In the end, it seems the math was quite simple: DRO reading (that is, the distance the computer thought it moved) divided by the distance it actually moved, multiplied by the "turns ratio"- that is, the number of turns the stepper motor shaft has to make in order for the axis to move one inch.
The spreadsheet makes it fairly simple, in that you can easily keep "fine tuning"- that is, test, change the ratio, test again, adjust the ratio again, etc. until you get it dead-nuts on.
The problem is it seems to "swing" a lot- you overadjust, then re-underadjust, then re-overadjust, etc. as you close in on the number. I admit I cheated a bit, and got close with the above method, then just made manual changes in regular amounts.
That is, the calculation might have given me 15.103462, but that still gives me an error of, say, 1.0065" instead of one inch. Rather than go through the calculation which would overshoot and give me a .997", I just changed the number by... well, blind stab.
I moved it to, say, 15.1035, which wasn't enough, then 15.104, which was a lot closer, then 15.1045, and so on. (All numbers just examples for illustration.)
Basically the same procedure we used in setting up Mach 3.