The best approach is probably not to cut spending to zero and then wait to see what happens. But perhaps it's reasonable to cut spending by 10-25% and then monitor the results?
Any project should have metrics to judge its success and a reasonable budget to accomplish the goal. If you were given a budget of nearly $300M for 7,500 homeless, how would you decrease the number of homeless? If you really think outliers take up the majority of the cost, how many homeless could you house with what budget?
I agree that metrics are useful, but using a terribly flawed metric just because it's the only one we have seems like a bad idea.
(I will not respond to your other points about budgets because I don't have any easy answers for them--this subject is not my main area of expertise. I will leave it to others who are more informed than I to provide actual statistics and actionable suggestions elsewhere in these comments.)
The number of homeless people is around 7,500.
This also doesn't include medical emergency services, private organizations, federal & state programs and many other sources of money. This is just the budget for the city program.
Give people money. Unfortunately, you could only do this effectively on a national level. In San Francisco, $40k isn't even enough to cover cost of living.
You could also give people housing vouchers, but you need to have available affordable housing for that to work.
[1] https://www.businessinsider.com/how-expensive-is-san-francis...
Let me ask you this: if someone doesn't have money, how do they get drugs?
Refusing help to people with substance use disorder is a form of toxic moralizing that does nothing to help them.