In recent years, small and medium-sized enterprises (SMEs) have increasingly adopted Social Media technologies with the purpose of fostering the bidirectional communication with customers or to facili-tate the collaboration between employees amongst each other. Thereby, customer posts in a company’s Social Media channels capture consumers’ current attitude towards product and service offerings or the enterprise as a whole. An automatic analysis of these posts does not only provide a firm with valuable knowledge on the customer relationship, but also frees up human resources in case the posts were screened by employees manually hitherto. However, posts in Social Media channels of SMEs are char-acterized by certain peculiarities such as regional slang or off-topic discussions amongst others. The study at hand investigates the impact of such characteristics on the accuracy of results received from an automatic sentiment analysis of corresponding posts. In this context, we revert to Social Media posts of five SMEs from southern Germany. The results show that an adaption of approaches used for senti-ment analysis to the specific language of customers and firms is mandatory for achieving a high level of accuracy.