KND vs Google Analytics


The Problem

A strange query came in the other day from one of our clients. They wanted to know why our custom statistics package was reporting more click throughs to their site than Google Analytics was. The margin of error I may add was in the order of 30:1.

Now seeing as I had personally written the statistics package and knew exactly how it worked I was more than intrigued by such a massive discrepancy. After a rigorous code review and trawl through the database I reconfirmed that my code was fine. I even logged time of day and user IP address, and utilized a PHP page not JavaScript to collect the stats.

I am left with an interesting credibility problem, could it be that there is something wrong with Google Analytics? But who is going to take my word for it? I mean thousands and thousands and thousands of sites use Analytics. How do I explain to the client that I am right and Google is wrong. It turns into a silly game of technical he said she said pretty quickly. Without the clients website server logs to back me up I am at a bit of a loss to prove my case beyond reasonable doubt.

And let’s face it, I am me, and Google well quite frankly is GOOGLE, our internet search lord and master.

I talked with the kids in the office and discussed several scenarios as to why a Javascript tracking system could be flawed. They raised some great points about JavaScript being disabled, and the Google code relying on a ONLOAD event to trigger. (So if you have a page full of content you could conceivably click away from the page before the page has loaded and before Google can collect the stats) But could this account for such a ridiculous disparity in numbers? I was skeptical.

I could not be sure that the Google code was always in the page nor that it has always been included correctly either.

The Challenge

Not being one to shy away from a challenge I thought I would conduct my own test using the KND code research project site

I would set up Google Analytics on the Index page of this site and customize the apache logging to collect maximum stats.

Then I would construct a series of tests to see whether I could create the same discrepancies.

Keep you posted…

I have just set up the Google Analytics on the site and I am creating the tests. I will post the full results of these tests and testing technique on this blog as I collect them.


Related Articles