MLM Rules and Regulations: What a Company Has to Get Right

This is general information about where the rules sit and what they require operationally. It is not legal advice. Direct selling regulation differs by country and, in the United States, by state. Take advice from a lawyer who practises in your markets before you launch anything.

There is no single MLM statute in most markets, which is why “is it legal” is the wrong shape of question. Six separate regulatory surfaces apply at once, each with a different regulator and a different failure mode.

Most of them are decided by what the software records at the moment of a sale, not by what a policy document says.

The six surfaces

1. Income representation

Any statement about what participants earn is a claim that has to be supportable. In practice three things follow:

• An income disclosure statement, published, current, and available before somebody enrols rather than after.
• The median rather than the mean, and a denominator that includes everybody who enrolled, not only those who qualified for something.
• Generated, not compiled. The figures should come out of the commission data by a repeatable process. A disclosure assembled by hand in a spreadsheet each year is a document nobody can reproduce, which is the wrong property for the one document whose accuracy is most likely to be examined.

The recurring failure is not a fraudulent number. It is a lifestyle-shaped implication — a slide showing a hypothetical organisation and the arithmetic of what it would pay, presented without the context that almost nobody builds it.

2. Product claims

A claim about a health product is regulated speech in nearly every market, and in this model it is made by thousands of people the company does not employ.

The control that works is an approved-claims library framed as a permission: here is what you may say about this product, in this market, with the evidence behind it. A prohibition list asks people to remember what not to say, which fails at scale. A permission list gives them something to use, which is what they actually wanted.

Two mechanical requirements follow: claims are per market, because clearance differs, and the library has to be versioned, so what was approved at the time somebody said it is recoverable.

3. Source of revenue

This is the question behind pyramid-scheme law everywhere, and it reduces to one thing: does the company sell meaningfully to people who are not participants?

Answering it requires classification at the point of sale — retail customer, subscribed or preferred customer, distributor purchase. That record is the evidence, and it is the one record that cannot be built later. An orders table without the column produces a company that genuinely cannot answer a question about its own history.

Two supporting mechanics belong in the same conversation:

• Inventory loading controls. Ordering patterns that suggest purchasing to qualify rather than to sell should be visible in reporting, not discovered in a complaint.
• A real buyback and return policy, applied by the system rather than by discretion — including for somebody leaving with unsold stock.

MLM versus pyramid scheme covers the legal distinction, and why pyramid schemes fail covers the arithmetic underneath it.

4. Contractor classification

Participants are independent contractors. Whether that holds is assessed on substance, and the substance includes the software.

Features that look like supervision become evidence of supervision: scheduling, shift assignment, time tracking, mandated activity, upline approval of what somebody sells. Well-built distributor tools deliberately omit them. What they do contain is reporting, training material and recognition — support without direction.

The distinction is not cosmetic. A platform that lets an upline assign work to a downline has introduced a management relationship into a structure whose entire legal basis is that no such relationship exists.

5. Data protection

A direct selling company holds an unusual quantity of personal data about people who are neither employees nor ordinary customers, and the genealogy makes relationships between them explicit and permanent.

The operational requirements are the same everywhere even though the statutes differ:

• a stated lawful basis for each processing purpose,
• consent recorded as an event with timestamp, version and scope — not a boolean that gets overwritten,
• retention limits that actually delete, because accumulation is the exposure,
• subject access, correction and deletion that can be executed without breaking the tree,
• cross-border transfer handled deliberately once you operate in more than one market.

In the United States this varies by state. In South Africa it is POPIA, and the deletion requirement genuinely conflicts with a naive “keep everything” genealogy design, which is why it belongs in the schema conversation rather than the policy conversation.

6. Payments, tax and consumer rights

• Contractor tax records per market, produced from the same data that produced the payment.
• Cooling-off rights on enrolment and on orders where local law provides them, enforced by the system rather than granted on request.
• Returns and buyback, with commission reversed through the same tree that paid it, including when the wallet has to go negative to recover it.
• Payment verification before money moves. A commission run that pays a wrong amount is a recovery problem across thousands of individuals, which is a materially worse problem than a delayed run.

Working checklists

At launch, before the first enrolment

• Plan document versioned, dated, and stored as the artefact the software runs from
• Order classification implemented at checkout, immutable afterwards
• Commissionable volume held per product, separately from price, tax and shipping
• Sponsorship and placement stored as two distinct relationships
• Distributor agreement with a versioned acceptance record per person
• Income disclosure process defined, generated from commission data
• Approved-claims library, per market, versioned
• Return, buyback and cooling-off rules implemented in the system
• Retention schedule defined per data category, with deletion that runs
• Product availability enforced on ordering and on volume, per market

Each commission period

• Run reconciles: total paid equals the sum of components, and the payout ratio is within the modelled band
• Rule set version stored with the run, so a closed period reproduces exactly
• Exceptions reviewed before payment, not after — caps hit, negative wallets, unusual jumps
• Reversals from returns applied through the tree that paid them
• Payment file verified against the run totals before release

Monthly

• Non-participant revenue proportion calculated and recorded
• Ordering-pattern review for purchasing that looks like qualification rather than sale
• Field material sampled for income and product claims
• Opt-outs and suppression confirmed to be honoured across every channel
• Complaints and chargebacks reviewed for a pattern rather than case by case

Per new market

• Local counsel engaged before any enrolment is accepted
• Product clearance per product, and availability enforced both ways
• Plan thresholds and settlement currency decided deliberately
• Claims library localised, not translated
• Data transfer basis documented
• Tax and contractor reporting confirmed against local requirements

If you are checking a company rather than running one

Four questions, and the manner of the answer is itself informative:

1. What proportion of revenue comes from customers who are not participants, and how is it measured? A company that tracks this classifies its orders properly. A refusal is an answer.
2. Where is the income disclosure, and what is the median? Not the top ranks. The median across everybody who enrolled.
3. What is the return and buyback policy for somebody leaving with stock? And is it applied by the system or by request.
4. Are the product claims specific and sourced? “Supports normal immune function, per the ingredient dossier” is a claim somebody stands behind. A testimonial is not a claim about the product.

MLM schemes explained sets out the public sources for checking a specific company in the US and South Africa, and how to start an MLM company covers the sequence in which these decisions have to be made.