Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
People can learn a lot through (collaborative) reflection at work: In organizations, staff debate experiences and due to issues every day, thus reflecting together and learning from each other. While this is desirable, it is often hindered by differences in time and space. Online discussions in community-like systems may provide a means to overcome this issue and enable staff to share experiences and learn from them. In this paper, we study two different data sets from two software systems to support online collaborative reflection in order to analyze what possible factors influence the occurrence of aspects of reflection in online discussions, like experiences, suggestions and learning. Our results include findings that partially approve existing models of (collaborative) reflection, but also others that add to or even challenge these models. Overall, we found that collaborative reflection may take different paths towards learning. From this we derive that facilitation mechanisms need to take up these paths, and that facilitation may be successful in different ways than anticipated from existing literature. We describe possible facilitation mechanisms and discuss their implementation.