I tend to divide it by seasons, usually only looking at it in smaller allotments of time only when prompted over it.
To attempt to measure it we'd need something a bit more concrete though. The seasonal effect is important to take into account but we'd able to see it by comparing past years' activity at the same month and day.
I'd figure that defining the mean of this site's activity would be how to find a bar, and even then I'd still insist on splitting it down to the seasons instead of just a single average.
Down to the day's how to get the data completely accurate, but is it worth that amount of effort when what it'd return wouldn't be too different from a larger span of time? I don't need to know how often Legga signs in every 14th of March to see website activity, and even that would be ignoring if something was different about one year's March 14th over another's.