Home : Topics : Traffic : Traffic statistics
Traffic statistics
Search engines
Spending money
Some myths
< Previous: Logs

Traffic statistics

There are several statistics used to count traffic. If you want to generate misleading figures, there is ample opportunity to do so. None of the statistics are 'accurate' because of caches and proxy servers.


Terminology varies, but most people consider a 'hit' to be a single request to the server for any kind of file.

If somebody retrieves a page with four images, this would constitute five hits; if somebody with images turned off retrieves the same page, this would only count as a single hit.

Hits are generally considered a particularly stupid way to measure traffic; they are most often used when a large number is desired.


A pageview is a single request to the server for a Web page (usually a .html file).

Since pageviews are not biased by the number of images on the page, they provide a useful basic statistic, especially if you are interested in views of a single particular page (e.g. front page).

The count is not accurate because of caching and proxies (see below).

Unique visitors

'Unique visitors' are calculated based on the Internet IP addresses of computers that requested data. If files were requested from 200 different addresses, then that counts as 200 unique visitors, regardless of the number of files requested from each address.

This statistic is intended to show the number of different people who visited a site. It can be unreliable: see below under proxy servers.


Web browsers 'cache' data, storing it for later use. If you view the same page twice immediately, the page is not generally requested from the server a second time. In other words, if a user looks at the same page twice in one session it will only normally count as a single pageview.


Most large Internet providers and many companies, educational organisations, etc. use proxy servers. When a customer's Web browser requests a page, the request is redirected: a request to www.google.com is not sent to www.google.com, but actually ends up at the proxy server.

The proxy server then does one of two things:

Requests from a proxy server appear to come from the proxy's IP address, not the user's. This messes up unique-visitor counting in two ways:

Proxies may also cause under-reporting because they cache information for all their users.