What is the right way to frame it? Total cost per passenger mile might be good. The transit systems that move the most people efficiently would do well on that metric.
Sometimes basic science research funding is framed in terms of "this program generated $10 of economic activity for every dollar spent." Social programs sometimes measured this way too. The term for this escapes me at the moment, but I think it would be more useful?
This type of cost-benefit analysis (or economic multiplier calculation) is also used to justify public subsidy of sports stadiums and the like. Unfortunately, these analyses always use overly optimistic assumptions, and fall victim to the broken windows fallacy.
Calling the police or fire department should incur service chargers.
Libraries should charge rental fees.
Schools should all charge tuition, including "public" schools.
If they're not at least breaking even, are we sure they're providing any public good?