The best way is to just do it manually. You can start by taping a newspaper to a wall and running through AFMA adjustments until you get the best image. Then go out and shoot a duck on the water to dial in the adjustment.
Makes sense...and that's pretty much what I do, although still using FoCal. Basically, I trust an image analysis program to determine which is the 'best image' more than I trust my own eyes (and I should point out that my day job for many years was analysis of microscopic images, both 'by eye' and with digital image analysis algorithms).
When I do a FoCal calibration, I use lots of light (my 600 II tests were in the EV 14-15 range) and a solid support (RRS TVC-33) on a stable surface (concrete or asphalt). I don't tether the camera and let FoCal do the iterative adjustments...rather, I take 83 shots per 'test' - 2 shots at even-numbered AFMA values from |20| to |12| with one starting focused at ∞ and the other starting at the MFD, then three shots at every AFMA value from +10 to -10 (inclusive), one from ∞, one from the MFD, and then one without moving the focus ring. I load the 83 images into FoCal in manual mode, and there are ample data to drive a curve fit. I run a test at 25x focal length and 50x focal length, and for zoom lenses, at each end and 1-3 intermediate focal lengths.
I then take the value(s) from FoCal and do real-world testing with the lens with AFMA applied, to make sure I'm getting sharp images.