In this paper, we create a research model addressing the antecedents of loyalty: service quality and cost reductions. The data was collected from Finnish industrial companies, buyers of the logistics service provider's services in the B2B domain. We found that service performance influences loyalty, but establishing a cost performance-loyalty influence is more problematic. We suggest that buyer heterogeneity gives rise to an insignificant relationship between cost performance and loyalty. Hence, we have extended our work with finite mixture structural equation modelling to reveal latent groups and their background factors: perhaps for the first time in the domain of logistics research. The paper provides new insight into the opportunities for customer analysis and segmentation, with powerful, up-to-date methods that are gaining a growing reputation within the research community. We found two latent loyalty classes, and in addition, as traditional customer segmentation seems to be insufficient, we suggest factors which explain latent behavioural classes.