Implementation of the advanced cost-effective methodologies for product and/or process quality improvement is an effective mean to fulfil or exceeds customer’s expectations. This paper presents the analysis of a performance of automatic enamelling process for a non-normal data distribution, conducted within the six sigma project implemented in a production system. Drawing on the process analysis results, process optimisation was performed using location and dispersion modelling. It proved its effectiveness in determining the significant effects of process factors on the response mean and variation, and in obtaining the optimal factors setting of the observed single-response process.