The following started as initial estimates of the causes of overhead. They happened to result in the overhead we assume for time-tracked research support. So far, updates haven't had much effect on the total. While originally intended for only (time tracked) research support, there's no reason why it shouldn't apply to all work.
Being just rough average estimates, you can fill in your own values to determine your overhead more accurately. Follow the new value with an "enter" to cause totals to be recomputed. Force reload the page (usually shift reload) to reset to the defaults.
Our time tracked research support is assumed to involve % overhead, which implies being available % of the time, i.e. hours/week. Our time estimates above imply % overhead, thus being available % of the time, i.e. hours/week.
So hours/week ( hours/day) is a reasonable long-term average for trackable work.
The hours/week estimate in the above can be misleadingly low, as it takes into account time away from work. Using just the "overhead hours per week" results in an estimate of being available % of the time, i.e. hours/week.
So hours/week ( hours/day) is a more reasonable goal for short term time tracking when at work.