I have a dataset, where multiple campaigns show that that campaigns have negative impressions. This however, at least of my understanding, should not be possible? In that contex
