What Do Intermediation Classes Not Decide?

In ISIC Rev. 5, published by the UN Statistics Division, an intermediation class is a statistical label for a kind of activity. It describes what arranging work looks like in aggregate statistics. It does not register a business, grant a licence, set a tax rate, or settle whether a particular platform is an agent, an employer, or a supplier in law.

Can a statistical class grant a licence or settle a legal status?

No, and the reason is structural. A statistical class is a label used to group similar work for counting and comparison. It is not a permission slip, a contract term, or a court ruling. A marketplace that falls into an intermediation class for statistical purposes has not thereby gained a licence to trade, nor has it settled whether it counts as an agent, an employer, or a supplier under the law of any particular country. Regulators, courts, and tax authorities use their own definitions, built for their own purposes, and those definitions can divide the world differently from a statistical classification without either side being wrong.

This matters most when a business is under scrutiny. A platform facing questions about worker status, consumer protection, or tax treatment cannot answer those questions by pointing at its statistical class. The answers will turn on contracts, control over workers, terms offered to customers, and national law, all of which sit outside the classification. The class can inform the discussion, by describing the arranging work in neutral terms, but it cannot close it.

What legal and tax questions stay outside the classification?

Employment and liability questions sit equally far outside. Whether a driver, courier, carer, or freelancer counts as an employee, a contractor, or an agency worker depends on working arrangements and law, not on how statisticians group the platform that matched them with work. Similarly, who is liable when a service goes wrong, who owes a duty of care, and who must refund a customer are legal questions answered from contracts and statutes. Two businesses in the same statistical class can therefore face very different legal answers, because their contracts and control arrangements differ even though their arranging work looks similar to a statistician.

Tax treatment follows the same logic. Value added tax, income tax, and platform reporting rules use tax definitions, not statistical headings, to decide what is taxable, who must collect tax, and what must be reported. A business that assumes its statistical class settles its tax position risks a costly error. Professional advice in its own country, not a classification guide, is the right source for those questions.

Why can two businesses share a class yet differ in almost everything else?

Because a class groups by activity, not by identity. Size, legal form, ownership, funding, business model, and customer base can all vary inside one class. A tiny local agency and a large multinational platform can both arrange the same kind of transaction, and statistics will group their arranging work together so the total scale of arranging can be measured. That grouping is useful precisely because it ignores differences that do not change the kind of work done. It is not a claim that the two businesses are the same in law, in risk, or in quality.

How should readers use the class without overclaiming from it?

Use it as shared language for describing work, comparing sectors, and reading statistics. A reader who wants to know how large arranging work is in a market, how it has grown, or how one country’s arranging sector compares with another’s will find the class genuinely helpful. ISIC Rev. 5, published by the UN Statistics Division, with 22 sections from A to V, 87 divisions, 258 groups, and 463 classes, endorsed by the UN Statistical Commission in March 2023, exists to make those comparisons possible across countries and over time. Its value is in that comparative job, not in settling individual legal disputes.

What does a marketplace example show about the limit?

Imagine a marketplace that wonders whether it is an intermediary for tax or consumer law purposes in its own country. Its statistical class can describe its arranging work clearly, and that description may help its advisers frame the legal question. But the legal answer will come from tax law and consumer law, applied to its actual contracts and practices, not from the classification. If the marketplace changes its contracts, for example by taking more control over pricing or delivery, its legal position may change even though its statistical arranging work looks much the same. That divergence is normal, and it is why guides like this one stress their limits rather than hiding them.

This guide therefore ends where it began, with a boundary. It is statistical classification only, and it decides no reader’s code, no principal activity, no registration, and no legal or tax question, and any formal code question must be confirmed with the national statistical institute or business register that holds the record.