I just wanted to see what other people's approach was. I do the same thing that you do Arima, and I set one small goal at a time. I'll concentrate on getting a menu to work right, or an animation to move properly, and then I'll move on.
Sometimes, though, there are several features that are interconnected and it makes it confusing on what to tackle first. Sometimes you have to develop features in parallel.
My biggest problem is when I stumble into a task, and I just can't get whatever it is to work. Beating my head against a desk over a tiny feature that won't work right for a couple of days sucks. I need to learn when to move on from a frustrating problem and come back to it later. After awhile I think I lose all objectivity and I need to come back later with fresh eyes.