Wait, hold on. You mean the paparazzi guy had a VEST on? Like a real press vest with the little pockets and everything? That is insane. Ngl, that detail is what really sells the whole scam to me because who actually wears those things unless they are trying to look official as heck. Honestly, it sounds like some low-level manager thought paying two dudes to fake snap photos would make their client look more famous than she is. Tbh, that is so pathetic it almost circles back to being funny, but also kinda sad for everyone involved, especially her. I bet the whole thing gets cut into some reality show blooper or a thirsty Instagram story later.

Kinda sounds like you were paying to train the algorithm on garbage traffic. Same thing happened to me with a pet product line. I cut spend by half, tightened the audience to just people who'd actually bought before, and the CPA dropped almost overnight. Your buddy's not wrong that two months is a small sample, but 3.4x vs 2.1x over 30 days is a pretty solid jump. Maybe the bigger issue is you had too many broad campaigns running and the system was scooping up window shoppers. I'd test, not with a full cut, but with a leaner audience and see if the trend holds for another month before you call it a fix.