We have other ways to control targeting that are based on user attributes available within your app as opposed to SQL in your data warehouse

The idea is to do as much as you can with targeting to limit experiments to the right people and then use segments and custom SQL filters during analysis to clean up anything needed after the fact.

Is it possible to get the sum of all values in the metric without dividing it by the sample population? I think the numbers are there but the metrics emphasize the calculated value (e.g total revenue / size of users)

This might just be me not understanding how metrics are calculated.

So in the above case, the overall number is bigger ($39K vs $34K), but that's just because it happened to have more users. The average is actually lower