Accuracy of Log File Analysis

Once a log file is poked and parsed is it worth anything?

[justgowithit]

I’ve got a partner that sends me referred traffic reports (them to me). The last report showed %99 more traffic than I can find when analyzing my raw logs.

To all you tracking buffs out there :)  - What value do you place on raw logs and analyzers?

[ogletree]

How much detail are they giving you? Is it just raw numbers? How did they get this information? They may be counting bots. There are lots of reasons why the referrer string is not logged. Using a tag based system like G analytics would give you pretty much the same data. I would expect there to be a difference between their records and yours. They know when somebody leaves their site. Some visitors don't show a referrer when they arrive at your site.

Server log file reports and tag based reports both have their strong points. Server log file reports include lots of information from files that don't contain the tags (images, pdfs, etc...) as well as information about users that don't run javascript. A tag based system will have more accurate user data. The server log file reports have to guess what a user is doing while the tag based system knows who the user is and what they are doing exactly.

[justgowithit]

How did they get this information?

This (critical piece) is yet to be seen.

Thanks for the run-through. I'm well-aware of the inherent variance with tracking in general. However, with this particular project my tracking data is limited for reasons beyond my control and I've been reduced to analyzing raw logs - not something I have a ton of experience in.

[cgrantski]

If you look in your logs at the small number of known referred visits (from the affiliates) and if for those visits the browser is never IE, but instead is FF etc, then you've got a blank referrer field situation and the numbers they are giving you could be correct.

Likewise, if you go to those affiliates and click on one of your ads, and if it opens in a new browser window, you have that problem also.

This is my best guess, give-the-benefit-of-the-doubt solution to the huge discrepancies you are seeing.

[justgowithit]

Likewise, if you go to those affiliates and click on one of your ads, and if it opens in a new browser window, you have that problem also.

That's it.

Wow.... I've made it this far without knowing that IE sends no referrer with a new window. Granted, tracking is not my specialty but I must say I'm a little embarrassed.

Thanks cgrantski.

[cgrantski]

you shouldn't be. it's actually a pretty obscure fact. Only hard core analysts like me appreciate it. It's one of the reasons why analytics is/should be a full time job.