3 ms·
There is no single analytics solution that would be fully satisfactory. The ones based on web bugs (e.g. Google Analytics) slow down your site and usually have
by bd 18y ago
There is no single analytics solution that would be fully satisfactory. The ones based on web bugs (e.g. Google Analytics) slow down your site and usually have clunky, slow and/or limited interfaces. Additionally, better ones are getting expensive for more serious use (also, I prefer one time fee over subscriptions).
Local analyzers of server logs do not filter out well robots and also logistics of transferring and updating logs is a pain.
I use Google Analytics, though not everywhere. Almost always I supplement it with some sort of local log processing. Over the time I tried several log analyzing tools, but I kept using only Visitors [1] and WebLogExpert [2]. I remember Web Log Storming [3] was also good: very fast and it allowed to track individual visitors.
Though often I just end up doing some custom logs processing, either by some Python script, or (more common) by some combination of Unix command line tools, directly ssh-ed into the server.
For example, to get real-time referrers:
cat access.log | grep "GET / HTTP" | grep -v '"-"'
If there is a lot of traffic, just use "tail -n XXX" instead of "cat".
[1] http://www.hping.org/visitors http://www.hping.org/visitors
[2] http://www.weblogexpert.com http://www.weblogexpert.com
[3] http://www.datalandsoftware.com/weblog/ http://www.datalandsoftware.com/weblog/