An affiliate programme looks like a link and a percentage. Underneath it is an accounting system that has to be right every time, because partners check their numbers and disagreements are expensive.
Attribution is the foundation
Each partner gets their own code. A visitor arriving through it carries that code through signup, where it attaches to the account permanently.
Three decisions shape everything downstream:
Attribution window. How long after clicking does a signup still count? Too short and partners feel cheated; too long and you pay for traffic that would have arrived anyway.
Last-click or first-click. If someone arrives through two partners, who earns it? Last-click is the common default and it under-rewards the partner who created the awareness.
What qualifies. A signup? A deposit? Sustained activity? This is the single most consequential choice, because it determines what behaviour the programme actually buys.
Revenue share beats bounties
A per-signup bounty rewards volume of signups. That is precisely the behaviour that fills your book with accounts that register once and never return.
Revenue share — a percentage of what referred users generate, for as long as they stay active — rewards quality instead. Partners with genuinely engaged audiences earn more from it, which is why the good ones prefer it.
Tiers that improve with volume give strong partners a reason to keep pushing, and cost nothing on partners who never reach them.
The calculation has to be reproducible
Six months from now a partner will ask why their October commission was what it was. The useful answer is a reconstruction from the recorded events, not a recalculation that happens to agree.
That means storing the inputs, not just the outputs: which accounts were referred, what activity they generated, what rate applied at the time, and what adjustments were made. Rates change, and a commission calculated under an old rate must stay calculated under it.
This is the part teams under-build and then cannot fix retroactively, because the data was never captured.
Fraud, and where it actually comes from
Self-referral is the most common: a partner signs up through their own link using new accounts. Device fingerprinting, payment-method matching and behavioural patterns catch most of it.
Cookie stuffing — dropping a referral code on visitors who never engaged with the partner — is harder to detect and shows up as an implausible click-to-conversion ratio.
The control that does the most work is unglamorous: a holding period before commission becomes payable. If a transaction reverses inside that window, the commission reverses with it, before money has left.
Payouts are an operational system
Partners across many countries want paying in different ways — bank transfer, stablecoin, local rails. Each is its own integration with its own failure modes and its own reconciliation.
What the payout layer needs: batching so you are not sending hundreds of individual transfers, a clear record of what each payment covered, retry handling for failures, and a statement the partner can check against their dashboard.
Pay reliably and visibly. Affiliates talk to each other, and a reputation for slow or disputed payouts cannot be fixed with a higher commission rate.
The dashboard is the retention mechanism
Live referred volume, commission earned and pending, per-link breakdown, payout history.
A partner who can see their numbers changing promotes harder than one waiting on a monthly email. It sounds like a minor feature. It is the main reason a good affiliate stays with one programme rather than moving to a competitor’s.
What we build
Tracking and attribution, the commission engine, partner dashboards, fraud detection, multi-rail payouts, and the admin console your team runs the programme from.
The plan itself — rates, tiers, what qualifies — is yours to design. We model it against real volumes first so you can see what it pays before you commit.
Common questions
How does an affiliate tracking system work?
Each partner gets a unique code or link. When a visitor arrives through it, the code is stored and persists through signup so it attaches to the account permanently. From then on, qualifying activity by that account generates commission for that partner. The hard part is not the tracking - it is deciding what counts as qualifying, and making the calculation reproduce the same answer months later when someone disputes it.
What commission structure works best for affiliates?
Revenue share for partners whose referred users stay active, because it rewards quality. Per-signup bounties reward volume and reliably fill your book with accounts that never return. Tiered structures that improve as a partner grows give your strongest affiliates a reason to keep promoting rather than plateauing, and cost you nothing on the partners who never reach the tier.
How do you prevent affiliate fraud?
Self-referral detection, velocity limits, cookie-stuffing checks, and a holding period before commission is payable. The most common abuse is a partner referring themselves through new accounts, which device and payment-method fingerprinting catches. The holding period matters most: it means a reversed transaction can claw back the commission before it has been paid out.
What does an affiliate need from a dashboard?
Live referred volume, commission earned and pending, a breakdown by link, and payout history with dates. Partners who can see their own numbers in real time promote considerably harder than ones waiting on a monthly email. The dashboard is not a nice-to-have - it is the main reason a good affiliate stays with one programme over another.
Does Corum8 build affiliate platforms?
Yes - tracking and attribution, commission calculation engines, partner dashboards, fraud detection and multi-rail payout automation, plus the admin console your team runs the programme from. We have built referral and affiliate infrastructure since 2016.
