The way in which people communicate affects their relationship building, social network structures and ultimately the knowledge they receive through their connections. For organizations, an effective knowledge exchange among employees is crucial for the competitive performance. Therefore, companies rely increasingly on social media platforms to facilitate communication and collaboration. To enhance our understanding of successful communication in enterprise social media, we apply human coding and quantitative analysis to the content and tone of 15,505 enterprise microblogging messages created by 1,166 employees of an international financial service provider. Specifically, we develop a model-based operationalization of communication styles and empirically derive two general communication types. Analyzing the quality and number of answers to questions from these communication types, we identify differences in the effectiveness of knowledge exchange. Our results suggest that a more factual-oriented communication type benefits from a higher knowledge exchange effectiveness compared to a primarily self-disclosing communication type.