But Amdahl's Law isn't totally applicable in this case, because if you manage to gain more subscriptions via paywalls or whatever, you raise the 3% figure but also the total revenue. It's possible (albeit unlikely) that subscriptions could increase to the point where they account for 50% of revenue, without necessarily decreasing the number of newspapers sold via newsstand or the advertising take.
In the reverse case, where you're trying to reduce one of them to reduce the total, it would apply well; your greenhouse gas example seems appropriate. Or the way the media reports political spending, for example; they like to focus on small examples that people relate to better, which make very little difference individually to the overall total.