To answer the research question of how semantic technologies can substitute and/or complement the manual coding of qualitative data, an experimental hypothesis testing is presented in the domain of customer satisfaction within a financial services company. Based on a traditional analysis of positive and negative customer statements, the use of semantic technologies is evaluated. On one hand, the utilization of existing ontologies and categories from the semantic web for substituting manual coding or improving predictive accuracy through semantic annotations is studied using different models. On the other hand, the semantics of these qualitative data are successfully incorporated into the analysis using the probabilistic topic model method. The results of the various approaches and methods are illustrated in comparison. The evaluation of the predictive accuracy of the models is tested using an evaluation dataset, and the differences regarding statistically significant deviations are analyzed.