The document outlines a 5-step approach for developing a sustainability data practice: 1) ideation to define goals and metrics, 2) technology development to select tools, 3) data centralization to pool and standardize data, 4) analysis and automation to extract insights, and 5) accessibility to share results. It also lists 10 golden rules for sustainability data science, including embracing open source technologies, pooling data, scaling through automation, and making workflows reproducible.