Got Google Analytics running on one of my websites. Google defines a
pageview as the most occuring event. Every refresh, every link clicked
to a different part of the page, even loading from cache is an
additional pageview
http://www.google.com/support/analytics/bin/answer.py?hl=en&answer=57164
Now it should be known that GA is JS based, and will report directly to
Google from the client’s browser that page X has been visited. This is
why it can also count hits from the client’s cache.
Now I have written my own little banner rotation system in one of my
projects, which also contains a viewcounter.
For testing purposes I have only 1 banner, and it is displayed on every
page that is also tracked by Google Analytics. The viewcounter for my
banners works serverside, and actually counts the times the banner was
requested (times rendered on serverside to be precise)
Now over a period of a month GA has accumulated around 150 views, where
my banner system reports close to 1100. Almost a factor 10 difference!
Now obviously both ways are pretty different. For example, my banner
system will miss hits from the cache, where GA’s JS will be blocked by
browsers that have JS disabled or are incapable of. And my system will
also count visits that are terminated before the website got fully
loaded where GA’s JS will only be executed at the complete page load.
Has anyone got any experience with pageview counters and can you tell me
if a factor 10 difference is acceptable? Because to me it seems it is
quite a big and unreliable difference.