Cutting off previews definitely isn’t perfect, but at its core I think a good heuristic is to use a 7day window as the denominator so you have one of each type of weekday (this also induces a nice straightforward correspondence between multi and avg weekly drop). That’s also why, later in the run, I might reference a forward multi from d4-d10, or d8-d14, etc.
With modern preview conventions, OWeek is in effect an 8day window, and true OWeek is a 7day window (real d2-d8). If I was forced to pick a method with included previews I would probably go for the reported D6 cume, which is a real 7day and includes one Th+one fri-wed instead of 2 Th+one fri-wed.
and yeah, then the other matter is just computational convenience in terms of looking up the figures and doing mental math on them (or spreadsheet if feeling serious).