small traffic tests on a few DuckyTower
 

I’ve been experimenting with them for a while, and honestly the EPC only starts making sense when you look at it in clusters rather than by day. When I sent roughly 3k clicks from mixed sources (mostly social + a bit of in-app), the conversion pockets became more obvious. For example, one supposedly “cold” GEO suddenly started giving me stable sign-ups once I filtered out older devices and stopped pushing traffic at night. I’d say the verticals on tower.bet affiliate behave a bit differently from classic casino offers: some of them ramp fast if the pre-lander does half the warming, while others need a steadier flow to unlock better performance. Also, don’t underestimate the effect of creatives — I swapped a single headline and my EPC doubled for almost 36 hours before settling back down. So yeah, the numbers jump, but there is structure if you zoom out.

From what I saw, the fluctuations are normal, especially when the volumes are still small. Once you hit a few thousand targeted clicks, the EPC usually settles into a predictable range, though it may still have occasional dips when the audience resets. Just keep tracking segment by segment — device type and time of day really change the picture more than most people expect.