Managing Unstable Demand in Manufacturing Systems

(Part 3/5) Plans to Policies Welcome to Part 3 of my mini-series on "Optimization Under Uncertainty." Acknowledging uncertainty requires us to move from producing a one-time “plan” to developing policies that guide decisions as new information arrives. A policy is not a single solution but a rule or mapping that adapts decisions in real time based on the current state of the system and the information available at that moment. For example, consider inventory management: 🔹 A plan might say, “Order 500 units of Product X now.” 🔹 A policy might say, “If inventory falls below 200 units and the supplier lead time is 5 days, place an order to bring inventory up to 600 units.” The policy reacts to the current inventory level (a state variable) and can adjust as demand, lead times, or supplier reliability fluctuate over time. This approach is critical for: 🔹 Dynamic pricing: Adjusting prices based on demand signals and remaining inventory. 🔹 Routing: Updating delivery routes in response to traffic and new delivery requests. 🔹 Capacity planning: Shifting production schedules as new orders and supply disruptions arrive. It transforms forecasting, simulation, and optimization into an integrated decision-making system that adapts to uncertainty rather than being broken by it. In essence, optimization under uncertainty is not about producing a plan; it’s about learning how to decide the plan. Policies allow you to operationalize uncertainty into systematic, responsive actions rather than brittle decisions based on point estimates. In the next post, I’ll explore how to build these policies practically and evaluate them systematically, showing how to test and improve your decision rules over time.

Material and Information Flow Analysis, or #MIFA, is a simple but powerful way to look at lead-time inside a plant—not by starting with production, but by focusing first on #withdrawal and #conveyance, and what information triggers these movements. In many factories, materials are moved around without much thought, often in big batches and then laid to rest where they arrive. MIFA starts by asking: Who takes what, from where, and when?—and just as importantly, What signal tells them to do it? Toyota engineers learned something surprising: if you increase how often you withdraw parts—even without changing batch sizes—you actually reduce the inventory you need to hold. Why? Because frequent withdrawal creates a natural rhythm that stabilizes the flow. It makes problems more visible and forces teams to react faster. Of course, to go further, you eventually have to reduce batch sizes as well, to move closer to true single-piece flow—the lean ideal. MIFA invites you to ask a few key questions: 1️⃣ First, does your production pace match your sales pace? If not, you’re probably building too much of what you can’t sell and not enough of what you can.  2️⃣ Second, are the product references going to a shop stock stable, or are they unpredictable? Unstable references drive up buffer stocks and chaos.  3️⃣ Third, is the production cell itself reliable enough to run in smaller batches, or even one piece at a time? These questions shift the improvement focus from production efficiency to flow efficiency—which is where lead-time really lives. MIFA doesn’t just map the plant; it reveals the logic behind the flow, shows how most material issues are created by how the information is structured and shared (what are the #timetables? What #increments? How stable is the #demand?), which helps teams redesign their system to be faster, leaner, and more responsive—by seeing what triggers what, and whether the rhythm matches customer demand. The deep insight behind the MIFA is that you control your plant through logistics first, rather than actual production, because a plant's real job is to deliver (good parts on time) rather than simply produce. #LeanIsAwesome

Companies are in for a double whammy. Tariffs are back, so everyone’s scrambling to buy inventory before costs jump. But at the same time, the economy is volatile and demand might be softening. Stockpile too much, and you’re sitting on cash you can’t move. Buy too little, and you’re paying a premium later. We’re already seeing the impact—inventory levels are expanding at their fastest rate since 2022, but costs are surging even faster. Warehousing prices spiked 18 points in late February, and firms are holding onto stockpiles rather than moving product. This shift from just-in-time to just-in-case inventory is putting serious pressure on storage capacity and operating costs. ➡️Buy what matters, not everything. Lock in critical, high-turn items, but don’t let panic buying clog up your balance sheet. ➡️Keep pricing flexible. If demand slows, offer strategic discounts or bundles to move inventory without slashing margins. ➡️Update forecasts aggressively. Don’t rely on last quarter’s data—adjust plans every 30-60 days to stay ahead of shifts. ➡️Cut Waste, Not Growth. Instead of knee-jerk cost cutting, optimize freight, reduce carrying costs, and streamline production to stay profitable. Stay agile. Stay resilient. Stay liquid. [Image source: Logistics Managers’ Index (February 2025)]

Forecasting Alone Won’t Save Supply Chains Anymore We’ve been chasing forecast accuracy for decades. And yet, supply chains keep breaking. Here’s why: Forecasting can’t beat variability. When demand spikes, forecasts react too late. When demand drops, forecasts overshoot. Every adjustment creates instability upstream. This is the bullwhip effect in action. It’s not new. It’s not rare. It’s embedded in traditional planning systems. The system amplifies noise: → Customer raises demand by 10% → Planner adds 15% safety → Supplier reacts with 25% → Their supplier panics with 40% Every actor is rational. The outcome is chaos. DDMRP changes the equation. Instead of chasing accuracy, it absorbs variability. Strategic buffers at decoupling points do the heavy lifting. They isolate variability before it cascades. They dampen demand signals instead of amplifying them. The results are counterintuitive but proven: → Less overall inventory → Higher service levels → Smoother supply signals upstream Not because we forecast better. Because we stopped relying on forecasts to solve the wrong problem. Buffers don’t eliminate uncertainty. They control it. After 60+ years, supply chains don’t need better predictions. They need shock absorbers. Time to stop amplifying the noise. Time to design supply chains that absorb it. #SupplyChain #DDMRP #DemandDriven #BullwhipEffect

“Sir, your factory runs 30 days — irrespective of orders or production.” That’s what I told the MD of a plastic molding factory who wanted to increase his productivity. He looked surprised. So I asked a few questions. 🧩 “What’s your plant capacity?” ➡️ “200 tons per month.” 🧩 “How much did you produce last month?” ➡️ “120 tons.” 🧩 “How many days did your plant run?” ➡️ “30 days.” Before that month? 180 tons. Before that? 210 tons. Still 30 days each time. That’s when I explained Parkinson’s Law — “Work expands to fill the time available for its completion.” So whether a factory produces 120 tons or 210 tons… it will still consume the full month — same manpower, same energy, same overheads. The only sustainable way I’ve found (after observing 200+ Indian factories): ✅ Always have demand for your factory of at least 85% capacity. When the factory is loaded, People work with urgency Systems stabilize Productivity automatically improves The MD said, “But my business is cyclic, I don’t always have orders.” I replied, 👉 “Then work on making your business secular — not seasonal.” He took it seriously. Started improving his NPD and Sales processes. Within a year, his monthly demand stabilized around 220 tons. 📈 Profit up by 3x. 💭 Reflection for factory owners: If your plant runs 30 days every month, ask yourself — Are you truly underloaded, or are you running on Parkinson’s Law?

🚀 Why Traditional MRP Falls Short in a High-Variability Supply Chain World In the dynamic landscapes of modern supply chains, where volatility, uncertainty, complexity, and ambiguity (VUCA) dominate, traditional Material Requirements Planning (MRP) systems often struggle to keep pace. Here’s why: 1. Reactive Nature: MRP systems typically operate on a push-based model, relying on forecasts that can quickly become outdated due to fluctuating demand and supply conditions. This often leads to either excess inventory or shortages, increasing costs and reducing responsiveness. 2. Lack of Flexibility: MRP is structured around fixed lead times and order quantities, which can be inefficient under rapid changes in market conditions. Adjusting these parameters can be time-consuming and complex, making it difficult for businesses to adapt swiftly. 3. Poor Response to Variability: MRP does not inherently manage variability in supply and demand. Without the capability to dynamically adjust to changes, businesses can face significant disruptions. 🌟 Enter Demand Driven Institute #DDMRP – The Game Changer for Managing Supply Chains in a VUCA World Demand Driven Material Requirements Planning (DDMRP) offers a robust solution by addressing the core shortcomings of traditional MRP: 1. Visibility and Responsiveness: DDMRP uses strategic decoupling points to buffer against variability, providing real-time visibility and faster response to market changes. 2. Demand-Driven Approach: By prioritizing real customer demand rather than forecasts, DDMRP reduces the risk of overproduction and underproduction, aligning inventory levels more closely with actual market needs. 3. Simplicity and Agility: DDMRP simplifies the planning process and enhances flexibility by focusing on flow and continuous adaptation, making it ideal for volatile environments. In conclusion, as businesses navigate the complexities of modern supply chains, DDMRP stands out as a clear winner, offering the agility and efficiency needed to thrive in a VUCA world. It’s time to move beyond traditional models and embrace a demand-driven approach for sustainable competitiveness. For more information of Demand Driven MRP education, consulting and software implementation contact WA Solutions SAS : John Melbye, DDPP, DDOP, DDLP, DDDP, CSCP : USA Derk Kuiper : Europe Gustavo Benítez Cárdenas Master CPIM CSCP Master DDPP DDLP DDOP: Mexico Juan Camilo Trujillo Cadavid : Colombia Hugues Boritiyca Silva : Brasil Herbert Braun Valle : Guatemala #SupplyChainManagement #DDMRP #VUCA #AgileSupplyChain #DemandDriven #wasolutions #avaluechain #inventory #customerservice #ddpp #ddlp #ddsop #adaptivesop 🚦

You can't solve bullwhip with better forecasts. You solve it by stopping the signal. Every supply chain director I meet is fighting the same battle: demand spikes at retail, panic ripples through distribution, manufacturing overreacts, suppliers scramble, and six weeks later you're drowning in inventory you don't need. We call it the bullwhip effect. And most companies are trying to solve it with more accurate forecasts, better communication, or faster data sharing. They're treating the symptom, not the cause. Here's what actually creates bullwhip: Fully connected planning systems. In traditional MRP, every demand change propagates through every level of your supply chain. A 10% fluctuation at the customer becomes a 20% swing at distribution, a 40% variance at manufacturing, and an 80% chaos event at your suppliers. The system is designed to amplify noise. The solution isn't better forecasting. It's strategic decoupling. DDMRP places buffers at carefully chosen decoupling points. These buffers don't just hold inventory - they absorb variability and stop demand signals from cascading. When demand spikes, the buffer catches it. The signal stops there. Everything upstream continues operating in its normal range, driven by buffer consumption rather than demand amplification. Think of it this way: → MRP = fully connected amplifier (every signal gets louder as it travels) → DDMRP = strategic circuit breakers (variability is absorbed locally) Buffers operate in ranges, not fixed points. They're designed to flex with demand while protecting upstream operations from the noise. The strategic shift: Stop trying to make your forecast perfect. Start identifying where to decouple. The companies winning on service and inventory turns aren't predicting better. They're absorbing variability better. What's your approach to managing demand variability - better forecasts or strategic decoupling? #DDI #DemandDriven #DDMRP b2wise