Took over the UGC on a $100M+ health tech account a while back.
Their static ads did great during sales. Outside of sales they'd burned through $20K and got zero purchases. Literally none.
So take the discount away and nothing was selling.
Most people would've just made more ads. I wanted to know why these only worked on sale, so I built stuff for full price instead.
First batch was 8 ads and all 8 were profitable (2.38x).
Five months later it was 41 ads and 5 of them scaled. Best one ran 54 days on roughly $19K and brought in $51.9K.
Whole thing was $97K spend, 651 purchases, $189K in tracked revenue. Revenue, not margin, before anyone asks.
Yeah, it's real creators and real editors, so it wasn't 40 ads a day like people do with AI UGC now. Still, 8 for 8 on the first try.
Check your own split
If you wanna know whether your ads work without the discount, split the last 90 days into sale days and normal days and compare what a purchase costs at about the same daily spend.
If normal days cost way more, the sale is hiding something.
The full numbers, the ads, and how the batches changed over five months are in the Withings case study.