The grid method is something that an artist should outgrow once leaving behind the beginner stage, because the longer you rely on it, the more it becomes a crutch. After developing your observational skills and eye-to-hand coordination for a while, you should have honed your ability enough to work without a grid system, or at the very least, only use a very basic one like a quadrant division or something. I’m not saying don’t use grids, just that think of it as the helper wheels on a child’s bicycle–once you get beyond the early stages, you need to take off the helper wheels and actually ride the bike for real.
As for starting a portrait painting digitally, there’s no real difference between analog and digital. It’s a matter or preference how an artist starts a figure. Some people do line drawing/sketching, some people block in the major shapes with solid values/colors first to establish the composition, and some people just make are few marks that indicate where the major landmarks are (eyes, nose, top/bottom of head), and then start painting in an alla prima manner (Richard Schmid is a master of this). I come from a comic book/animation background, so my habit defaults to drawing first–that’s just personal preference though.
I think for less experience artists, having the roadmap of a line sketch/drawing will make them feel more secure in proceeding into values/colors, because they can make sure the sketch/drawing is sound before going further, whereas if they worked with valued/colors right away and make mistakes with proportions that need correcting, it’ll take more work to repaint than it does to redraw.