CSCMP's Supply Chain Quarterly
October 22, 2018

Improve your odds

No matter how hard you work to reduce uncertainty in your supply chain, you'll still have to make critical "bets" before key data become available. This three-step approach to decision making will improve your chances of making the right calls.

Highly volatile supply chains are similar to betting games: in both, it is important to understand the odds of various outcomes and the commensurate payoffs. The similarities continue when it comes to decision making. Like gamblers who make plays without knowing what cards their opponents hold, companies often must make critical supply chain "bets" long before key data become available. Mobile phone manufacturers, for example, must place orders for memory chips up to two years before new handset designs reach the market. Food growers must make decisions about which crops to sow even though yields will be determined by weather conditions that occur months later. In recent years, high-tech companies have had to decide how to allocate their investments in the production of liquid crystal display (LCD), lightemitting diode (LED), and plasma technologies without knowing which will succeed in the market.

Most companies struggle with making big decisions in the face of uncertainty. These difficulties constantly erode their profits—sometimes dramatically, as when their predictions turn out to be spectacularly wrong, and sometimes insidiously, as when they consistently over- or underproduce. Perhaps that is why so many seek some sort of "crystal ball" to help them manage supply chain uncertainty.

Article Figures
[Figure 1] Sources of bias in supply chain decision making
[Figure 1] Sources of bias in supply chain decision making Enlarge this image
[Figure 2] Significant impact from end-to-end supply chain improvements
[Figure 2] Significant impact from end-to-end supply chain improvements Enlarge this image

What they need instead is a deep understanding of their actual business risks and a systematic decision-making process that relies on robust data and aligns key decision makers. With that (and our betting analogy) in mind, we recommend the following three-step approach to dealing with uncertainty:

Stack the deck in your favor, reducing the level of uncertainty through such traditional actions as latestage configuration, flexible production, and demand control.

Understand the odds by assessing the probability and impact of remaining risks on both the demand and the supply side.

Place smarter bets by aligning the incentives within your organization and educating your people on how to make the right choices when uncertainty exists.

The impact of this three-stage process can be profound. One company that adopted this approach, for example, achieved a 2 percent increase in revenue and cut its inventories by 20 percent. In this article, we will explain the three steps and how they can help companies similarly improve their chances of making the right supply chain decisions in times of uncertainty.

Why is it hard to cope with uncertainty?
Before examining this three-step approach in more detail, we should consider the reasons why companies often perform poorly in the face of uncertainty. When we look at the techniques companies use today to cope with uncertainty, we find that problems arise not so much because the individuals charged with making decisions lack the appropriate skills or knowledge, but rather because they lack sufficient information, or there is some bias in the culture or management environment that leads to incorrect decisions.

One of the most common types of bias is the "action-oriented" bias. This is the natural tendency of an individual to base a prediction regarding the next event on the outcome of the last one. A supply chain manager who failed to order enough of last year's popular toy, for example, may well order too much for this year's holiday season. Likewise, a beverage company executive who has had to explain the costly disposal of thousands of tons of rotting fruit may be reluctant to accept his sales team's optimistic projections for sales of a new fruit drink.

An organization's incentives frequently produce bias. Sales teams, for instance, are often motivated to increase volumes but are not penalized for the costs of excess production. Keen to keep service levels high in order to minimize the risk of lost sales, these teams may consistently produce unrealistic demand forecasts, leaving their companies with unsold or obsolete products. Sometimes the commercial manager, who is in charge of the forecasts, and the supply chain manager, who is in charge of production, do not communicate about their decision-making process and rationale. It's not uncommon, moreover, for the sales organization to develop a forecast and "throw it over the wall" to the supply chain function, where a skeptical supply chain manager may execute a completely different plan in order to maximize the utilization of production facilities, or pile on an additional forecast buffer to avoid negative consequences later on.

Sometimes the multiple steps required for decision making along the supply chain introduce additional complexities—the classic "bullwhip" effect. If, for example, an agricultural products company's sales force predicts a certain level of demand for the next year, its processing unit may add a certain percentage to the figure it gives to its growers to account for production problems or damage in storage. The growers, in turn, may add their own safety buffer to allow for bad weather depressing yields. In practice, the chance of both bad weather and big production problems happening in the same year is extremely low, so it is very likely that the company will end the year with excess inventory.

Another example of bias that leads to incorrect decisions is the "social" or "champion" bias, where decision makers tend to follow a higher-ranked individual's expectations or advice. There can be enormous pressure on managers to follow a superior's lead. Suppose you are responsible for ordering the touchscreen LCDs for a new consumer electronics product that is about to launch. Your company's CEO will debut the new product in front of a global audience, and sales expectations are high. How will you make the decision about order quantity? Will you believe the hype or be conservative? If you choose to be conservative— and then the product is wildly successful and shortages occur—you could lose your job. But even if the conservative order turns out to be correct, your caution could be blamed for slowing the adoption of the product. (See Figure 1 for some other sources of bias in supply chain decision making.)

Stacking the deck
Smart companies start their battle against uncertainty and irrational decision making by "stacking the deck"—taking advance action to ensure a favorable outcome. (Although this expression usually refers to an unfair advantage, we are using it in a positive way here, of course.) In a supply chain context, that means using a variety of strategies that help to reduce their risk exposure. Take, for example, the risk posed by volatile demand. Strategies that improve a supply chain's speed, flexibility, or reliability can help companies avoid problems caused by this type of uncertainty.

Some apparel companies, for instance, have accelerated their supply chains in recent years, leading to the development of "fast fashion" supply chains. Using a segmented approach to decide between onshore and off-shore manufacturing (generally on the basis of margins and demand volatility), these companies have been able to reduce lead times where it matters. As a result, they can take design ideas from the catwalk to the consumer in a matter of weeks. This fast fashion strategy also allows them to rapidly adjust production up or down in response to demand, maintaining high service levels without the risk of being caught with excessive unsold inventory. (In less volatile segments, the supply chain can be optimized to minimize cost rather than focus on speed.)

Where it isn't possible to accelerate the whole production process, companies can sometimes improve flexibility through postponement. Under a postponement strategy, they may manufacture "vanilla" (generic) systems or components, which they can convert or assemble into any one of a number of final product configurations once real demand is known. High-tech companies can, for example, move more product features from hardware to software; this allows them to switch more features on or off just before final delivery. Postponement offers an effective way to deal with uncertainty because it balances the impact of demand variability across portfolios of products, thus reducing the risk of over- or undersupply of individual items.

Other risk-mitigation strategies include supply-side measures like product design flexibility, which allows manufacturers to use alternative components if first choices are not available. On the demand side, everyday-low-price policies can reduce demand volatility because they eliminate the demand spikes caused by periodic price promotions.

Ultimately, most companies will have to cope with a certain level of variability when there is no way to mitigate it, or when mitigating actions would be prohibitively expensive. They have no choice but to make a decision about production capacities and inventory levels in light of uncertain supply or demand. If they bet too short and manufacture too few products, they will miss out on sales opportunities and risk upsetting customers. If they bet too long and produce too many products, they may end up with excess inventory that they will have to destroy or offload at a loss. But, as we explain in the next section, assessing the probability and impact of the remaining risks improves the chances of making the right decisions.

Understanding the odds
To win in the supply chain game, you need to understand the likely spread of product demand, not just the average; incorporate the economic impact of betting short or long; and then make decisions that optimize for that economic impact.

Developing the right supply and demand probability distributions starts with a good understanding of the underlying drivers of variability. Common sources of demand-side variability include seasonal fluctuations, innovation and new product introductions, competitors' activities, and wider economic conditions, to name just a few. Supply-side variability can be affected by factors such as weather conditions, fluctuations in the price or availability of commodities, and by the actions of suppliers, among others. As these drivers will vary widely by industry and market, initial lists of drivers should be built using input from internal managers, external partners, and consumers, and then be statistically tested for relevance by comparing historical fluctuations in each driver with actual changes in supply and demand.

Understanding the underlying factors that drive demand can materially improve a company's ability to forecast. For a pharmaceutical company that needs to estimate demand for its vaccine products, for example, it can be easier to estimate key drivers—such as the chances that the disease will spread to wide geographies, a competing product will come on the market, and the government will approve the vaccines— than to attempt to estimate the distribution of final demand.

To factor in the economic impact of supply chain decisions, companies need a clear understanding of how going long or short in forecasts will affect their margins. For excess inventory, this model should take into account the carrying costs, the costs of disposing of any surplus inventory, and any costs associated with diverting product to alternative markets (by repackaging or reworking, for example). For short inventory, companies will need a good understanding of consumer behavior in order to incorporate the effects of product substitution when estimating lost sales margin. If customers can easily be steered toward alternative products—perhaps by giving a discount on a premium version—then margin loss can be contained. Once companies have developed their supply and demand distributions and understand the margins that are at risk, they can use statistical modeling techniques like Monte Carlo simulations (which assesses the probability of certain outcomes by running simulations using random variables) to model the likely impact of different scenarios and identify the most profitable inventory and production quantities. However, feeding the model with the right inputs requires tapping into the full breadth of knowledge that exists throughout the organization, and this, in turn, may require quite profound cultural changes.

It is essential to incorporate the best possible knowledge into any analytical model that weighs trade-offs. Yet this knowledge often lies not with those people who do the modeling or make the inventory decisions but in the hands of those who are closest to the front lines. Product managers in different regions, for example, may have unique insights into the willingness of customers in their areas to accept substitute product in the event of a supply shortage. Similarly, growers in different regions may be able to offer better crop-yield predictions than a single model of the country as a whole could provide. Companies can capture this kind of data by allowing regional managers to enter their own estimates into the model.

When looking for the best sources of knowledge, companies should also look beyond their own organizations and leverage that of customers and suppliers. Consider the experience of one consumer packaged goods company that struggled to predict the effect of promotional events on sales through a large retail customer. This company developed a joint monthly sales and operations planning (S&OP) process with its customer, reviewing point-of-sale data, sales lifts from similar events in the past, display strategy, and the customer's business and supply chain constraints. This allowed it to develop a forecast that created longerterm visibility, improving service levels by one percentage point.

Another technique to improve a model's forecast accuracy is the use of "prediction markets" or "wisdom of the crowd." For example, when managers are asked to predict their company's performance, the average of all managers' predictions may be more accurate than the official company forecasts, which may suffer from bias, personal judgments, and the need to set business expectations.

Placing smarter bets
Even the best available model of an unpredictable situation is only a guide for decision making. Managers will need to use their business judgement to make the final call on their big supply chain bets. By becoming more systematic in their approach to this part of the process, however, they can increase the likelihood that they will make the right decisions. Systematic decision making requires the right people to be brought together with the right data, in a regular formal planning process. By doing this, companies can ensure that decisions are made with the best available information and with input from all stakeholders.

To continuously improve the quality of the information they have gathered and of judgment-based decisions, organizations should also create feedback loops that allow them to learn from past experiences. This includes recording the rationale behind choices and later—when the results are in—reviewing those choices with decision makers. Reinforcing the feedback system with formal mechanisms such as incentives linked to forecast accuracy or the ability to execute against a demand plan can further accelerate the learning process.

Because optimal performance in variable and uncertain environments may depend on the decisions made by many individuals across the organization, it's important to help people make the best choices. This may require changing their mindsets and behaviors through education and the use of incentives.

That was the tactic employed by one company that experienced frequent overstocks because its sales teams consistently overforecast demand for high-margin products in order to ensure that supply would always be available. The company organized a series of workshops to educate sales managers about the costs associated with excess inventory and explain why demand could not always be fulfilled. The managers then conveyed this message to the front-line sales organization. Adding incentives that were based on overall margin further reinforced the company's message.

Metrics and incentives that are aligned with the overall business strategy can help every function understand its own impact on supply chain performance. Metrics drive appropriate trade-offs—among cost, service, and inventory levels, for example—and ensure that functions work together to deliver the best result for the business as a whole. The results of some of our recent supply chain research bear this out. Companies that achieved best practice in performance management (which included appropriate use of metrics and incentives) were 1.7 times more likely than poor practitioners to be top service performers, for example. At the same time, they delivered this performance with better inventory levels and were 1.6 times as likely to be a top inventory performer.1

However, managers must recognize that in an uncertain environment, they are not going to get it right every time. Effective decision making requires building confidence in the robustness of the processes that deal with these uncertainties; this is essential for preventing the organization from lapsing back into old habits when the occasional outlier event does occur.

Stronger in the long run
Improved decision making in volatile supply chains delivers benefits that come from a variety of sources. First, it improves sales and service. We have seen companies use the three-step approach to lift their sales by as much as two percentage points in highly competitive, volatile categories. Second, it can have a dramatic effect on supply chain cost and complexity, helping companies to cut 5 to 10 percent from their overall supply chain costs while also reducing complexity and lead times. Third, better forecasting and risk management practices help companies to cut inventories, by as much as 40 percent in some cases, and to smooth their production processes, such as by improving the utilization of production equipment (see Figure 2).

Companies we have worked with that have adopted this approach to dealing with risk have enjoyed other benefits, too. At one biotechnology firm, the sales staff's level of confidence in the supply chain as a competitive advantage more than doubled yearover- year. (One sales representative even demonstrated that recognition by sending the director of supply chain a Christmas card thanking him for helping the group make its sales targets.)

Some companies have been so impressed with the success of the three-step decision-making process that they have extended the same approach to other business decisions where uncertainty is a concern. Based on this logic, for example, one pharmaceutical company decided to make a significant investment in a capacity expansion that it believes will be a competitive advantage in years to come.

Although still in the minority, more companies are beginning to implement this systematic approach to dealing with supply chain risk. After doing what they can to improve the responsiveness of their supply chains, they are using analytical tools to deliver a deep understanding of the end-to-end impact of supply chain decisions. Then they are systematically tapping into sources of field intelligence in order to complement supply chain science with organizational experience. Finally, they are taking steps to change their culture so that individuals are motivated to make the best decisions for the organization. While recognizing that some of their supply chain bets may not pay off, these companies also know that their systematic approach to risk will make their organizations better off in the long run.

1. "The Race for Supply Chain Advantage: Six practices that drive supply chain performance." Brian Ruwadi, Joshua Wine, Bruce Constantine, Martin Losch, Alex Niemeyer. McKinsey & Company report (2008), p. 24.

Jan Henrich is a principal in McKinsey & Company's Chicago office and leader of McKinsey's consumer packaged goods operations practice in North America. Nitin Khanna is an associate principle in McKinsey's Atlanta office and a leader in the firm's supply chain practice. Evren Ozkaya is an associate in McKinsey's Atlanta office and is also a member of the supply chain practice.

Join the Discussion

After you comment, click Post. If you're not already logged in, you will be asked to log in or register.

Want more articles like this? Sign up for a free subscription to Supply Chain Executive Insight, a monthly e-newsletter that provides insights and commentary on supply chain trends and developments. Click here to subscribe.

We Want to Hear From You! We invite you to share your thoughts and opinions about this article by sending an e-mail to ?Subject=Letter to the Editor: Quarter 2011: Improve your odds"> . We will publish selected readers' comments in future issues of CSCMP's Supply Chain Quarterly. Correspondence may be edited for clarity or for length.

Want more articles like this? Subscribe to CSCMP's Supply Chain Quarterly.