10 min read
Key takeaways
- The headline revenue-share percentage is not the cost of the deal.
- Negative carryover decides who absorbs the months when refunds and discount costs exceed revenue.
- A CPA deal can cost more than revenue share once day-30 (D30) quality is in the table.
- A hybrid deal without a written floor is two deals arguing.
- Value each partner on net revenue after quality filters, not on last month's new customers.
How do revenue share, CPA and hybrid deals compare?
CPA buys the new customer and leaves you the tail. Revenue share rents the tail to the partner. A hybrid tries to cap your downside in month one while still paying for the tail. All three can be rational. None of them are if the customer was only there to use a discount.
| CPA €45 | Revenue share 30% | Hybrid €20 + 20% | |
|---|---|---|---|
| Month 1 net revenue per customer | €28 | €28 | €28 |
| Month 1 cost | €45 | €8.40 | €25.60 |
| Months 2–6 net revenue | €55 | €55 | €55 |
| Months 2–6 cost | €0 | €16.50 | €11 |
| Total cost | €45 | €24.90 | €36.60 |
Over months 1 to 6, net revenue per customer totals €83 (€28 + €55). After deal cost, CPA keeps €38 (€83 − €45), revenue share keeps €58.10 (€83 − €24.90) and the hybrid keeps €46.40 (€83 − €36.60). On this customer, revenue share wins. On a customer who stops at day 3 (D3) with €8 of net revenue, a €45 CPA loses you €37, while revenue share costs you €2.40. The deal has to be priced on the customer mix you actually get, which is why affiliate management starts with quality, not recruitment.
What is negative carryover for?
Without negative carryover, a partner who sends discount hunters in March, whose refunds and promotion costs push the month below zero, and a few high-value customers in April keeps the April revenue share, while you absorb March's loss. That is a one-way option against you. Carryover means the negative March balance is set against April before the partner is paid. That is not harsh; it keeps revenue share a share of a real business.
If a partner refuses carryover, price them on CPA or walk away. The rule belongs in the insertion order. A Slack message is not a contract.
What happens with a realistic mix of 100 new customers?
Take 60 low-quality customers at €6 of D30 net revenue each and 40 decent customers at €90 each. Total D30 net revenue is €3,960 (€360 + €3,600), a blended €39.60 per customer.
- CPA at €40: you pay €4,000 for €3,960 of net revenue, a gap of −€40 before discount cost is even counted.
- Revenue share at 35%: you pay 35% of €3,960 = €1,386 and keep €2,574, which is survivable provided discount cost is already deducted inside net revenue.
- Revenue share at 35% without carryover: if refunds and discount costs push the low-quality months below zero while the decent customers arrive in a different month, the partner is paid 35% on the good months and none of the losses come back to them. Your blended model no longer holds.
The table is the argument. Screenshots of volume are not.
Where does this meet media CPA?
If paid media's honest target is €46, as worked out in our article on target CPA and customer lifetime value, an affiliate CPA of €55 is not better performance. It is a more expensive channel. Align the two, or accept that some affiliate spend is really a tax on your own brand searches.
Read next: How to set a target CPA that still works in month three. Or see how we manage affiliate programmes.