Under-funding the test is the most expensive way to save money I know.
The setup
— New CPL offer, $13 payout, push traffic.
— I gave it a "safe" $50 test budget to see if it had legs.
The move
Killed it at $50 spend, 0 conversions, ROI -100%. Filed it under "dead offer."
The numbers (illustrative)
— At a realistic ~1.5% conversion rate and $0.30 CPC, $50 buys ~167 clicks: expected conversions, about 2.5.
— Getting zero on that sample is pure variance, not signal. The math says I needed ~$150-200 just to see a believable rate.
— A peer ran the same offer to $300 and hit 41% ROI.
The lesson
A test budget below a few times the payout tells you nothing but noise. Killing offers on under-powered samples means you'll reject winners at random and never know it. Cheap tests are the most expensive habit in arbitrage.
What I'd do differently
Size the test so you'd expect at least 5-10 conversions if the true rate is decent, roughly 5-10x the payout minimum. If I can't afford a real test, I can't afford the offer yet.
Arb Files
@ArbFiles
I tested with $50 and "proved" a winning offer was a loser
This post was published in the Telegram channel Arb Files (@ArbFiles). All its posts are on the channel page.
Open on Telegram