Upline and Downline in MLM

Upline and downline are the two easiest terms in direct selling. Sponsor and placement look like synonyms for them and are not — and the difference is the most consequential data modelling decision a launching company makes.

The easy part

Upline — everyone above you in the structure. The position immediately above you, the one above that, up to the top of the tree.

Downline — everyone below you. Your frontline, their frontline, and so on down.

Leg — one branch of your downline, viewed from you. A binary distributor has two legs; a unilevel distributor has as many legs as they have frontline positions.

Level — how many steps a position sits below you. Your frontline is level 1, their frontline is level 2.

Commission generally flows up: volume generated by a downline position pays the uplines above it, to the plan’s paid depth, subject to qualification.

The part that gets conflated

Two different facts are true about every position in the tree, and they answer two different questions.

In a unilevel these coincide, because everyone you sponsor goes on your frontline. That is why the mistake is so easy to make: a company launching with a unilevel can store one field and appear to be fine.

In a binary, the frontline holds two positions. Sponsor a third person and they must be placed further down a leg — under someone who did not recruit them. That is spillover. In a forced matrix, the same thing happens because width is fixed.

So the recruit is sponsored by you and placed under someone else. Both relationships are real, both drive different money, and both get asked about.

What happens when you store one field for both

The failure is not immediate, which is what makes it expensive.

Month 3. Everything looks fine. Nobody notices.

Month 9. A distributor asks why they did not receive a matching bonus on someone they personally recruited. The system has that person placed under a different distributor, so the sponsorship relationship no longer exists anywhere in the data. Support reconstructs it from enrollment emails.

Month 18. A rank promotion depends on a count of personally sponsored active distributors. The count is wrong for anyone whose recruits were spilled. Nobody can say by how much, because the input is gone.

Month 24. A dispute about a commission run from last year requires knowing who was sponsored by whom at that time. The answer is not recoverable.

None of this shows up in testing. It shows up as an unfixable historical problem, because the data was never captured in the first place.

What the genealogy model has to hold

Both relationships, separately. Sponsor and placement as distinct stored fields on the position.

Timestamps and attribution on placement. Who placed this position, when, and under which rule. Spillover placements are made by the system and that decision has to be recorded, not recomputed later — recomputing gives you today’s answer to a question about last March.

Full change history. If a placement is ever moved, the previous state, the reason and the approver are part of the record. Closed periods are never recalculated against the new structure.

Deterministic fill order. “Next available position” must be one rule, applied identically every time, and reproducible when two enrollments arrive in the same second.

The tree as it stood at each run. This is the requirement people miss. A commission run is a calculation against a structure at a point in time. Compression, in particular, makes the effective tree different every period, because it depends on who qualified in that period.

If your engine recalculates historical runs against the current genealogy, you do not have reproducible closed periods — and every historical question becomes a negotiation. This is the property described in commission software, and it is the main thing that separates a platform from a spreadsheet.

Compression, briefly

Compression closes the gap left by an inactive or unqualified distributor, so the positions below them move up for commission purposes and the qualified upline is not penalised for someone else’s inactivity.

Dynamic compression is evaluated per run, against that period’s qualification state. The same genealogy therefore compresses differently from month to month, and the compressed view belongs to the run rather than to the account. Store it with the run or you cannot answer why a level paid differently.

For the field, not the database

Distributors experience upline and downline as relationships rather than as a data structure, and two practical points follow from the mechanics above.

Your upline earns on your volume, which means their support is compensated rather than charitable — a reasonable thing to understand rather than to be cynical about.

And spillover is not a business plan. Positions only spill into a leg someone is actively building. A distributor waiting for their weak leg to fill itself will wait a long time. The plans that market spillover as a substitute for selling tend to attract people who do not sell, which is a field-quality problem before it is anything else.