This article consists of a post-analysis of twenty-four learning games, developed by master students within a dedicated course, over a period of twelve years. From this longitudinal qualitative analysis of serious game development, we are able to identify critical factors that make these games successful or not. We have grouped these factors into three axes: project organization, game mechanics and graphics. The variety of games produced in this academic context opens the way toward a more systematic analysis of this corpus, aiming at producing useful guidelines to design better serious games in an efficient manne