We get a boatload of questions from our clients when switching or comparing analytics packages and reports. Those questions are easily addressed, but sometimes we come across a real stumper that makes no logical sense and causes use to lose confidence in our analytic packages. For this post, I’m going to pick on Google Analytics.

In a limited test of a particular landing page we received the following results from Google Analytics during the timeframe of February 1st, 2010 and March 17th, 2010.

Aside from being unimpressive; it’s also not trustworthy. (Below) The view of the “conversion funnel” for the same set of data. Note the discrepancy between the number of conversions / forms submitted?

To check our sanity we compared the data to Google Adwords. AdWords reported 5 conversions; ok at least that is consistent with the first graphic. But here is the crux, Google AdWords *should* have been reporting a higher conversion rate because we tested the forms multiple times from several IP addresses. Google Analytics is filtering out our tests based on a few selected IPs, therefore Analytics’ numbers *should* be lower. The second problem is the change in results data from switching views from the “Goal Overview” page to the “Funnel Visualization” page. Why would Google Analytics show 5 in one view and 1 in the other view, using the same data set?

We’re not the only ones struggling with Google Analytics apparent data discrepancies. It’s discouraging that problems like this go unaddressed by Google. I understand the service is free, but Google’s lack of addressing these problems on their forums is exactly why more corporate clients have returned to tools like Webtrends, Woopra, Clickly and Mint.