So I have to do some major research using Google Analytics for a site I am inspecting. I've come to the conclusion that there isn't anything really simple about the application because it's very difficult to understand the assumptions. I am constantly bombarded by people giving me stats that seem contradicted by the next graph on the list and guess that much is because people don't really understand what they are saying. To some extent, that has included me.

I've spent the better part of a week trying to figure out how to meaningfully assess traffic and I've discovered that in many instances it is best to look at just one day and that will give you an idea of some consistency of users. But again, there are some assumptions to be made. For example, take the Visitor Loyalty graph which says:

"This graph tracks the number of visits by visitor frequency. The number of visits by visitors who never returned (no loyalty) are indicated on the left of the histogram. The number of visits by visitors who returned over 200 times (very loyal) are indicated on the far right."

Now I have to guess that the number of visitors who returned over 200 times means it's over the entire range of when Analytics was first installed in January of 2006. The problem with this is that if I measure the entire range from install to current date, I get far more visitors than I do absolute uniques, which makes no sense. After all, those numbers should add up.

Another example, of those visits we just read about, what is a visit? Are we talking about 30 minutes of inactivity and then another action? Another session? Etc.

I haven't really seen anyone that can tie together all the assuptions that are being made and go far beyond the simple text explanations there. I'm curious if anyone has seen a source or manner to get the answer to these questions instead of taking stabs at what these numbers really purport to be.