At Decision Spot, we specialize in assisting clients on their supply chain transformation journey using AI/ML, Decision Intelligence, and other mathematical tools. To help you establish an efficient, effective, and future-ready supply chain, we have compiled a set of recommendations:
Set clear goals and define measurable success: It is of utmost importance to set clear, long-term goals and effectively communicate them throughout the organization. While setting goals is a subjective process, it is essential to establish objective metrics for measuring success. This may involve a series of iterations to ensure comprehensiveness and effectiveness.
Challenge existing processes: While supply chains often focus on optimizing individual processes for efficiency and agility, true global optimization requires considering an end-to-end network perspective and challenging existing processes. Supply chain network design models have proven effective in evaluating strategic and tactical decisions holistically. Here are a couple of recommendations regarding these models:
Optimal Baseline: Network models offer significant value in optimizing supply chain network footprint and capacity expansion decisions. However, even the process of building these models and running an ‘optimized baseline’ provides many insights and end-to-end visibility into your supply chain, including true cost-to-serve a customer, identification of inefficient product flows, and bottlenecks.
Repeatability: Building a network design model requires time and effort, particularly in data validation, input data preparation, and results interpretation. Automating this process is essential to enable easy model reruns with changing timeframes and parameters. This allows running multiple what-if scenarios to make more robust decisions. Moreover, tactical network optimization models can become the backbone of the Sales and Operations Planning (S&OP) process.
Automate processes: Identify decisions made through manual processes (spreadsheet analyses, dashboards, emails, meetings) in your supply chain planning and execution, and automate them. This doesn't mean complete automation of the supply chain, but rather involving humans to address and override exceptions instead of performing routine calculations. Process automation provides the following benefits:
Enhanced efficiency: Code can perform calculations far more efficiently than humans, so leveraging code accelerates processes. The gained efficiency can be used to evaluate what-if scenarios in planning use cases.
Improved accuracy: By minimizing human errors, individuals can focus on handling exceptional cases that may not (yet) be handled by code.
Streamlined information: When people transition out of their roles, a lot of important information tends to get lost along the way. Embracing automation helps manage data effectively and record valuable information in dedicated databases. This change ensures data integrity, reduces the risk of losing crucial information, and provides a reliable foundation for effective decision-making.
Implement intelligent processes: Simply replicating existing processes and logic into code would be a missed opportunity. Utilize this opportunity to apply Decision Intelligence approaches, enhancing the science of decision-making. In a previous blog post, we discussed an approach for categorizing project opportunities based on their impact and frequency of use. For high-impact and/or high-frequency use cases, leveraging AI/ML and Decision Intelligence approaches is recommended. For low-impact and low-frequency use cases, automating existing logic is a reasonable option, particularly if data science resources are limited.
Create a chain of models: No single model can make all supply chain decisions. It requires a series of strategic, tactical, operational, and execution-level models working in harmony to optimize supply chains holistically. Focus on selecting the appropriate granularity for decision-making within these models. Ensure accurate assumptions and seamless data flow between models and fine-tune them to achieve the global optimum rather than local optima.
The Walmart network design team, the recent winner of the Edelman Prize, shared their two-tier model framework highlighting the tradeoffs between strategic and tactical decisions. My mentor, Mike Watson, shared valuable learnings from this video in his blog. According to Mike, “When they (Walmart) iterated through different scenarios, they could find optimal solutions for the system but not for any one component—a good reminder of all the trade-offs in a supply chain.” Treat each model as a micro-service connected with other models to build a chain. Once this chain moves your end-to-end process efficiently, focus on identifying and replacing the weaker links rather than replacing the entire chain.
Measure, detect, and tune: Establish an automated process to periodically measure the impact of models. Acknowledge that initial projections may deviate from actual results, and therefore, monitor input data and assumptions for any drifts. Categorize the root causes of deviations to facilitate continuous improvement. Supply chains are dynamic, so ongoing tuning and improvement of models are crucial for maintaining their effectiveness.
Achieving success in this entire process requires a blend of art and science. The required scientific knowledge and technology are already accessible and constantly improving, making them increasingly user-friendly. However, required art can’t be created by 21st-century “Dali” (yet?).