Data-Driven Supply Chain Optimization

Data-Driven Supply Chain Optimization

Q: What is data-driven supply chain optimization, and why is it important?

A: Data-driven supply chain optimization applies prescriptive analytics in mathematical programming and operations research (OR) to provide decision support for supply chain decision-making at the strategic, tactical, and operational levels.

By properly capturing the decisions to make (decision variables), the performance metric to optimize (objective function), and the requirements and/or limitations to satisfy (constraints), an optimization model can prescribe the best (or better) solutions for supply chain network design (strategic), supply chain configuration and production planning (tactical), and resource allocation, routing, and scheduling (operational).

Its data-driven feature makes it possible to separate a model from its input data, and to prescribe optimal solutions and recommendations that are adaptive to today’s changing and volatile business environment. Custom designed data-driven supply chain optimization tools, tailored to the unique business setting and decision needs of a company, are key to the company’s competitive advantage and success.

Q: What are the new trends and opportunities for data-driven supply chain optimization?

A: First, fast advancement of information technology (IT) and vast availability of big-data, in volume, velocity, and variety, make it possible to address and solve innovative supply chain optimization problems that were not solvable before. Examples include advanced manufacturing with Internet-of-Things (IoT), climate-smart food and agriculture supply chain, and efficient and resilient supply chains in healthcare, energy, and telecommunication.

Second, three dimensions of complexity for supply chain optimization are emerging: dynamics, uncertainty, and multiple decision-makers (game-theoretic setting). These call for the integration of multiple techniques in analytics and artificial intelligence (AI): descriptive, predictive, and prescriptive. Moreover, modern supply chain optimization applications must address multiple (often conflicting) performance metrics, e.g., efficiency, cost, profitability, equity, resilience, and sustainability.

Q: What is the best practice for developing industrial-strength supply chain optimization applications?

A: First, successful development and deployment of supply chain optimization applications require collaboration and concerted work of a team of subject matter experts, optimization modelers, and software developers, supported by stakeholders and leadership.

Second, incremental modeling is recommended for model building. That is, start with modeling the core decision needs to get a prototype for proof-of-concept; then progressively add new features and components with expanded complexity, e.g., those addressing dynamics, uncertainty, or the game-theoretic setting.

Last but not least, engage users and stakeholders from the beginning to the end. Seek their inputs and feedback for modeling building, use case, scenario analysis, graphical user interface (GUI) design, and most importantly, evaluation and assessment of the decision-support and managerial insights provided by the optimization application.