On my 10" f4 Newton I use to perform AF routine every 40 minutes since the focus is continuously drifting over the night.
The average drift between two consecutive AF routines is around 50 steps (more than the CFZ of the scope). It is not really correlated to the temperature drop, probably due to differential thermal dilatation between the carbon tube, the mirror cell and the focuser.
To account for this I could perform AF every 20 minutes or so… at the expense of 3 minutes for each AF.
It could be better, in this case, to decide that the AF result must be offset by a certain amount (positive or negative). In this way, the focus could be set optimum for the image in the middle of the interval between two AF, instead of being optimal for the first one only, the others being more out of focus.