Yes, I agree. That's why my example was a randomized trial. But you could also compare with a baseline from results in the previous year or whatever.