i find that for me personally, writing out the whole thing and then solving for the answer is quickest. for example, if you are working on stoichiometry, instead of doing unit conversions one unit at a time, i'll write out all the conversions, plug everything into the appropriate equation, and then start cancelling out variables/values. one downside to this approach is that you spend time on stuff that you could have done in your head-- but i think it's helped me in terms of accuracy. also, it makes it easier to check your answers later if everything is written out.
it's cetainly not something everyone follows, but it's worked for me fairly well.
gluck!