【Opinion】Operational Uncertainty: The Mean is Mean but the Variance is Meaner

By John Saldanha , Sears Chair in Global Supply Chain Management at WVU John Chambers College of Business and Economics

West Virginia University

As a graduate student at my first meeting with an industry research partner I still recall my incredulousness when I realized that a popular enterprise resource planning (ERP) system missed a critical statistic for measuring the variance or uncertainty of an operational process. The academic literature often correctly assumes that some types of operational uncertainty can be ignored e.g. we can assume lead times are fixed for truckload shipments delivered consistently in less than a day. However, in other practical settings, variability of lead times cannot be ignored such as, with the diversion of all Suez Canal sea freight around the African Cape due to the Houthi attacks of shipping in the Red Sea. The question is what constitutes operational uncertainty and when do we need to pay attention to its different facets?

In my intro to logistics class I define operational uncertainty as the “risk of not knowing.” In supply chain management specifically, it is the risk of not knowing the exact outcome of normal operations. Consider a supplier product service agreement (PSA) that specifies the supply of a quantity of widgets within a certain lead time at a specified fill rate and acceptable quality level (AQL). It may be the case that from a statistical analysis of the history of shipment receipts you will receive an amount that averages close to the PSA quantity, fill rate, lead time and AQL. However, without valid data and some more involved analysis you may not know with any certainty how far from this average each shipment’s lot quantity, lead time and AQL will be. In other words, what is the variance of these metrics. Lack of this information can affect short-term production schedule and resource utilization decisions as well as long-term profitability outcomes.

It is important to distinguish operational uncertainty from unexpected disruptions or black swan events I wrote about in my May 2024 essay. Operational uncertainty is the prevailing uncertainty expected in normal operations. Black swan events are unexpected events that can rarely be anticipated ahead of time and are typically low-probability high-impact events. A key difference is that operational uncertainty can be influenced by the members of a supply chain. Typically, this is through the terms of the PSAs and the selection of different suppliers and service providers (carriers and third-party logistics providers or 3PLs) and routing e.g. ports of entry.

Why does the variance matter or what did H. Randolph Bobbit and colleagues mean by “The mean is mean but the variance is meaner” (Bobbit et al. 1978)? Consider the Figure 1 with two normal distributions of mean 50 and standard deviation of five (black) and ten (red) representing the distribution of demand from two customers, A and B, respectively. Both customers have the same average of order quantity shipper. However, it is clear that there is less variation from customer A who is very unlikely to order less than 30 units or over 70 units.

On the other hand customer B is very likely to order less than 30 units even as little as 20 units and over 80 units on the high side. Variation such as this can affect production scheduling as well as transportation economies, requiring less-than (full truck or carload) rates and can be costly for the firm.

The primary function of any supply chain is customer service. In other words, to meet end consumer expectations. This requires delivering the 7Rs of supply chain management: delivering the right product to the right place at the right price to the right customer in the right condition at the right time in the right quantity. For this to happen each supply chain firm must understand and fulfill the demand of its downstream customers, which means dealing with the customer demand uncertainty. Consider the growing demographic of 18-24 year olds in key global markets who use e-commerce with an emphasis on value and focus on sustainability and health (McKinsey 2024). In an attempt to build brand loyalty firms will want to both drive and ensure they manage demand uncertainty to meet demand for their products with this key demographic.

Demand Uncertainty

A primary operational uncertainty is demand uncertainty. Assuming a B2B customer, demand uncertainty can be minimized or become baked into the clauses of the PSA. The ideal case of course is where a PSA specifies that customers place a fixed order quantity at fixed intervals for a limited number of items, something I have only seen once for inter-plant material transfers of a single company. Typically, there is uncertainty in when, how much and what customers order.

A first step in managing demand uncertainty is forecasting demand. Depending upon the nature of demand: fast or slow moving, small or large batch sizes, seasonal, trending up, down or steady different forecasting approaches are available. These forecasting approaches can differ in sophistication. An important trade-off in managing demand uncertainty is between forecast sophistication and the cost of forecast inaccuracy. Greater forecast sophistication will drive up costs of technology to access higher quality data and improved forecast accuracy. Forecast inaccuracy can drive up costs with either over forecasting that drives up inventory and obsolescence costs or under forecasting driving up penalty costs for non-compliance of PSA terms, loss of market share and the resulting lost revenue.

Besides forecasting, firms can employ other strategies such as vendor managed inventory (VMI) where firms manage their own inventories in their customers’ facilities. Thus, VMI gives firms insight into their customers’ usage patterns. Collaborative planning forecasting and replenishment (CPFR) is another program that allows firms to align forecast information across suppliers, original equipment manufacturers and customers in order to improve demand forecasts all along the supply chain. Naturally, both VMI and especially CPFR come at a considerable technological cost for higher forecast accuracy. Such strategies would be ideal to synchronize all partners in the supply chains where brands focusing on sustainability and health are building loyalty in the key 18-24 year-old demographic.

Operational Uncertainty

A primary reason to be able to reduce or eliminate demand uncertainty is to be able to synchronize the internal upstream operations of the firm. Lack of synchronization can be costly. For example, sales can over-promise exceeding the capabilities of logistics, distribution, procurement and/or manufacturing or exceed the finance budget. Or inventories at each echelon, raw materials, work-in-process finished goods disposed around the distribution network are not aligned to firms’ customers’ goals.

The best approach to synchronize internal operations is through a sales and operations planning (S&OP) process. An S&OP process ensures all functions including sales, marketing, distribution, production, procurement, logistics and finance have an understanding of the firm’s master plan to serve its customers. The outcome of an S&OP is that each function has the capability to do meet the firm’s master plan and the budget is aligned to ensure that the plan is effectively executed. In doing so there are several key areas where operational uncertainty needs to be managed.

Production Uncertainty: The uncertainty that production managers have to deal with is ensuring production yield meets customer demand. In doing so the managers must fulfill customer demand while balancing plant capacity, labor, plant, materials availability quality, maintenance and unscheduled breakdowns.

Procurement Uncertainty: The goal of procurement is ensuring high quality materials is available for production. Dealing with procurement uncertainty begins articulating clauses to minimize lead times and quality defects in the PSA. Thereafter, it continues with monitoring supplier performance focusing on scorecard metrics that have the greatest effect on downstream operations and costs. Today’s geo-political tensions and conflicts such as in the Red Sea and Ukraine have made contributed to increased procurement uncertainty.

Logistics & Distribution Uncertainty: Transportation uncertainty mainly manifests itself as variability in transit times of transportation and damages/shrinkage in transportation and warehousing operations. Typically, these variations affect materials or product availability and are best buffered against by inventory. Contracting and monitoring contracts and holding 3PLs and other service providers responsible for anomalous service can mitigate these uncertainties. However, disruptions such as the Baltimore bridge collapse, geo-political factors such as the Houthi Red Sea attacks and climate-change such as the Panama Canal drought can add to the uncertainty of logistics and distribution.

Summary

While operational uncertainty can originate in procurement, logistics, production, distribution and demand uncertainty the resulting variation usually has three key outcomes. Firstly, variation results in a lack of the correct materials for production and finished goods to fulfill customer demand. Secondly, variation results in resource misalignment or operational synchronization, either too much or too little of resources. Thirdly, it results in poor customer outcomes such as poor service and product quality. Often inventory is used as a means of buffering against many of these sources of operational uncertainty. However, if the correct data and metrics are not available in ERP systems variance reduction strategies can be ineffectual and expensive.