Inforegister’s Credit Score is now an internationally recognised credit management tool in use in a number of foreign credit insurance groups. The Credit Score was developed in co-operation of data scientists from Softwade Development and Applications Competence Center and the University of Tartu. We used machine-learning methods in developing the model, including unsupervised learning algorithms that can discover hidden connections and based on these predict whether a company is likely to become insolvent. The model is highly accurate – 99.5%!