Why do classifications need revisions?

Economies change, and a classification that never changed would slowly stop describing the economy it is meant to measure, which is why ISIC Rev. 5, UN Statistics Division, replaced Rev. 4. New activities appear, old activities shrink or merge, and production methods shift, so categories, definitions and boundaries need periodic review. A revision is therefore a maintenance task for statistics, not a verdict on the previous revision, which served its own period.

Why would a good classification ever need to change?

A classification that never changed would slowly stop describing the economy it is meant to measure, and that slow drift is more dangerous than a visible revision, because it is invisible. New activities appear, and a structure with no place for them must either force them into ill fitting old categories or leave them uncounted. Old activities shrink or merge, and categories built for a once large kind of work begin to hold a residue that no longer resembles their titles. Production methods shift, so that work once done one way is now done another, and the old boundary between two categories no longer tracks the real difference in how work is done. ISIC Rev. 5, UN Statistics Division, replaced Rev. 4 for those reasons, and the replacement should be read as maintenance, not as a verdict. Rev. 4 served its own period, and Rev. 5 serves a changed economy. A future revision, when it comes, will do the same for its own period.

How does a reader work safely across a revision boundary?

Working across a revision means treating the boundary as a fact to be recorded, not as an inconvenience to be smoothed over. Start from the activity described in words, then move through the hierarchy from broad to fine, checking each level against the UNSD structure for ISIC Rev. 5, UN Statistics Division. For classifications need revisions, at each level, consider whether the definition still fits, whether an excludes list redirects the activity, and whether a finer level adds meaningful detail. For classifications need revisions, an unclear answer at any level means staying at the broader level that is clearly correct and recording the question, rather than picking a fine category that cannot be justified from the evidence available. The broader level is particularly valuable across a revision, because broad groupings often survive a revision intact even when fine categories move, and a broad comparison across the boundary may remain valid where a fine comparison would not.

Why does a revision ripple through the whole statistical system?

ISIC does not stand alone, so a revision to it cannot stay contained. It connects to national accounts, to business registers, to product classifications and to the classifications used for occupations and trade, and its value grows when those connections are kept clean. A change in ISIC therefore ripples into the tables that use it, which is why revisions are planned, documented and phased in, rather than introduced silently. National offices need time to update questionnaires, coding manuals and registers. Compilers need time to map old categories to new ones and to document breaks. Readers need time to learn new definitions before they are asked to interpret new figures. A revision introduced silently would deny all three groups that time, and the resulting tables would be uninterpretable for a period nobody could date.

What routine keeps a revision from becoming a misreading?

A short routine, applied whenever figures from different revisions appear together, prevents most revision errors. Name the activity in words first, find the section and division in the UNSD Rev. 5 structure, read the group and class notes, and record ISIC Rev. 5, UN Statistics Division, beside the result:

Step	What to do	What it confirms
1	Word the activity plainly	An activity is being classified, not a product or a job title
2	Find section and division in the UNSD Rev. 5 structure	The broad grouping is a fit
3	Go through the group and class notes	No excludes entry redirects the activity
4	Keep ISIC Rev. 5, UN Statistics Division, beside the result	The version remains checkable later

Do not skip the note reading step, because that step is where borderline cases are decided, and revision boundaries create borderline cases in abundance: activities that sat comfortably in one category under Rev. 4 may sit astride two under Rev. 5, or may have moved entirely. Reading the new notes, rather than trusting the old memory, is what lets a reader land on the right side of the new boundary.

Remember, finally, what kind of document this guide is. It describes a statistical system and its maintenance, not a decision about any reader, and for a decision about a specific record, the national statistical institute or business register that holds that record is the authority to consult, since this guide determines no reader’s code, principal activity or registration.