The advancement of technologies and the enormous volume of available data have made Data Science (DS) indispensable for addressing complex social issues. In the public sector, Evidence-Based Public Policies (EBPP) seek to support decisions through data analysis, increasing transparency and governmental effectiveness. However, the adoption of Data Science in public policy faces obstacles such as data quality, the need for collaboration between different areas, and institutional resistance. This article proposes the BEPP-DS methodology, which structures the entire process—from problem identification to policy evaluation—with a focus on transparency, reproducibility, and scalability. The model serves as a reference for governments wishing to use data science to build more effective policies, promoting the use of artificial intelligence, advanced analytics, and citizen engagement in data governance.