I did some (quite a bit actually) MDA (over large data samples).
I looked at the output, condensed it (probably over simplified, but integration is more important than accuracy for me right now) and applied it over a tiny sample of play (10k).
Given how differently I'm playing than I was previously, I'm honestly a bit lost as to how and when to attempt to determine which of these changes I should keep and which I should not.

Right now I'm so far out of my comfort zone it's not even true, on top of which I'm struggling to interpret the feedback I'm getting.

Thoughts on methodology for comparing the efficiency of changes (and samples required)?