This makes sense in some cases, but there are several functions of government that are fundamentally money-losing prospects that wouldn't change when optimizing for impact. I agree that the metrics around government efficacy can be improved, but arguing that increasing impact will always or almost always lead to profitability isn't feasible in many government cases. The fear I have with such an approach is that if governments indirectly promise profitability via maximizing impact and they fail, then you still have the same problem you started with.
A perfect example of a project that will only ever lose money but is very important to the public is the Hanford Nuclear site[0]. The US government discovered recently that the nuclear waste site was leaking radioactive chemicals and threatening to taint the Columbia River water supply, which millions depend on. Cleaning up the radioactive waste will cost tens to hundreds of billions, and the best possible outcome is that everyone still gets clean drinking water. That should be the most desirable outcome, but even if successful, the government won't generate a profit from it, nor should they. Utilities shouldn't be priced to be profitable, they should be priced to be sustaining and accessible to those who need them.