Software is often poorly captured in economic statistics because of at least three main reasons. First software is evolving rapidly to take on new forms (e.g. software as a service). Second, it is a non-physical good and a part of it is created and exchanged outside the monetary environment. Last, there are lots of cross-border innovation activities and exchanges with software products. In order to develop policies that support software innovation appropriately, it is necessary to better understand several critical aspects of the innovation process. More emphasis on improving of measurement and data collection can help strengthen the understanding of policy development for software.