In my opinion.
Once in a while I’ll read something about mixing engineers and how they create different mixes for CDs and digital music services. The arguments are that encodes don’t recreate certain frequencies well, like anything below 100hz. So because of that they may suggest boosting the low-end on anything that will be used for an MP3 encode.
I think this is a bad idea.
I think mixing and mastering engineers should work on music for ears, not encoders. These days, bitrates for digital music stores are about 256k, which may as well be lossless for most people. And someday we all may get lossless encodes – which leaves a catalog of mixes that were poorly engineered in the hopes of making up for the weaknesses of an MP3 encoder.
Plus, there are so many variables to think about (which mp3 encoder? bitrate? AAC/WMA/OGG?) that trying to create the perfect mix for each of them means chasing your own tail, and it doesn’t account for the fact that I, and many others, will be ripping the official mix to a digital format anyway.
Besides, mixing for a certain encoder means giving those encoders a pass. Better to let the flaws of the encoder shine through so that the market can choose to not use that encoder and move on to something better.
However, I think you should arrange your album for digital formats. Please spare us hidden tracks or 5 minute stretches of silence that were a trend during the 90s. Downloading a 10meg file of 256k CBR encoded silence is obnoxious.