The payment model is not an accounting formality but a way to distribute risk. Who pays if the cohort turns out worse than expected is decided right here.
CPA: risk sits with the operator
A fixed amount for a player who reaches the deposit. The team gets the money right after the hold and after that does not depend on how the player behaves. The operator pays up front for a promise of quality.
When it fits: a quick test of a new geo, clear economics, a predictable spend is needed. Where it breaks: the team has no reason to care what happens to the player on day thirty. Without a qualification threshold and clawback, CPA is an invitation to incentivized traffic.
RevShare: risk sits with the team
A share of net revenue from the players brought in. The team earns only if the player plays, so its motivation matches the operator's interest. In exchange the team funds media buying for months.
When it fits: the operator has strong retention and reliable payment methods, and the partnership horizon is long. Where it breaks: on negative carryover and on opaque NGR. If bonuses, payment processing fees, taxes and an "administrative fee" are deducted from revenue, the real share turns out to be half of what was stated.
Hybrid: risk split in half
A fixed fee per player plus a reduced share of revenue. The fixed part is enough for the team not to stop buying, and the share makes it think about quality. The operator pays less up front than under pure CPA.
When it fits: almost always, if both sides are working for the long term. Where it breaks: harder to count and reconcile — two payment flows instead of one, and both have to be closed on time.
How to choose
- New geo, no data yet. CPA with a strict qualification threshold and a short hold: you see fast whether the campaign setup works or not.
- Proven team, known cohort. Hybrid: a below-market fixed fee plus a share — both sides care about D90.
- Strong product, small budget. RevShare with no negative carryover: you pay out of what was earned.
- Seasonal launch. CPA, because RevShare won't have time to ramp up before the season ends.
What to agree on in any model
Whatever the scheme, four things are fixed in the annex: what counts as a qualified player, the hold and the clawback window, exactly what is deducted from NGR, and whose data export is primary in case of a discrepancy. Without those four lines a dispute over money is a matter of time, not of probability.
Our terms for all three models are on the page “Terms”, payout rates and available volume by geo — in the section “Geos and models”.
