This post represents a major deficiency that is common when people think about reflections and speaker-room interactions: people think in 2 dimensions and flatten the room, collapsing the vertical dimension.I want to play a game.
Everyone take out some graph paper. Draw your listening room to scale, speakers at the top, positioned as you have yours.
Draw in their approximate radiation pattern.
Now draw the first and second reflections of those speakers as additional virtual sources on the other side of the front and side walls. Draw in the radiation patterns of the virtual mirror image speakers. What happens to the mirror image radiation patterns? Reposition your speakers. What happens to the virtual speakers?
Now formulate a theory of what causes the size and shape of the soundstage based on radiation pattern and speaker positioning.
Welcome to Image Model Theory!
In reality, the height is typically the smallest dimension, and floor and ceiling reflections (including higher order reflections such as sound bouncing first off the floor and then off the ceiling) are responsible for the vast majority of reverberant sound degradation (particularly for rooms with significant square footage). Build a speaker with a very narrow vertical beamwidth (less than 60 degrees: aka +/-30) and you find clarity much, much farther into the room than with a conventional speaker (likely all the way to the back wall).
Since the post by geickmei concentrates on soundstage, it makes some sense to prioritize the horizontal reflections, except that you have to include the virtual vertical sources to properly understand the distribution of your virtual sources even in the horizontal plane. Those vertical reflections reinforce the sound and don't widen the soundstange at all. Therefore your high-angle horizontal reflections that can widen the soundstage have to compete with more sources than the ones diagramed based on the instructions.