I record everything to Google App Engine data store, then have a script to export the data into a simple text file on my machine that runs from cron. Then I can

I record everything to Google App Engine data store, then have a script to export the data into a simple text file on my machine that runs from cron. Then I can run Python scripts on my own computer to just go through the text. I get about 60MB of events / day. Soon I will have to start deleting old events from App Engine, because I have to pay Google for data store entities stored. At first I was planning on doing some clever incremental thing to get all the stats on App Engine, but it's a lot simpler if you can just have everything locally and rerun any analytics quickly when I make mistakes. It only takes a few minutes to load a week's worth of data and go through it, so doesn't make sense to spend days trying to be really clever about doing it on App Engine. My favorite thing to extract is kind of a life story of users. When a user joins, what do they do on average in the first 24 hours? How about the next day? This is a pretty different point of view than Google Analytics, because it is from the user point of view and not historical of the whole app. So does this point of view give you some new insight? At least in our case. For example we are now testing which of two profile box designs is better on our MySpace app. Just a simple stat would reveal that the new design has a better clickthrough rate. BUT, when I look at the whole story, it seems that users are more likely to remove the app when they have this new design, so end results may be worse. Additionally I use Google Analytics events tracking, but I end up looking more at the life story than those events, because it is difficult to pull out how many events happen per visit, or how many events happen per visit in first 24 hours since user joined etc.