Interesting - 2 ways to skin a cat.
I use CD Architect for final levels. It really only does a dB or so gain or reduction, which should be enough for final tweaks. But sometimes songs that seem loud need to come down a notch (even if the meter reads -.2 dB like others), while "softer" songs need a bit of a boost so they feel the same volume. It is pretty easy to skip on the timeline.
I use Sound Forge for "mastering," tho before it gets there it has been through analog compression/EQ. I used to use Voxengo in Sound Forge, and can still send it through elephant if needed there. Otherwise, it is just top n tail and that stuff to get it ready. And then I do all my rate/bit changes in SF, using the final "master" for CDs and mp3s.
That is how I learned to do the job, although Bristol's method looks good, esp. if you don't have SF/CDA or similar programs. I wouldn't use Pyro - I use that for home comps, etc. But either method works - pick the one that suits you best.
@