Analytics programs such as Google Analytics often fail to track various type of users, these may include:

1. visitors that erase cookies after each session or at some other interval
2. visitors that block cookies outright
3. visitors that rewrite cookies for fun
4. visitors that arrive via a proxy service
5. visitors that click on your link or are referred to your site some other way and then leave your page before your analytics code can load and capture information or open their cookie (assuming there is one); Google has come up with a good antidote for this with their asynchronous tracking JavaScript, which is now available in a beta release.
6. visitors that arrive via multiple redirects
7. visitors that randomly visit your site from home, the office, on any other extra computers, their phones, ipads, game consoles
8. visitors that share their computers with two or more household members that all happen to visit your site and similarly multi user such as those at internet cafes and any jumping point where a new or different cookie must be used and where cookies might be shared between two or more people on the same computer who all happen to visit your site
9. visitors who proxy code for themselves, rewriting JavaScript and other parts of a sites code to meet their desires.
10. visitors that spoof their referrers or any other means that will derail validity

Surely there are more scenarios, but these alone are reason to believe that clickstream analytics is handicapped out of the gate. The multi user per computer or multi computer per user scenarios are in particular the most troublesome. These certainly account for a substantial portion of the visitor mix (or mix up as it were). I use at least 4 different devices to surf. I often visit the same site from multiple machines, but I tend to convert (if I buy or trigger the appropriate event) on one of two computers, so there are situations where I know that I appear to be about 4 different users on varying IP addresses (even on the same device), and only one of us/me will ever convert. What does that do to my conversion rate if I’m the site in this case?

This prevalent type of scenario is and will be a fundamental flaw in analytics for the near term at the very least.