We started our journey by trying to understand why ie9 had higher figures than ie10. As we also have an analytics on the server side, we extracted a raw log. W

We started our journey by trying to understand why ie9 had higher figures than ie10. As we also have an analytics on the server side, we extracted a raw log. We used our existing tools that can filter "fake" or "foe" trafic by its behavior. We then manualy analyzed representative filtered samples (still by analyzing its behavior). It took some time, but was worth it, because we found an opposite conclusion than GA: most of the real traffic ($) from our ie9-10 samples is on ie10. So now the dev. guideline is: forget ie9 and make it not too ugly/broken on ie10. ;) PS: We had to do a manual work of filtering because our tools do not filter all the traffic that we know being fake or foe because we often have false positive.