Had the loudness vs. dynamic range discussion with a client last night. He choose "louder" arggggggr.
There are simple "solutions" to this:
-Match the RMS levels of the files and then let your clients listen to them.
-After that let them listen to 128kbit/s mp3 files of the squashed and dynamic masters at the same RMS levels.
If the client then chooses the squashed version its a lost cause.