Defect causal analysis (DCA) has shown itself an efficient means to improve the quality of software processes and products. A DCA approach exploring Bayesian networks, called DPPI (Defect Prevention-Based Process Improvement), resulted from research following an experimental strategy. Its conceptual phase considered evidence-based guidelines acquired through systematic reviews and feedback from experts in the field. Afterwards, in order to move towards industry readiness the approach evolved based on results of an initial proof of concept and a set of primary studies. This paper describes the experimental strategy followed and provides an overview of the resulting DPPI approach. Moreover, it presents results from applying DPPI in industry in the context of a real software development lifecycle, which allowed further comprehension and insights into using the approach from an industrial perspective.