Emphasizing the importance of Health Management Information Systems (HMIS) as well as their overall weak performance in providing reliable data is becoming a stencil of the discussions on health systems in developing countries. Many efforts have been made to improve data collection in these settings. This talk presents ongoing reflections on ways to improve HMIS data usage, focusing on data management and data analysis innovations. Based on current research at the Institute for Health Metrics and Evaluation and at CSE’s Data Science Incubator, we present two examples of HMIS data leveraging. A first example will present how metadata from Excel spreadsheets have been used to compile and standardize batches of reports from Kenyan HMIS. A second example shows how data from Open Street Map can be matched with HMIS data from Nigeria to estimate geo-localization of health services. Finally, we will offer preliminary reflections on how the increasing availability of structured HMIS data changes the way information and decision making should be linked. Using Alain Desrosières’ typology of the use of statistics for policy making, we will propose the possibility of a Learning State as being most adapted to the data available through HMIS.