A widespread shortage of skilled employees, which also includes the increasing difficulty to retain talented employees by limiting voluntary employee turnover, impedes company success. To gather insights on voluntary employee turnover, previous research has conducted explanatory research focused on theory-driven hypotheses, or utilized predictive models, aiming to predict which employees are most likely to leave the company based on historical data. In contrast, we propose a methodology that combines predictive data mining methods, causal decision trees, and an expert validation to yield firm-specific actionable explanations for employee turnover. The proposed methodology is applied to a real-world case of a mechanical engineering company. Here, our data-driven causal analysis identifies patterns of voluntary service technician turnover, which are validated by domain experts and used to derive targeted measures for reducing future employee turnover. The results are shown to provide valuable insights, adding to the a priori knowledge of the experts, revealing discrepancies between subjective opinions and quantitative results, and substantially informing the company's decision-making. A follow-up study demonstrates that almost all of the derived measures are being realized and indicates early positive effects on employee turnover.