I think the Netflix piece, and what I was hoping to convey, is that the decisions about how to implement something, or change something, should come from a mix of humans and data. This isn't the decision to actually do something, but more, if we need to change a form/page/module, how or why do we make that decision? Is it performance related, as in the thing is too slow? Do we think the there is a better way to implement the feature? Are users confused and this causes problems?
I think often we might look at something and think we know the problem, or that we have a solution that will work, but making decisions without gathering data, or involving more people can lead to what person A wants, but not what might be better for a larger group.