Mastering audio issues and better ways to deal with them

I have a question for mastering engineers working with audio files. When done mastering an audio file/mix/etc., in whatever mastering suite/application your using, when converting to say an MP3, what is your process ? i have been mastering it as if there will be no formatting to lesser container/lossy format, applying dither i think would be best for the mix/song, then converting to say mp3, afterwards, i compare the master and the mp3, then go and make adjustments to the master that i think will make the formatted file sound as close as possible to the master file, which usually means some equalization work, compression work and dither application review, then reformat again, recheck against master, and repeat until satisfied. the issues seem to be;

• -Loss of samples, and Bits (word length)

this will change the complex wave,

freq. content, and representations accuracy

-Compensating for the loss of data

-Compensating for Round off error,

overflow, underflow, Quantization error,

Sample loss, Bit loss, Interpolations,

Formatting algorithm(codec, container) errors, Etc.

Also, masked frequency removal by the format codec, and removing frequency ranges covered by louder sounds/frequencies.

Any thoughts on more advanced concepts to dealing with the loss of info when formatting, and possible solutions in MatLab itself ?

Thanks...