My experience with AA occurred inside schools. Grutter v. Bollinger was the case that everyone had to comply with. AA used to be a POLICY, and schools that applied it did so by keeping track of that policy with statistics. We used to do the same with people of different economic classes as well. It's kind of like making sausage, people don't really want to see how it's done, but we used to assign grades to applicants taking economic status (a + if you had more money), background (a + if you were a legacy), race (a + if you were a minority), etc., into account. Grutter v. Bollinger prevented that from happening. We no longer could assign preferred status statistically to races (through use of quotas AND/OR grades). So what happened? minorities were admitted on an ad hoc basis. At a place like Penn State where the African American student body flatlined at 3% (and in some years lower) in a state with 15% AA population, we got a big boost by making minor, simple ad hoc adjustments that were no longer based in any defined policy (though the policy existed all the same).
But I added the part about legacies and economic background to give you a sense of what happens in a need-blind admissions policy. The first phase was admittance based on academic record. Then we doled out 35% of all tuition money to the students. After looking at academic record, we combined the record with academic need, and the more need you had, the more your grade would go down. For instance, a stellar student with straight As and a good amount of need would be assigned a grade of A. Landing in the A category meant we would attempt to fulfill 100% of your need. If you were a A-/B+ student with the same amount of need as the kid with an A grade, you would receive an A-/B+ and that meant (on the chart) we'd offer you 75% of your needs, and so on down the line. The kids without any need were admitted solely on academics obviously, but they were highly valued since the truth is we redistributed 35% of their money to the kids in need. That was the system. The legacy kids got an A regardless, we didn't even bother looking at their records. Their files just had a big rubber stamp on top from an administrator--we'd see the file and say, MOVING ON!
The point is, no one in that room was allowed to go back and count minorities AFTER the decisions were made. You can't eff with the process after your realized there was no longer a systematic way to ensure that your campus was diverse. Since Grutter v. Bollinger in 2003, the African-American population at PSU has dropped from 6% in 2000 to 3% now.