I used to be a big proponent of soffit mounting. Now I'm more on the fence about it.
With soffits you can create an early reflection free (never completely) zone. And this is unquestionably a good thing. The other thing that soffit mounting gets you is a flat power response to go along with your flat on-axis frequency response. Freestanding speakers require a -6dB high frequency shelf in order to compensate for the "baffle step": the fact that the high frequencies radiate hemispherically while the lows radiate omni-directionally due to the finite size of the speaker cabinet.
And it's this flat power response that now has me questioning its validity. A flat power response would seem to be a good thing theoretically, but the vast majority of listeners have freestanding speakers. So, from a translation perspective, soffit mounting seems to be less than ideal. You can imagine that a freestanding power response could be simulated by adjusting the relative HF and LF absorption levels in the room. However, you don't want to make the room more dead in the high end, so you have to make it more live the low end. Then you're faced with the issue of enhanced room modes and such.
I think the ultimate solution is a soffit mount arrangement in a room with widely splayed walls and ceiling, whose reverberant response has a relative +6dB LF step. This way you get the best of both worlds: a reflection free zone along with the more translatable power response of a freestanding speaker. Next best is freestanding speakers pulled far way from the walls (preferably splayed) in a large room.