Unique visitors and IP addresses

[ALarson]

So many metrics programs use IP-based user data for analysis. This data, however, is far from accurate because many hosts are using floating IPs or the same IP for a large number of users. Is there any way to clarify this other than using cookies? I'd be really interested in the alternatives.

Thanks!

[gregbo]

Using cookies instead of IPs is not much better. Do a search for Rebecca Lieb's How Many Users are "One" article for more info.

Some people think using the combo of IP address, HTTP user agent, and cookie gets you to a reasonable point as far as determining unique users. I disagree, but you may find it is a good enough approximation for what you're trying to do.

[digitalv]

The way the web is going, the IP address is going to be as good as it gets. The next generation of web browsers will be blocking third party cookies by default, so a lot of cookie-based web analytics services are going to have to do some major re-design since their cookies will be automatically rejected.

[gregbo]

In general, I think some people have unrealistic expectations of tracking and identifying users on the Internet.

[Dijkgraaf]

Even if the user accepts cookies it won't solve everything, as a user may use computers in different locations (one at home, one at work). If you want to accurately count them as a single user you will actually have to get the user to identify themselves (eg. login)

[nmattheij]

Well, using 1st party cookies seems like the way to go. Members might log in from different locations, but we're mostly interested in prospects.

I don't expect prospects to look at our site from different computers, so most returning visitors will visit twice or more from the same computer. Obviously you can never get the exact numbers.

It also depends on the goal and target audience of your site.

If you keep your standard for measuring unique visitors your numbers might be a little off, but you can still observe changes in visitor behaviour.