The Economist this week (above) gets to grips with the puzzling world of web visitors, page views, download stats, "time spent", and what they all mean for the curious publisher, blogger or author in terms of how many people are reading their output. The piece gets it right to my knowledge -- for example a "hit" is a less useful statistic than a "page view" because the former measures every graphic on a page as a separate visit. However, page views themselves, as well as session-length metrics, have their own problems, as outlined in the Economist article. And don't forget the stories of the hard-to-navigate websites that pride themsleves on the length of time visitors stay -- while all the time the visitors are desperately looking everywhere, not being able to find the information they need.
One thing is for sure: only believe a publication's visitor statistics claims if the statistics are independently verified -- for example "Counter compliant". (Here is some information about Counter.)