I agree with the "change one variable at a time" method of load development.
I've found that increasing powder by as little as .1 of a grain can have some drastic impacts on groupings at 100 yards. I'll generally load a range of charges and see which ones perform the best. I've also found that more does not necessarily equal better.
For instance, 44.5 grains of 4895 works well with a particular bullet, but 44.8 grains tends to not group as nicely out of the same rifle, using the same components.
So work with one variable at a time and work up from there.