The potential capabilities of computers to support analysis of interaction data have attracted the attention of the CSCL research community. This has led to the proposal of a number of interaction analysis tools, which process interaction data to meet different purposes. These may range from supporting researchers in ethnographic studies to providing advice to the students. However, after several years working with classroom-based CSCL experiences, we have found that both researchers and practitioners meet many difficulties to apply these potential benefits to their CSCL settings. Thus, the first goal of this chapter is to provide a systematic analysis of the problems that can be found when trying to apply interaction analysis tools to CSCL settings, which are then classified at into three levels, namely: application, architecture and design levels. Then, we outline the path for possible solutions to face these problems. According to this, the issues identified at the design level call for an IA-aware design process where we distinguish between co-design approaches that directly integrates the diverse needs of learning and analysis, and multi-perspective approaches that treat them independently at an initial stage. On the other hand, the problems at the application and architecture levels must be faced by technology-driven solutions, such as the use of decoupled architectures, either based on inter-process communication or on interchange of log file information. Several open issues have also been detected that need adequate solutions, as e.g., the semantic integration of log-files when multiple self-contained learning tools are used for an integrated analysis.