That's a great point. Maybe for example:
* level of negative externalities due to homeless people in city X goes down by Y%
* number of short-term homeless people in city X goes down by Y%
* number of chronically homeless people in city X goes down by Y%
* number of homeless people in city X goes down by Y% of which no more than Z% may be due to decreased arrival rate or increased departure rate of homeless people
* number of homeless people in city X goes down by Y% without raising net tax burden in city X by more than Z%
* tax burden of homeless services primarily met by those who benefit financially from increased rents
* housing stock in city X increases by Y%
* median apartment rent in city X decreases by Y%
* average apartment rent in city X decreases by Y%
* apartment rent in city X keeps place with inflation
* Gini coefficient (income inequality) in city X decreases to at least Y
* unemployment rate in city X decreases to at least Y
* government refrains from conditioning benefits or services on acceptance of drug treatment
* government refrains from conditioning benefits or services on cessation of drug use
* government refrains from conditioning benefits or services on acceptance of mental health care
* government requires acceptance of drug treatment or mental health care as a condition of receiving services
* no increase in arrests for crimes often committed by homeless people
* substantial increase in arrests for crimes that significantly affect other people's property or quality of life
(this isn't even including broader-based changes to unemployment insurance, welfare, or health care systems)