Comment by hruk

No, you definitely have to pick your battles. Something that you want to continuously optimize over time makes a lot more sense than something where it's reasonable to test and the commit to a path forever.