While ray tracing (2D on paper or 3D with lasers) tells you some good information, there are enough factors complicating matters that you really need a simulation tool, such as EASE, to design coverage with any kind of useful accuracy. My comments will overlap a little with what others have said.
If a loudspeaker is 2-way, the stated coverage pattern refers only to that of the HF horn, which likely only goes down to about 1.5 to 2.5 kHz. Only if you have a 3-way speaker with a large MF horn will you achieve any kind of directivity down to, say, 300 Hz.
Also, the stated coverage pattern is as it is applied to the surface of a sphere at some distance. Go read the article linked on the PSW homepage about designing distributed ceiling speaker systems. It gives good examples of how a stated 6 dB down at, say, 60 degrees off axis actually results in more like 13 dB down when your listening plane is flat, not a sphere. And that's when the listening plane is perpendicular to the speaker's axis. Now imagine what happens to the response when the listening plane is at a shallow angle to the speaker's axis, as it is in most PA applications. That's why when you have a single speaker covering the depth of a room (no delay speaker), you typically aim the axis of the speaker at the last row of people, thus "wasting" the entire top half of the horn's coverage pattern, making it hit the back wall instead of people.