Cost-per-acquisition deals are attractive because they appear to move all the risk onto the affiliate. You pay only for depositing clients. What could go wrong?
The variance problem
At low volume, conversion is dominated by variance rather than by quality. If your true conversion rate is 2% and you run 100 leads, you expect two deposits — but the realistic range runs from zero to six. A month with zero deposits tells you nothing about the source, and yet it is exactly the month in which most brokers cancel the deal.
Below roughly 200 leads a month you simply cannot distinguish a good source from a bad one inside a single billing cycle. You are making decisions on noise, and the affiliate knows it.
What the affiliate optimises for
A CPA affiliate is paid on the deposit event and nothing after it. Their rational strategy is to maximise deposits regardless of what happens on day 31. That is not dishonesty; it is the incentive you designed. Expect aggressive bonus messaging, expect deposit-and-withdraw behaviour, and expect your retention team to inherit a cohort that was never yours.
What works better at small scale
Buying verified data outright and running it through your own floor gives you three things a CPA deal cannot: you keep the records, you learn what converts, and your sales team builds the relationship from first contact rather than collecting a hand-off.
The trade is that you carry the risk. That is a fair trade when the data is documented and replaceable, and a terrible one when it is not — which is why the replacement guarantee, not the price, is the term to negotiate hardest.