If this Stephan Franke guy hasn't retired since posting this message...
I would suggest that your usership is divided into three primary segments:
DJ/remixer, taking existing material and modifying it.
Electronic musician/composer, writing new music and designing new sounds.
Theorist/educator, establishing new models for understanding music.
Recording engineer/sampler, capturing live performances or found sounds
Now, of course, none of these is mutually exclusive, they are maybe quadrants on a graph, where one vectors somewhere in between. In fact, it is likely most Live users do not fall firmly into one quadrant, because there will be some quadrants that are more purely covered by other applications.
In terms of each of the above, there are sub-areas:
for example (For EXAMPLE, not exhaustive),
DJ-- turntablist, knob-twiddler. These may want radically different controllers modeled after what they are used to.
Electronic musician-- keyboardist, vocalist, guitarist "models". As a guitarist myself, I am very used to holding a necked instrument and using a pedalboard to activate effects, and am very comfortable with this mode of interface. A vocalist/guitarist would tend to use the same mode. A keyboardist/vocalist may prefer more controls near the keyboard, and few or none for the feet.
Theorist/educator-- this is someone who is likely to build his/her own, or to want only a highly configurable modular system. They would also like 'utility components' that convert any kind of message into a MIDI message.
Recording engineer-- would want a remote interface that monitors key metrics of what's going on, and provides a total level of assurance that he/she is capturing what he/she is hearing. An example-- if you are recording with a laptop and your adapter becomes unplugged, you may not be aware of this. While the OS may warn you, it may not be obvious enough.