How to determine uniqueness of clients from http requests?

Question

I notice that when http requests are made from clients through a proxy server, then the IP address of the requests is always that of the proxy. So if many clients from a huge corporation with a proxy server access a web site, I cannot tell if the requests are from unique clients or not. Is there any way to determine uniqueness of clients if the http requests are through a proxy? I know that the mac address is not included in the http request, so I have just about ruled that out.

Answer 1

Provide a cookie to each new user with a GUID. You can track that and even include the GUID in your server logs.

We do this with our public web server to track "unique paths" through our site.

Answer 2

The simplest way would be to set a cookie on the response, and check it in the request. If it's there, then you've seen that client before (and you could include some identification in the cookie). Of course, this relies on the clients being cookie-aware and the user not having disabled cookies (or clearing them manually).

There's also the issue of some clients which may be cookie-aware, but will effectively start from scratch each time - for instance, if someone's running a program to scrape your site, it will probably start with a fresh cookie jar each time, no matter how you set the cookie.