I'd agree with functional group first. If you know your target, it will undoubtedly contain some functional group transformation.
Then I'd the go by reaction mechanism. If you were making a alcohol, I'd find it logical to look at reduction reactions separately from reactions of aldehydes, ketones or from alkenes, for example. Also think umpolung.
I never worried too much about stereochemistry. I generally thought that for every reaction that gave one isomer, there would be another reaction that gave the opposite isomer. Because I was in the area of chemistry in which a lot of chemistry involved aromatic rings, regiochemistry was more important. Using a different chiral ligand is not going to reverse the course of a reaction. None the less, I wouldn't organize about regiochemistry per se. I'd stick with mechanism.
I was in chemical discovery. It may be interesting to hear what Disco would say as he was more involved in large scale syntheses.