How do you check a code in the UNSD structure?

Checking a code in the UNSD structure for ISIC Rev. 5, UN Statistics Division, means opening the official classification detail and tracing the code through section, division, group and class, reading the explanatory note at each step. The check confirms that the code exists in Rev. 5, that its title matches the activity described, and that no excludes list sends the activity elsewhere. A secondary site can help a reader orient, but the UNSD structure is the source to confirm against.

What does tracing a code level by level actually involve?

Start at the top and work down, rather than jumping straight to the finest level. Read the section, then the division, group, and class, pausing at each step to ask whether the activity described still fits the scope you have just read. A title alone can mislead at any of these levels, because short titles reuse familiar words that carry a narrower or broader meaning in the classification. The explanatory note is what settles the question, not the impression left by the title. If the note at the division level already points elsewhere, follow the pointer and start again from the branch it names. A secondary site may help you find which section to open first, but once you have a candidate code, confirm it on the UNSD detail page for ISIC Rev. 5, UN Statistics Division. If the two disagree, trust the UNSD structure, because a table built on a copied code inherits whatever error the copy contained.

What should you do when two candidate codes both look plausible?

Read both notes in full, side by side, and look first at the excludes lists. Often one note will name the very activity you are checking and send it to the other category, and that single line resolves a contest that the two titles left open. Where neither note decides the matter, look at the includes lists for the concrete activities each category names, and at any residual category nearby, which exists for activities that fit the broader level but do not fit any detailed category under it. Do not resolve a tie by picking the category whose title sounds more modern, more specific, or more familiar from an earlier revision. Rev. 4 habits are a frequent source of error here: a code correct under Rev. 4 may have been split, merged, or reworded in Rev. 5, so an old placement is evidence to investigate, not a shortcut. When the notes still leave genuine doubt, the defensible move is to report at the broader level that is clearly supported and to record the open question in your notes, so a later reader can see exactly where certainty ran out.

How do you record a check so another reader can repeat it?

A check that lives only in one person’s head cannot be audited, so write it down in a form that lets someone else retrace it. Keep four facts together with any code you carry forward: the version label, ISIC Rev. 5, UN Statistics Division; the level of the hierarchy you are quoting, whether section, division, group, or class; the reference period the figure covers; and the source you checked against. A division total and a class figure can share a word in their titles and still answer different questions, so the level is not decoration, it is part of the meaning. The reference period matters for a similar reason: revisions and national adoption dates can fall inside a series, and a figure that spans such a change needs a note saying so. A short methods note stating these four facts at the time of use turns a private check into a public one.

One checker, one code, four pauses

Picture a reader who finds a code quoted in a report and wants to reuse the figure. She opens the UNSD Rev. 5 detail, finds the section, and reads its scope. It fits. She moves to the division and reads its note, which narrows the scope but still fits. At the group level she pauses, because the excludes list names a neighbouring activity that resembles the one in the report, and she checks whether the report’s activity is the excluded one or the included one. It is the included one, so she continues to the class, reads its note, and finds the scope still fits. Only then does she copy the code into her own table, together with its title, its level, and the version label. The exercise takes minutes; correcting a published table that quoted an unchecked code takes far longer.

That habit is the safeguard: open the official structure, read the note at every level, follow any excludes pointer, and write down the version, the level, and the source before the code leaves your notes.

This guide explains statistical classification only, and it settles no individual case. It does not fix any reader’s code, principal activity, or registration, and anyone needing the code that applies to a particular record should confirm it with the national statistical institute or business register that holds that record.