PDF File: 

ASC6_wp_nechleba.pdf

4R Systems, Inc.

While technology is helping eliminate artificial uncertainty it is exposing natural uncertainty. For retailers, many decisions must be made within the context of uncertainties that cannot be eliminated or abated.

Over the last 20 years, we’ve seen significant changes in retailing and manufacturing. There are many factors driving these changes; increased margin pressures driven by a competitive marketplace; crowded marketplaces where sustainable differentiation is increasingly difficult; technological innovation driving shorter product lifecycles; consumers who are more sensitive to price in the face of abundant supply.

Concurrent with these increasing business pressures, technology has created new capabilities to capture and analyze more information at a level of detail that was previously unavailable. Where once businesses made decisions on monthly data by region, we now use daily store-level information and, with the adoption of radio frequency identification (RFID) technology, potentially up-to-the-minute information. And with the increased power of technology, we’ve been able to create data warehouses that unify information resident throughout the supply chain.

With this bounty of information, one would expect that we should be more certain about our business and how it runs than ever before. In some ways, we are. These unified data warehouses are delivering increased visibility to the entire supply chain, exposing inefficiencies and inconsistencies that generate significant costs. In this regard, technology has reduced uncertainty by increasing the completeness of information that drives key decisions and exposing many false assumptions that we’ve held or used as a foundation of our planning – a type of bad forecast. In unearthing these bad forecasts, modern supply chain data and technology have helped us remove artificial uncertainty and its associated negative economic consequences.

However, not all uncertainty can be removed. As technology has enabled the capture of data with increasing granularity (when, what and where), we’re finding a more chaotic, uncertain world rather than an orderly one. What appears to be relatively predictable sales on a monthly, departmental, and regional basis is actually much more uncertain at an item, weekly, or store-level basis. Just as physics went from an orderly deterministic world in the Newtonian model to one based on chance with the quantum model, business is discovering that more granular information reveals uncertainty rather than removes it.

As a result, while technology is helping eliminate artificial uncertainty, it is also exposing natural uncertainty. For retailers, many decisions must be made within the context of uncertainties that cannot be eliminated or abated.

Uncertainty Within the Supply Chain

Why should retailers care about uncertainty? The short answer is that uncertainty adds cost. Let’s take a retailer who sells a perishable food product that is only saleable for one day. If the retailer knew exactly how much they were going to sell each day, they could provide the exact quantity to satisfy demand and the cost to service customer demand would be minimized. However, if demand were to vary randomly each day, the days where demand fell short of supply would incur spoilage costs while the days where demand outstripped supply would incur the opportunity cost of a lost sale.

So, if uncertainty adds cost, one strategy for reducing cost is to reduce uncertainty. Indeed, a key benefit of a modern, information-rich, technology-enabled supply chain is the identification and elimination of artificial uncertainty and, as a result, a significant reduction in cost. Much of what has already been realized in savings in the supply chain has been based upon this premise.

However, retailers have a variety of sources of natural uncertainty with which they must contend as well. Some uncertainty is associated with events where changes in the status quo create uncertainty. Whether it is the introduction of a product, promoting a product, changing price, or markdown for clearance, these changes have both an expected outcome and a degree of uncertainty in that outcome. Beyond that, the uncertainty is not limited to the particular product whose price has changed. In fact, that price change can impact the sales of other products and makes their sales less certain as well.

Beyond these events, retailers must deal with systemic uncertainty as well. For example, let’s compare two retailers. One sells 500 units per week through an Internet site while the other sells the same quantity through 50 stores. Given typical randomness of sales found in retail stores, the store-based retailer would need to position 799 units in order to provide a 90 percent service level (see Figure 1). Comparatively, the Internet retailer might only need 540 units to provide the same service level. As each item has an expected level of sales and uncertainty in its sales level at a given point of sale, these characteristics describe one facet of a retailer’s systemic uncertainty.

Uncertainty Throughout the Product Lifecycle

Retail products have three distinct phases of life – product introduction, replenishment, and end of life. The uncertainty a retailer faces throughout the lifecycle is also distinctly different.

At product introduction, the key driver of uncertainty is the degree to which consumers are interested in the product and its value proposition. Pricing, product characteristics, and competition are just some of the factors that affect the demand for a product, and while historical product launch data can be a guide, actual demand often varies significantly from prelaunch forecasts. This uncertainty often creates lost sales, as there is too little available supply to service demand, or costs associated with obsolete products, as too much supply is bought relative to lifetime demand.

Once a product has been introduced and its general selling pattern has been established, generally products go on replenishment. As shown in Figure 1, the random fluctuations of demand at a given point of sale introduce a degree of demand uncertainty. Moreover, marketing efforts such as temporary price reductions, promotions, and special displays not only change underlying demand but change the uncertainty in that demand as well.

Finally, at end of life, there are a number of uncertain events. When will a product be discontinued? Depending upon whether it happens earlier or later, less or more inventory in the supply chain is needed. Once a product is discontinued, how long will it take to sell off the remaining inventory? How can markdowns change that time and generate the most total dollars that can be generated?

Strategies for Dealing With Uncertainty

Eliminate artificial uncertainty – Probably the easiest savings in the supply chain involve eliminating artificial uncertainty. Researchers have richly illustrated how supply chain myopia has created artificial uncertainty (bad assumptions in many cases) that is at the root of significant costs. For example, Campbell’s Soup for many years ran large promotions during peak season assuming that these events were profitable. As it moved toward continuous replenishment, it identified the true costs of servicing the peak season promotions and realized that the promotions created more costs, for itself and retailers, than the margin they generated. In many regards, the cost assumptions that drove the planning process were a poor forecast that led to a less than desirable outcome.

Reduce natural uncertainty – Many of the mechanisms retailers employ are uncertainty reducers. For example, holding inventory at a distribution center that can quickly replenish stores is an uncertainty reduction strategy. By scaling back-up inventory in relation to the uncertainty of sales for a large aggregation of stores, the amount of inventory and the cost of inventory can be substantially less than the amount needed if inventory was held in stores.

Economically optimize decisions to chart a profitable course in an uncertain world – Given a forecast of demand and uncertainty in the outcome, economic optimization selects an inventory policy that weights the likelihood of different demand levels by the economic costs incurred in servicing those levels of demand. Thus, products with higher gross margin should be inventoried at higher levels as the margin from capturing incremental sales would justify the cost of positioning more inventory. Alternatively, products with lower margin would tend to be stocked at relatively lower levels, as the potential gains from those products would not justify the increased costs associated with stocking these items.

For example, in our experience working with a large, multibilliondollar retailer, we economically optimized store model stocks for 5 million store/sku combinations on a weekly basis, yielding gross margin increases that amounted to 1.3 percent of sales and an economic benefit in the tens of millions of dollars.

Adopt economic metrics to measure inventory efficiency – As retailers adopt economic optimization to drive more profitable inventory management practices, we’ve seen a change in the metrics they use to track and manage inventory. Traditionally, retailers have relied on metrics that were easily collected and calculated. Metrics such as turn rate and in-stock levels are retail standards. These metrics are certainly correlated with profitability and provide valuable inputs in identifying significant problems. However, they inadequately address a number of issues that are key in generating profitability. We’ve seen retailers adopting new metrics focused on lost sales, margin dollars, and forecast error when they adopt economic optimization technology to manage their inventories. By measuring lost sales and forecast error, retailers are able to better understand the impact of their inventory management decisions on company profitability. Moreover, by actively managing to these metrics on a daily basis, they can positively impact their quarterly and annual results.

Understand what factors optimization considers and doesn’t – There’s been considerable interest in markdown optimization tools and they can be quite effective at recovering more revenue from excess product inventory. However, markdown optimization comes with two significant liabilities.

First, it is fundamentally a loss mitigation approach. Often, markdowns are focused on minimizing the cost of a prior, suboptimal decision (poor buying or allocation). While an effective tactic, it is a more expensive solution than reducing the amount of product that needs to be liquidated. One could say that markdown management is “locking the barn door after the horse has been stolen.” By focusing on understanding and anticipating product end-of-life, inventories can be intelligently optimized to drive a loss avoidance strategy. Coupling loss avoidance with loss mitigation allows retailers to both lower the amount of markdown activity in addition to extracting the maximum economic value from excess inventory. In this case, synchronizing supply optimization with markdown optimization can create a more effective, profitable solution.

Second, markdown optimization leverages insight into price elasticity to generate the most revenue in liquidating an inventory position. Left out of this optimization decision is any erosion in brand value that comes with significant discounting at a particular point of sale. As a result, markdown optimization does not generate a retaileroptimal (global) but merely a product-optimal (local) decision. Again, those that execute both a loss avoidance strategy (optimizing inventories for product exit) with markdown optimization are less likely to erode their brand’s value and, ultimately, their sustainable profitability

Comparatively, optimizing supply to service uncertain demand (replenishment) is something a retailer can globally optimize. By understanding the trade off between gross margin and the cost of inventory, retailers can develop a consistent inventory policy rather than one that oscillates between programs that cut inventory to reign in costs and programs that add inventory to drive better service levels. With inventory optimization, there are no hidden, unconsidered factors.

The Retail Optimization Opportunity Map

Across the product lifecycle, substantial profit opportunities exist in optimizing decisions. Given the amount of uncertainty that exists at each stage of the product lifecycle and the magnitude of costs, we can estimate the potential payoff associated with optimizing different retail functions. Figure 2 depicts both the degree of difficulty and the expected increase in profitability through optimizing product launch, replenishment, and end of life. In addition, it represents the potential impact of a rigorous testing mechanism in reducing uncertainty as a driver of reduced costs.

One can see that all areas have the potential to create substantial incremental profits. It is important to note that the difficulty in realizing these incremental profits varies across these lifecycle steps. In our experience, replenishment is simpler to implement (although there is nothing simple about the underlying technology) in a retail organization when compared to other lifecycle steps. First, it is not an exception process like initial buying or end-of-life but rather a systemic process that occurs often for most items and needs to be decided for each store. Thus, it is best managed en masse, across all replenishment products and is primarily a data-centric exercise.

Initial buying and markdown are exception processes applied to products at specific points in their life. For example, in the initial buying process, a planner typically looks at information but also overlays significant judgment and tacit knowledge. At markdown, the planner also holds significant tacit knowledge that can help in extracting maximum revenue from the remaining supply. Compared to replenishment optimization, these activities change people’s work, require training, and often change business processes. Successful implementation of these solutions requires a good understanding of the organization and its characteristics.

Conclusion

In a very competitive world, pressures on retailers and their suppliers to lower costs continue unabated. No longer can they afford to have low-productivity, excess inventory available to handle unexpected demand. With these pressures, the supply chain is increasingly called on to service consumer demand on a just-in-time basis. Unfortunately, the uncertainty of consumer demand makes servicing on a just-in-time basis difficult. In the drive to substitute information for inventory, retailers need to leverage better mathematics designed to model uncertain demand to help them optimize their decisions and generate more profit.

References

1 Marshall L. Fisher, Ananth Raman, Janice H. Hammond, and Walter R. Obermeyer, “Making Supply Meet Demand in an Uncertain World,” Harvard Business Review, May-June 1994.

2 Marshall L. Fisher, “What is the Right Supply Chain for Your Product?” Harvard Business Review, March-April 1997.