Webtrends dodgy stats

[adamcmoran]

i think this may have been answered at some point. But does anyone else have the feeling that Webtrends statistics are a little dubious, slightly on the high side to say the least. If anyone has any views on this matter it would be appreciated. And maybe why they are high!

Thanks
Adam

[seth_wilde]

If you get a lot of your traffic from AOL your probably seeing inflated numbers. Because of the way AOL's proxy cache works most stats packages count one visitor as 3 or 4.

[tedster]

Deutsche Telekom also uses a similar scheme -- dynamically assigned IP addresses for various page element requests from the same visitor. I've heard there are a few more big ISPs using this "method" as well.

[ritualcoffee]

also consider this...

a couple months back i noticed a 5X jump in monthly views - turns out that engineering had let loose a testing spider to monitor the site - thus inflating my stats...of course I wasn't told..found out the hard way.

I've always found Webtrends to be on the high side but, most of the tracking tools are "best guesses"

[guysmy]

try [stats4all.com....] I dig 'em.

[rogerd]

I tend to agree with Seth - I've seen some Webtrends reports where the number of visitors exceeded the number of page-views! (There are several things that can cause this, but in this case it appeared to be AOL using as many as 10 IPs to pull the images from a single page.) A cookie-based approach is probably more accurate, although cookies aren't perfect, either.

[netcommr]

...kind of off topic, but may be of help to someone

Something to try if you want to see actual page views even if a cached page is viewed by an AOL user or system cache by hitting the back button.

1.use javascript to generate a random number.
2.use javascript to document.write an image tag.
3.the filename of the image includes the random number.
4.the url of the image tag pointing to a server side script to record the hit.
5.server side script returns a 1 pixel gif.

works great for tracking ad impressions...

[retro]

I know both me & a m8 have seen some spurious results in WTL, although we've only used the freebie Personal edition so I don't know if that has any bearing. Country of Origin stats were very wide of the mark (the latest update to WTL has meant that us freebie users don't now get the functionality we once did :( ).
Worst of all was that for a site I've been keeping an eye on for 12 months or so, the results upto and including August have disappeared from the yearly view of 'hits-per-month' :(((

R.