I find this technique works well
1) start First by adjusting the right side screw looking only at the first test point, getting it to 205 ignoring the other points during this step. Multiple runs may be required.
2) Next use left side screw to raise/lower point two to match point three. This is the key, two and three need to be equal values, ignoring that they are higher or lower then point one, points two and three just need to match each other.) again multiple runs may be needed.
3) Finally use the front screw to raise/lower points two and three to match point one which should have remained at 205 during steps 2&3.
When I work it this way I Finish at 205,205,205 saved to the machine.
note: calibration routine on rev g firmware gives calibration output results ONLY after a failure, so I went to rev J firmware that displays the results after every pass. I don't have a reseter for filaments so not worried about that.
edit: I'm hearing 250 is the magic number to shoot for. but as long as I get a good even number on all three saved is my main goal. eventually I'll get a good depth probe and start dialing them in better and ignore the numbers altogether.