(a) Visits for today are regularly updated as per normal, while the overall total remains static. So instead of these two figures growing consistently – as in n + 15 being matched by m + 15, n + 412 by m + 412, and so on – m (the total visits) remains at m all day while there are regular increments to n: n + 15, n + 412, n + 802 etc.
(b) At midnight, the total figure, m, does increase but not by the number of visits for the day just ended; rather for the same day a week ago. So, for example, if yesterday was Saturday 25 August, the increment to the total is the number of visits for Saturday 18 August.
(c) The shortfall in the total caused by the initial ‘loss’ at the beginning of August remains; it hasn’t been made good.
Which, I opined at the time, is probably due to the multiple databases involved getting out of sync.
These results were first observed on my own Sitemeter, but I have now also studied those of six other bloggers, with ‘open’ Sitemeters, and the results for them are identical (except that I don’t know the extent of their initial ‘loss’). That doesn’t, of course, show that the results hold good for all Sitemeter accounts, but it does suggest that the malfunction I describe is somewhat general.
I am one of the six Norm studied. My own observations are generally consistent with his. Today’s “Summary” will show 2,297,230 all day, though the actual meter at this writing indicates 2,328,640. This is a difference of 31,410, which has varied very little in the last four days. I continue to believe that the higher figure is correct, and that the lower ones resulted from failure to post changes during a period of site “upgrades.” The weekly report emailed me on Saturday had the correct Saturday total. Of course, I can’t prove it, and usually the only way to get through to tech support is to have your payment go troppo.