One issue is there's infinite metrics to measure outcome, and any measurement you choose will have certain groups excel in compared to others. Once we achieve equality in a certain metric we will always have new ones to work on essentially forever. This may be a good goal for people/institutions but it can't contribute to discrimination being indefinite in scope.

Another issue is we're talking about group level outcomes here - which means we're already accepting that there will be biases in measurements, otherwise the group wouldn't be a 'group' (unless it's literally just skin color which is - a pretty arbitrary/racist way to group people all else being equal).