Making bad assumptions about data seems more likely than deliberate malice. If you A/B test different designs, and you see a huge increase in "interactions" with some design, you'll tend to assume that's a positive thing, unless you spend the time to interview a lot of customers to figure out what's going on.