It sort of makes sense if the experiment is about judging not just individual user reactions but the wider reaction of the media, advertisers, "the conversation

It sort of makes sense if the experiment is about judging not just individual user reactions but the wider reaction of the media, advertisers, "the conversation" to a change. I don't think that makes it a good idea though, seems like each failed "experiment" poisons the pool and certainly makes it possible to do experiments in isolation.