Multi-Level Affiliate Commissions Explained

Single-level affiliate programs are easy to reason about: a partner sends a click, the click converts, the partner earns a percentage. Multi-level programs change the question. Now a partner also earns when the people they recruited convert, and sometimes when those recruits' recruits convert. The payout is no longer a flat rate on one sale. It is a split distributed across a chain of partners who each contributed, directly or indirectly, to the conversion. Get the model wrong and you either underpay your best recruiters and watch them leave, or you create an uncapped liability that compounds with every new signup. This piece walks through how commissions flow up a referral tree, how level splits are defined, and the rules that keep the math both fair and survivable.

The referral tree is the source of truth

Every affiliate has exactly one parent: the affiliate who recruited them. Follow those parent links upward and you get a path from any affiliate to the root of the program. That path is what a multi-level commission walks. When a conversion lands on an affiliate, the commission engine reads the ancestry of that affiliate and pays out level by level until it runs out of eligible ancestors or hits the configured depth limit. Argus Grape stores this ancestry as a PostgreSQL ltree path, so resolving the full upline for a conversion is a single indexed query rather than a recursive crawl. The affiliate who owned the converting click is level 0. Their recruiter is level 1. The recruiter's recruiter is level 2, and so on. The tree, not the individual link, defines who gets paid.

What a level split actually is

A level split is the schedule that says how much each tier of the upline earns on a conversion. It is expressed per level, and each level is computed against the conversion value, not against the level below it. A typical three-level schedule on a 100 dollar conversion might look like this:

• Level 0, the direct affiliate who drove the sale, earns 10 percent, or 10 dollars.
• Level 1, the affiliate who recruited them, earns 4 percent, or 4 dollars.
• Level 2, one step further up, earns 1 percent, or 1 dollar.
• Levels 3 and above earn nothing, because the schedule defines only three paying tiers.

Total payout on that conversion is 15 dollars across three people. Note that the percentages are independent: level 1 is not 4 percent of level 0's commission, it is 4 percent of the conversion. Computing each level off the base value keeps the rates legible to partners and keeps your effective commission rate, the sum of all active levels, easy to cap and forecast.

Why commissions flow up, not down

The economic logic is recruitment. A multi-level program pays you for the value of the network you built, not just the clicks you personally sent. When your recruit closes a sale, you contributed to that sale by bringing them into the program, so you earn a smaller share of it. This is also why depth limits matter. Without a cap, a conversion deep in a large tree could pay dozens of ancestors, and your effective rate would balloon with the height of the tree. A fixed schedule of two to four paying levels keeps the total commission bounded and predictable no matter how deep the tree grows.

The first level pays for the sale. Every level above it pays for the recruiter who made that sale possible.

The rules that keep it honest

Flowing money up a tree is an attractive target for abuse, so the commission engine has to be defensive. A few non-negotiable rules: ancestry is resolved at conversion time, not signup time, so backdated tree edits cannot rewrite who gets paid on past sales. Shadow-banned affiliates are skipped silently in the upline walk rather than removed, so a fraudster cannot tell their commissions stopped flowing. And conversions are only counted once, enforced by idempotent, signed webhooks, so a replayed event cannot pay the entire upline twice. The combination matters more than any single rule. A level split decides the shares, the referral tree decides the recipients, and the fraud controls decide which recipients are real. Skip the last layer and a multi-level program turns into a payout pump the moment someone scripts fake recruits beneath themselves.

Modeling it before you ship it

Before launching a multi-level schedule, compute your worst-case effective rate by summing every paying level, then multiply by your expected conversion volume to size the liability. If the total scares you, reduce levels or rates rather than capping payouts after the fact, which erodes partner trust. Our pricing page breaks down which commission features and tree depths are available per plan, and the docs cover wiring the conversion webhooks that feed the engine.