Counting Web Traffic

A couple of old 'web traffic' chestnuts cropped-up today.

First, hits are no valid way of counting real traffic to your website. Sure, they are always the highest of web metrics - being the total of all files (pages, images etc) accessed in a given time period - and as such, there's always the temptation to say "we've had 250,000 hits this week". What this really means is, we've had a total of 250,000 files accessed, it could be 50 people or it could be 5,000 people, but we like 250,000 because it's way way more impressive than 5,000. Hits tells you so very little about your website traffic. Instead, unique visitors are the true test of web traffic. For each unique visitor there will be one or more page views, and since each page is usually composed of several files such as images, one visitor will generate several page views each one of which will generate several file hits. And the total in bytes of all these files is the bandwidth usage.

And this leads on to the second chestnut of the day: (graphical) page counters. Don't use them. They tell your visitor nothing useful, and provide you with no useful information at all.