Web Analytics: Logfile Analysis

Ok, so if you’re like some of the people I know, this might be your approach to web analytics…

Dr. RandE: “Hmm, I would like to know how many people are visiting the blahblah foundation website.  I’ll give my website guy, Chuck, a call.”

Ring…Ring…

Chuck: “Hello”

Dr. RandE: “Hi Chuck, Dr. RandE here.  I need to know how many people are visiting the blahblah foundation website.  Can you help me do that?”

Chuck: “Sure, no problem.”

Chuck might be a great website guy and there is no doubt he will give Dr. RandE the numbers he wants, but depending on how Chuck gets these numbers, they could be very different.

There are two main methods used in Web Analytics, page tagging and logfile analysis; like all methods, neither are perfect.  Each have their pros and cons but you should know the difference.  We’ll talk first about Logfile Analysis.

Every website lives on a server.  That server collects data and records it on a log.  This log keeps track of the files being accessed (your website is just a bunch of files) and who is accessing those files.  Logfile analysis, is literally just the analysis of these log files.

Logfiles are indiscriminate in what they track.  Anything that pulls a file from the server is recorded, this includes search engines and a variety of bots that spend their time patrolling the web.  Logfile analysis programs use IP addresses to determine unique visits.  IPs can be shared by more than one person, and in some cases they are shared by large groups of people.

Ever heard of a cache?  The thing you occasionally empty on your web browser.  Well, when your web browser stores info you basically don’t need to access the web server each time you visit a website.  Sometimes you get the info a second time from your computer, meaning that visit is not always shown in a logfile.

Logfile analysis is server focused, so it’s great at figuring out what’s happening with the server.  Things like how much bandwidth is being used and if the site is being effectively crawled by search engines.  Dr. RandE’s web guy Chuck might use it all the time.  Of course, that doesn’t mean it is the appropriate approach for your needs.  There are cases when logfile analysis is the right way to go, but it’s better for you to make that decision knowing that there is a choice.