I attended two sessions at Evaluation 2011 where Google Analytics was discussed. In both, the emphasis was placed on how you could use GA in evaluation practice; questions you could answer and how you would go about answering them. At some point or another a hand would raise and a technical question would be asked. How do IP’s factor into the counts? How do you know these are unique counts? Is that an accurate number?
In one of the sessions, the term Squishiness was used to describe the unknowns in Analytics. The suggestion was to establish a baseline and compare within.
My opinion is that we need to understand the Squishiness. Analytics are not perfect numbers, but we should have an understanding of why they are not perfect numbers. I’m going to do run some tests over the next 6 months to explore the issue, hopefully coming up with some practical suggestions you could use to reduce the Squishiness, or at least understand it. Let me know if you would be interested in helping out. Also, if you know any good resources, please share.