Cognitive diagnosis models received growing attention in recent psychometric literature in view of the potentiality for fine-grained analysis of examinees’ latent skills. Although different types and aspects of these models have been investigated in some detail, application to real-life data had so far been sparse. This paper aims at addressing different topics with respect to model building from a practitioner's perspective. The objective is to draw conclusions about examinees’ performance on the Austrian baseline testing of educational standards in math 2009. Although there is a variety of models at hand, the focus is set on the easy to interpret deterministic input, noisy ‘and’ gate model. A possible course of action with respect to model fit is outlined in detail and some conclusions with respect to test results are discussed.