We look at window boxes differently than we do free standing containers. The classic formula for container gardening of thriller-filler-spiller or, the less sexy, focal-field-trail is tried and true. It works so well because our brains are wired to 'focus' on a prominent feature, usually the thriller, and then the eye moves downward to other elements that complement or contrast, but in essence, support that focal element. We still have the same TFS formula in a window box; however, now that the view is wider, the placement of our elements becomes more important.
The more the eye moves while viewing something the more interesting it is. I think we can all appreciate the layer cake design of different strata of complementary or contrasting elements. Though this design moves the eye up and down, it doesn't however, move the eye forward or horizontally. The problem with the layered design in a window box is that layered strata are meant to be read vertically, where window boxes are naturally read horizontally.
Think of the window box as a music score. With all the elements laid out side-by-side, as in stratified layers, the elements read Do-Do-Do-Do. How would Beethoven's Fifth sound if it were Re-Re-Re-Re instead of Re-Re-Re-Do? That counterpoint placement is the key. In the horizontal design of a window box, moving the elements high and low, like notes, is as important as the elements chosen. This creates a more melodious Do-Re-Me and your eye will hear that music.