Continuous Improvement has long been integral to business strategy, a focus intensified by the advent of Industry 4.0 and 5.0, where automation, data exchange, and digitalization have transformed operations. Data is not only a tool for analysis, but also a foundation for empowering employees to lead evidence-based continuous Improvement. Building on this premise and following the Design Science Research methodology, this Doctoral Thesis contributes two primary outcomes. First, it presents four original data-driven models designed to reveal “hidden” improvement opportunities, even within seemingly optimized environments. Developed and validated in real manufacturing and logistics environments, their core innovations include: a Bottleneck Detection model built under the assumption of minimal information; a Process Performance Analysis model that incorporates a static and a variation-oriented version of performance; a Root Cause Analysis model combining Machine Learning and eXplainable Artificial Intelligence techniques that indicate the most relevant variables that explain (in)efficient processes/workers; and a Human Performance Variation Prediction model emphasizing worker-centered process improvement. Second, the Thesis introduces ImproveXpert4.0, a digital platform to drive data-driven Continuous Improvement. It includes two supporting design elements, the DMAIC-PDCA Structural Flowchart and its High-Level Architecture. The flowchart illustrates the platform’s practical use and its integration with the developed algorithms, merging top and operational KPIs towards organizational alignment and cross-functional collaboration. The architecture provides an overview of the platform's technological components and interactions, supporting software and hardware specialists in implementation. The goal is to move beyond statistics-level dashboards to uncover hidden wastes and inefficiencies, enabling organizations to unlock untapped performance potential.