What's with seeking averages for everything? When you have completely different situations the average is not useful information:
When I go to a book store I pick up a lot of books, skim them, and put them back because they are boring or not what I wanted. Eventually I will find an interesting one, find a hidden courtesy chair, and read 50 pages. You could study this information and say on average I read 3 pages of all books. That's correct, but it's wrong and misleading.
It's not really worth critiquing the rest of the study since it's just not rigorous enough from the outset, but a couple more points:
1. He threw out visits lasting longer than ten minutes. Obviously that's going to affect his conclusion.
2. They have no idea what percentage of the text on a page was actual content. I'd guess 25% of the words on any newspaper site page are not the actual article. He should correct for this.
3. He tacks it on the end, but it seriously challenges the usefulness of this entire study: the content matters. People will read what appears interesting to them and ignore what doesn't.
† Surely knowing their behavior is being monitored alters their behavior. If these are work computers then they're probably wary of spending too much time on the internet if it isn't directly related to their work. Maybe they bookmarked articles to read at home. Or maybe they altered their browsing because they're worried about being judged.