First of all, design should be working. I have seen many times a brilliant design with meaningless content (I started working on a PC in 1987, at DOS, in textonly mode). And people have different liking. One likes watermelon, another one likes a gristly pork...
This one was that bad that it eventually became very popular. So, I'd use the conversion rate as an ultimate metric of good/bad design (for the websites).
For the applications, things are a bit more complicated - you can measure productivity (including errors rate), but should also include factors like stress.
Bottom line: if you exclude ethics, the revenue is a very reliable (objective) metric.