in both directions, out of its old group and into its new one, is a small discipline that prevents a large class of misreadings. Tables that show group and class detail together make this tracing possible. Tables that show groups alone hide it.

How should the 18 intermediation classes be read at class level?

Within the class total of 463, the 18 new intermediation classes stand out because they represent a field given its own fine detail for the first time. At class level, intermediation is not a single heading but a set of distinctions, and that matters for anyone working with detailed data. A coder assigning a class code now has a specific home for intermediation work, instead of choosing among headings that fit it only approximately. An analyst reading a detailed table can see the field’s internal shape, rather than a single lumped figure.

The presence of 18 new classes also warns against treating the class count as a simple growth story. Some of the 463 classes are new, some are revised, some are carried over, and some Rev. 4 classes were retired or merged. The net total, like the division total, hides movement in both directions. A class level methods note should therefore name the kinds of change affecting the classes its tables actually use, instead of quoting 463 as if it described every class.

A worked pass through a detailed table

Imagine an analyst opening a detailed Rev. 5 table for the first time, with its groups and classes laid out beneath a familiar division heading. The first pass should be structural, not numerical: which groups appear that did not appear before, which class rows sit under different groups, and which headings carry new explanatory notes? The second pass is the correspondence pass, matching each unfamiliar row to its Rev. 4 origin and marking splits, mergers, and retirements. Only the third pass is numerical, comparing values where the correspondence allows it and footnoting breaks where it does not.

This order feels slow, and it is the order that prevents the most common detailed-table error, which is to compare class rows by title alone. Titles recur across revisions with changed scope, new classes carry titles that sound like old headings, and retired classes leave gaps that a hurried reader fills by assumption. The three-pass habit, structure first, correspondence second, numbers third, turns a detailed table from a trap into a tool.

This guide is general information about statistical categories only. It does not assign a code to any reader, identify anyone’s principal activity, or settle a registration matter; for a specific record, the national statistical institute or business register that holds it is the authority to confirm.