Looking it, up I can find some equations for deciding on the components. So all that said, yer pretty much just saying that i start with my original signal, split it into a low-passed and a high-passed, Send the low to an amp/speaker and send the high to the next set of filters and then repeat until I get my number of speakers? Really it seems simple, just a matter of getting design well down.
Yes, that's generally the idea. You can combine them if it makes sense, but I might choose a more parallel layout to reduce the number of filters in any given signal's input chain... so, picking some random ranges, for example:
0-50Hz, 50-150Hz, 150-1000Hz, 1000-5000Hz, 5000-20000Hz
I might go from an input buffer to high-pass stages at 50Hz, 150Hz, 1kHz, and 5kHz, plus the lowpass at 50Hz. The 5kHz highpass goes straight to an output, and the 50Hz, 150Hz, and 1kHz highpass filters cascade into their respective lowpass filters (at 150Hz, 1kHz, and 5kHz).
This way, you reduce noise and phase distortion by the number of components in any particular signal's path.
I believe the links you gave are the same calculator I used when I designed a variable-frequency lowpass filter for a little powered sub. Sallen-Key isn't the best filter in the world, so if you have no need to vary the frequency (or you want a different filter slope), Linkwitz-Riley might be a better choice. Then again, if it's just academic, SK is more than adequate.