What does webtrends miss out?

[musicales]

Does anyone know anything that webtrends misses out - our host says we're using 14GB bandwidth a day, where webtrends shows 2GB. There's very little email or ftp so we've ruled those out.
Thanks for any help

[jeremy goodrich]

I'd contact their support department, that sounds like Webtrends is missing out on *an aweful lot* of data.

[Receptional]

14 GB a day sounds like an AWFUL LOT of data transfer if there is little FTP downloads etc. Let's think - if the average page with pictures is 100K that is 140,000 page views per day - assuming EVERY PAGE and IMAGE was unique code.

If the site is dynamic, I would guess many more page views than that.

Unless your site is so picture heavy that you ARE the world wide wait.

I agree with Jeremy - they should justify their measurements. You may find that downloading the log files themselves or some web based backup might be driving up the bandwidth.

[musicales]

We do get a lot of page views - over 200,000 a day so I'm beginning to worry he may actually be correct - but I still can't see how Webtrends could miss the remaining GBs - it's all html, pdf, asp, standard stuff. Strange thing we had more like the 2GB per day figure on a different server and a totally different stats package.

[europeforvisitors]

I've noticed the same disparity between FastStats and Sawmill. FastStats might report 300+ Mb of data transfer in a day while Sawmill reports more than 600 Mb.

On the other hand, the disparity between the reported page views and unique visitors isn't very significant at all. (FastStats reports slightly more unique visitors on most days, while Sawmill reports slightly more page views.)

Sawmill's numbers for data transfer are usually very close to what my hosting service's control console shows, by the way.

[musicales]

Hi europeforvisitors
I've still not worked it out. The only news here is I discovered some strange folders with names that repeated themselves several layers deep. I looked some of folder names up on Google and it seems they may be warez related. I'm trying to see now if there's anyway some scumbag has been stealing my bandwidth - though I've no idea how to check that (I'm on a windows server, so no doubt everyone here will shout 'told you so'!)

[musicales]

europeforvisitors - one question - which of the two figures do you suspect is accurate (which for example looks like a sensible pageview/bandwidth ratio)?

[hayluke]

Could you get old of some other log analysis software and see what that comes up with? At least then you'd know if the problem is with Webtrends..

[jm_uk]

You should also take a look to see if your WebTrends bandwidth reports include the bandwidth that is being used for images in your pages.

Some log analysis software will exclude requests for certain types of files (such as images, frames, style sheets etc.) before reporting.

[europeforvisitors]

Musicales wrote:

europeforvisitors - one question - which of the two figures do you suspect is accurate (which for example looks like a sensible pageview/bandwidth ratio)?

Well, the disparity isn't in the unique visitor and pageview counts. Those counts are fairly close for both programs. The only big disparity is in the amount of bandwidth consumption that's being reported. Sawmill's figure is pretty much what my hosting service is reporting, so I assume it's correct.

EUREKA! It just hit me: I have FastStats set up to exclude graphics hits from its daily summary. So it's probably just excluding graphics from its daily bandwidth calculation, too.