This document discusses how data science projects often fail due to a lack of business adoption and the gap between data and business needs. It provides statistics showing that 85% of big data projects fail and 80% of AI projects do not scale within organizations. Common reasons for failure include solving the wrong problem, having the wrong data or skills, and not clearly defining the business purpose. The document then discusses how organizations are addressing these issues through design thinking, design sprints, and focusing on actionable insights and simple, prototype models. It provides an example of a predictive maintenance model that scheduled maintenance to detect HVAC failures and improve customer service.