Why Pyramid Schemes Fail

General information, not legal advice. Whether a specific arrangement is unlawful in a specific market is a question for a lawyer practising there.

A pyramid scheme does not fail because it was managed badly, marketed badly or unlucky with timing. It fails because of arithmetic that is fixed before the first participant joins.

The only money in the structure comes from participants, and each level needs more participants than the level above it.

The arithmetic

Take the common shape: to be paid, each participant must recruit six people, each of whom must recruit six.

Twelve levels exceeds a third of the human population. Thirteen exceeds all of it.

Nobody has to reach level thirteen for the structure to fail. It fails at the point where the required recruitment exceeds the population that can plausibly be reached — which, once you account for the fact that recruitment happens inside social networks rather than across the whole world, arrives far earlier than the table suggests. Most schemes exhaust their reachable population somewhere in the single-digit levels.

Change six to three and the same wall arrives a few levels later. Change the requirement to two and it still arrives. The multiple sets the date, not the outcome.

Why the collapse is sudden

This is the part participants consistently describe as happening without warning, and there are three reasons for it.

Payments come from current inflow, not reserves. There is no buffer. The money arriving this week pays the people owed this week. When inflow drops below what is owed, payments do not shrink proportionally — they stop.

Success is the recruitment mechanism. Every early participant who genuinely got paid is evidence, offered in good faith, that the thing works. So growth accelerates until it doesn’t, and the steepest growth is immediately before the wall.

The first failure is self-reinforcing. A missed payment converts recruiters into claimants within days. Recruitment stops exactly when the structure most needs it, and the remaining inflow disappears. The curve up is smooth; the end is a cliff.

Who loses

At the moment of collapse, most participants have paid in and not recovered — because in any pyramid the bottom level is, by construction, larger than everything above it.

That distributes losses by join date rather than by judgement. Somebody who joined in month nine of a scheme that failed in month ten loses everything. Somebody with identical reasoning who joined in month one profited. Nothing separates their decisions except sequence.

This is also why “you just had to get in early” is not a defence of the model. It is a precise description of what is wrong with it: the returns of the early participants are the losses of the later ones.

Four things that look like growth and are not

For anyone assessing something in progress:

1. Accelerating enrolment with flat product movement. Growth in participants without growth in units going to non-participants is the exhaustion curve, not a sales curve.
2. Reorders that track qualification deadlines. Purchasing that clusters just before a period closes is people holding a rank, not customers buying a product.
3. Rising average enrolment cost. If the entry price, the required starter package or the required monthly minimum keeps increasing, the structure needs more per head because it cannot get more heads.
4. New markets opened faster than the last one was established. Geographic expansion is the standard response to an exhausted population. It resets the recruitment pool and postpones the arithmetic without changing it.

On “current pyramid schemes”

We do not publish a list, for the same reasons we do not publish rankings of companies: we cannot substantiate a claim that a specific named company is operating unlawfully, we build software for this industry and are therefore not a neutral party, and any such list would be out of date and would carry real consequences for anyone it named wrongly.

What exists instead are primary sources, which are better than a list because they are current and attributable:

• United States — Federal Trade Commission enforcement actions and press releases; state attorney general announcements; SEC actions where an investment element is alleged; court records and filed complaints.
• South Africa — the National Consumer Commission; the Financial Sector Conduct Authority’s warnings, which cover the unregistered-financial-services version of the same problem; reported judgments.

MLM schemes explained sets out how to use each of these, and what to read in an income disclosure statement.

The one thing to avoid is treating the absence of an enforcement action as a clearance. Enforcement is slow, and by definition it arrives after the losses.

What separates this from a legitimate plan

One structural fact, and it is not the diagram.

Legitimate plans have trees. Legitimate plans pay on multiple levels. Legitimate plans reward recruitment indirectly, because a recruit who sells generates volume. None of that is the distinguishing feature.

The distinguishing feature is where the money comes from. If the payout is funded by margin on products sold to people who are not participants, there is an external revenue source and the structure does not depend on recruitment continuing. If the payout is funded by participant money, it does.

Which means the practical test is a records test: can the company state what proportion of its revenue came from non-participants, and produce the order-level classification behind the number? A company that cannot has not demonstrated that it is a pyramid — it has demonstrated that it cannot show it is not, which is a position it chose by not recording the data.

MLM versus pyramid scheme covers the legal framing in detail.

What a company builder should take from this

Three design consequences, all of which are ordinary engineering decisions:

Classify every order at the point of sale. Retail customer, subscribed customer, distributor purchase. This is the evidence for the only question that matters, and it cannot be reconstructed from a table that never had the column.

Bound the payout and fund it from margin. Model the total payout as a proportion of commissionable volume across every component, including the ones that only pay occasionally, and check that it sits inside what the product margin can support in the worst case rather than the average one. The plan calculator exists for this.

Make recruitment worthless on its own. If a position can be paid without product moving to someone outside the plan, the arithmetic above is available to your plan too — not as a legal question first, but as a business one. A revenue line that only exists while recruitment grows is a revenue line with an expiry date.

MLM rules and regulations covers the operational side of getting all three recorded properly.