+1 to what Terry said.
Some setups have automatic voltage setting. The gcode sets up the proper cut height (the important bit) and the THC is switched off. Now, assuming your plate is nice and flat and parallel with your XY rails, the cut begins at that precise cut height, and doesn't care what the voltage is. The system takes multiple samples of the voltage for a certain time then averages those readings to give the SET VOLTS. It then passes that voltage to the THC, effectively saying, "At this cut height, feedrate, airflow, amps, etc, etc, THIS is the voltage you need to maintain your proper cut height".
That's also why the voltage can change when the electrode hafnium starts eroding away with use. This effectively increases the arc distance and hence the voltage. So as your electrode wears you need a higher arc voltage to maintain the same torch to work distance.
Another method of getting the correct voltage is to do a test cut with THC off and simply note the cut voltage. The feedrates, etc can be tweaked and the voltages noted at each test run, and the parameters from the best cut are what is used. But once again you must have a flat sheet laying parallel with your XY rails (no lumps of slag on the slats, lifting up the sheet) for these tests to be accurate.
So at the end of the day the arc volts is ONLY a representation of the cut height, which is the critical factor that Hypertherm for instance base their best quality cuts on.
If a different cut height to the cut charts gives you a better quality cut, then leave me out of that one, that's something to argue with Hypertherm, etc
Keith.