Finding a common ground between academic and industry training is a challenging task. However, despite all the differences between the two worlds, there is one important common aspect for the master students and industry trainees in the field of conceptual data modelling: the mistakes they make when creating data models. This idea paper describes the existing differences and similarities between two academic and industrial training approaches for teaching conceptual data modelling. We propose ideas for improvement of training quality on both sides by analysing and tackling the ground errors in conceptual data modelling made both by novices and professionals. Additionally, we propose the ideas for methodology exchanges between the two training types.