This document discusses best practices for developing data-driven products. It advises that data scientists should focus on writing batch processes and machine learning, while engineers build production frameworks. For a data science team to succeed, it requires competent management, cross-functional collaboration between data scientists, engineers, product managers, and investment in infrastructure. When hiring data scientists, companies should select for intelligence, practical skills, and communication over degrees or past employers. Successful data-driven products often have an elegant key insight, can be proven quickly, and use a batch processing framework.