8. Bringing It Back to Demand/Supply Integration: Managing the Demand Review

Chapter 1 focused on the “super-process” of demand/supply integration, or DSI. Chapters 2 – 7 examined the subprocess of demand forecasting, which along with a variety of other subprocesses such as supply planning, inventory planning, and financial planning, make up the super-process of DSI. This concluding chapter focuses once again on demand/supply integration, but, the discussion centers on the demand review. One way to think about the demand forecasting process is that it is a month-long exercise in preparation for the demand review. All the steps discussed in the book so far, from statistical forecasting, to qualitative forecasting, to performance measurement—all are either foundational capabilities that must be in place, or specific pieces of the subprocess that lead up to the demand review. This chapter presents the typical process flow that leads up to that demand review, as well as the most effective way to conduct the actual demand review meeting, paying considerable attention to that step I call “gap analysis.” This critical step transforms DSI from a tactical exercise in supply chain planning to a strategic element of the overall business planning process in the firm.

Figure 8-1 shows a graphical representation of the Demand Forecasting Process flow, which consists of three distinct phases. Phase I is perhaps the most laborious and time consuming. The outcome of this phase is the initial forecast, which follows from the consolidation of various subprocesses that have been discussed in previous chapters. In Phase II, the demand forecaster identifies the gaps between the initial forecast and the overall goals of the firm, and creates a series of gap-closing strategies in preparation for the demand review. Phase III is the actual demand review meeting. The following sections describe the work that needs to be done to effectively complete each of these phases in the demand forecasting process flow.

Figure 8-1. Demand Forecasting Process flow

Phase I: Preparation of Initial Forecast

The entire process usually begins with the step labeled “Baseline Statistical Forecast” in Figure 8-1. I say that it usually begins with this step because as Chapter 3 detailed, some relatively rare situations exist where an analysis of historical demand patterns is not particularly useful. Aside from these situations, the baseline statistical forecast requires access to demand history. I have spoken in various places throughout this book of the importance of using demand history as the source of the demand forecast. As discussed in Chapter 7, world-class companies construct their demand history using three separate pieces of information: shipment history, adjustments for backorders, and adjustments for unrecognized demand, or lost sales. Frequently, creation of this demand history data file requires a monumental effort that involves not only system enhancements, but behavioral changes by people who work with customers. Both sales and customer service must be trained, and incentivized, to document those instances where customers were ready but unable to buy, because the product or service was not available at the time or place required by the customer. Individuals in these departments require access to this demand history file so that these lost orders can be documented. All this data—shipment history, backorder adjustments, and lost order records—should be stored and professionally maintained in the firm’s data warehouse. Refer to Figure 2-2 or Figure 7-1 for a refresher on the appropriate system infrastructure that supports demand forecasting.

After accessing the demand history, analysts then can apply the procedures described in Chapter 3 to “look in the rear-view mirror” for patterns that might exist in that historical demand, and then project those patterns into the future. However, before applying whatever statistical models might have been used in the past, reviewing those statistical models to ensure that they are still of value is important. Chapter 6 discussed how performance measurement techniques can be used as diagnostic tools to evaluate the usefulness of various statistical models. Examples were offered of percent error charts that revealed flaws in the forecasting techniques that had been applied; also before finalizing the statistical forecast, the analyst should look at previous periods’ performance metrics to identify models that should be adjusted or rethought.

After the baseline statistical forecast is created, the analyst must begin to consolidate the various sources of data that are used to answer the question, “Will the future look like the past?” Various inputs are used to answer this question. These inputs, which the forecaster must consolidate, include a top-down forecast, which is created with macro-level market intelligence as described in Chapter 5. As noted in that chapter, and in Chapter 7 during the discussion of the “Approach” dimension of forecasting management, the most effective process is one that encompasses both a top-down and a bottom-up perspective. Recall that a top-down perspective is one where an estimate of industry demand is combined with an estimate of market share to arrive at a forecast of demand. Macro-level information is needed to create that top-down perspective. Recall from Table 5-1 (Micro versus Macro Market Intelligence) how market intelligence can help to inform that critical top-down demand forecast. As discussed in Chapter 5, forecasters often struggle to include this macro-level market intelligence in their demand forecasting process. However, including this step in a process flow such as that depicted in Figure 8-1 helps to remind forecasters of the importance of looking at macro-level information on a regular basis, and using that information to continuously analyze and document those critical assumptions that underlie the forecast.

Another input that must be included in the “Input Data Consolidation” step is a bottom-up forecast. Chapter 4 covered the importance of qualitative judgment, which is usually gathered from sales, marketing, and product management in a manufacturing context, and merchandising in a retail context. This qualitative judgment constitutes the key element of the micro-level market intelligence discussed in Chapter 5. Insights about customers that come from sales, as well as information about promotional activity that comes from marketing, product management, or merchandising, is critical in creating this bottom-up forecast. The final piece of input that analysts must consolidate during this phase is customer-generated forecasts, which Chapter 5 also discussed. That chapter discussed the best way to choose which customers should be providing forecasts, and some of the risks and opportunities that are involved in using customer-generated forecasts. In many situations, though, these direct customer insights are extremely useful. Thus, the “Input Data Consolidation” step in Figure 8-1 consists of data from the baseline statistical forecast, the top-down forecast generated through macro-level market intelligence, the bottom-up forecast generated through micro-level market intelligence, and customer-generated forecasts.

At this point in the process is where competent demand forecasters look completely different from excellent demand forecasters. Competent demand forecasters are capable of pulling together this information and compiling it into a database or spreadsheet. Excellent demand forecasters are able to take the forecasts created from the different perspectives—statistical, top-down, bottom-up, customer-generated—and interpret the biases, understand the various agendas, evaluate the different levels of quality, apply their own intuition and insight, and create an initial demand forecast that will be ready for the next step in the process—gap analysis.

Phase II: Gap Analysis

The entire purpose behind all the work that is completed in Phase I of the demand forecasting process depicted in Figure 8-1 is to create the best, most accurate, most credible forecast of demand in future time periods. It is intended to be a dispassionate assessment of the level of demand in the marketplace for the firm’s goods and services. Arriving at this place takes a lot of effort by a lot of people—but it’s not the end of the job, because, as discussed in Chapter 1, the case might be that the best, most accurate, and most credible forecast of demand results in the conclusion that the firm will not achieve its objectives. If that is the result of the demand forecasting process, then identifying the gaps becomes the responsibility of the demand forecasters, as well as preparing gap-closing strategies that can be discussed at the demand review. This section discusses the concept of gaps and tries to bring clarity to the cause of those gaps, as well as identifies some of the possible gap-closing options.

Chapter 1 discussed the difference between forecasts and goals. Recall that a forecast is the best guess about what will actually happen, given a set of assumptions. A goal is the outcome that the firm hopes will happen. Goals can be expressed in different ways. A firm can have market share goals, margin goals, inventory goals, cash-flow goals, revenue goals, or any other of a variety of goals. Many organizations have an overarching set of goals, typically financial in nature, which is stated in annual or quarterly “buckets.” Usually referred to as the Annual Operating Plan, or AOP (even though it is really a goal and not a plan), this “master-goal” often forms the foundation for all goals established by the firm. It is commonly the case that the forecast—what we think will actually happen—falls short of the goal—what we hoped would happen, and what we planned our business to be able to have happen.

So what are the consequences of failing to generate enough demand to achieve the goals expressed in the AOP? The two primary consequences are financial and operational. The financial consequence is that for publicly traded firms, investors tend to value the firm based upon their expectations of the firm’s performance. Regardless of the firm’s strategy for communicating expectations to investors, typically the case is that when actual performance fails to reach stated goals, investors won’t be happy, and the stock price might suffer. The operational consequences are that from a planning perspective, the firm will typically acquire enough supply capacity to allow it to deliver the goods or services that must be sold in order to achieve the AOP. Thus, if insufficient demand exists in the marketplace to actually generate the revenue stated in the AOP, then unused capacity might result. Raw material and work-in-process (WIP) inventory might stack up, workers might need to be laid off or furloughed, and investment in fixed cost capacity expansion might be wasted. In other words, across a variety of dimensions, failure to generate enough demand to achieve the AOP goals is not a good thing.

Unfortunately, what often happens in this situation is what I described in Chapter 1 as the most insidious aberration to an effective DSI process—plan-driven forecasting. When the forecast fails to be as high as the AOP, the forecast is simply raised up to the point where it is consistent with the AOP, and the firm deludes itself into thinking that everything is okay. This is insidious because it removes all credibility from the forecasting process. “Customers” of the fore-cast—those procurement planners, production planners, inventory planners, transportation planners, financial planners, and so on—begin to ignore the forecast because they don’t believe it is based on actual demand in the marketplace. This is why the gap analysis phase of the demand forecasting process flow is so important. Without the disciplined analytical activities that accompany this gap analysis, the firm not only runs the risk of failing to achieve its objectives, accompanied by the resulting consequences, but it also runs the risk of removing credibility from the forecasting process as a whole.

Three separate steps are involved in an effective gap analysis. The first step is to examine the assumptions underlying the AOP. Although the possibility exists that the AOP was determined by something as simplistic as “Our plan for the upcoming year is to increase everything by 10%,” one would hope that a more comprehensive analysis was done, with assumptions underlying those analyses. Typically, assumptions that underlie the AOP include the following:

• General business climate. Macroeconomic assumptions can include statements about economic growth, unemployment rates, interest rates, or whatever general business indicators are relevant for the business being planned.

• Market share. When the firm is making overall business plans, it needs to make assumptions about its market share in different markets. General business climate assumptions will inform overall industry sales predictions, but market share assumptions are needed when the firm is planning its expected long-term demand.

• Industry growth. Beyond the general business climate, the industry in which the firm competes might grow or contract at a different rate, or in a different direction, than the general economy. Assumptions must be made about overall industry growth when planning the business.

• Competitive activity. Assumptions about competitive activity will underlie the market share assumptions noted previously. Market share is likely to remain stable if neither the firm nor its competitors do anything different than they’ve done before. However, in most cases, neither the firm nor its competitors will remain static.

Each firm, and each industry, has its own set of assumptions that underlie their AOPs. The more completely these assumptions are documented, the easier is the job of demand forecasters when doing their gap analysis. If the forecast does not reflect the level of demand found in the AOP, there are really only two possible reasons: either the firm’s performance has not reached expectations, or the industry-level assumptions underlying the AOP have not in fact occurred as they were planned. Determining which of these root causes is the real reason behind the gap is critical. If the industry-level assumptions have not occurred as planned, there might be little that can be done, at least in the short run. However, if the gaps are caused by firm performance issues, then gap-closing actions are probably available for consideration.

The second step involved in gap analysis is to document the magnitude, and the level, of the gaps. Several different categories of gaps can exist between the forecast and the AOP, and an understanding of them can help to guide the demand forecasters in their recommendations for gap-closing strategies. These different categories are

• Timing gap. In some cases, demand is likely to materialize, but the timing of the demand is not consistent with the expectations that underlie the AOP. For example, AOP assumptions might include incremental demand associated with new product launches. Having new product launches be delayed, for any number of reasons, is not uncommon. In this case, the demand assumptions might still be valid, but because of the launch delay, the forecast will not match up with the AOP. Another timing gap might revolve around project-based businesses. Again, the assumption behind the magnitude of demand might be valid, but customers who have awarded large projects to the firm might be experiencing delays in the implementation of these projects, and this might affect the timing of their actual purchases. This might be reflected in the demand forecast. In either of these cases, there might be no need for the demand forecaster to suggest any gap-closing strategies, but rather, to simply update others in the firm about these timing issues.

• Volume gap. In some cases, the overall volume of demand might be reasonably close to the AOP goal, but the mix of SKUs, or even brands, that constitute the overall volume, might be highly uncertain. This uncertainty can have a substantial impact not only on revenues, but also on profits. As discussed in Chapter 2, when a company forecasts at the SKU level, there will inevitably be more error, because lower levels of the forecasting hierarchy usually have considerably more variability of demand.

• Regional gap. It is frequently the case that some regions of a company’s market area will experience demand in line with expectations, while others will not. For example, a company might create a forecast for demand in Germany that is consistent with the AOP, while during the same period, demand in Spain or Greece would be far below expectations due to continuing economic woes in those countries. In those cases, suggesting gap-closing strategies would be very helpful for the demand forecaster, such as to increase demand in regions where economic conditions are more favorable.

• Customer gap. Just as situations might exist where one region is meeting planned targets while another is falling short, there might also be certain customers whose demand levels are meeting expectations while other customers are buying at far less than anticipated levels. Gap-closing strategies might be available to increase demand at some customers, because demand is lagging at others.

The bottom line then, from this discussion, is that understanding the source of the gap between AOP and forecasted demand is critical. Without such understanding, any gap-closing strategies are likely to be misdirected and ineffective.

The final step in the gap analysis process is in preparation for the demand review, to prepare a set of alternative gap-closing strategies to present for consideration in that meeting. One important element is that this stage of the process is demand focused. In other words, the gaps that have been identified up to this point are gaps between what customers would buy from us if they could—remember, that’s our definition of demand—and what our firm had planned for in its Annual Operating Plan. No discussion should occur during the demand review of gap closing strategies related to supply. The focus at this point is on the question of how can the firm influence customer demand to bring demand shortfalls into alignment with the firm’s overall goals? To answer this question, I return to a discussion from Chapter 1. When demand is falling short of expectations, a variety of “levers” can be pulled. Some of these levers are very short term–oriented, such as

• Promotional activity. In many companies, additional demand can quickly be acquired through sales promotion efforts. In consumer packaged goods (CPG) companies, either trade or consumer-based promotions can have a dramatic, although short-term, effect on demand. In business-to-business firms, promotional activity might take the form of salesperson incentives to increase demand in certain product categories to certain customers or channels. Demand-side executives should always keep in mind the fact that the demand spikes that often accompany these promotions can be highly disruptive to the supply chain, creating peaks and troughs of demand that can be quite costly.

• Pricing actions. Because most demand curves are downward sloping (at least I think I remember that from my economics courses long ago), firms can usually expect that a price reduction increases demand, and a price increase decreases demand. The amount of the expected demand change is, of course, determined by the buyer’s price elasticity of demand. The firm also must take into consideration any strategic implications of pricing actions, especially on brand reputation. For example, I wouldn’t expect that executives at a company such as Rolex would approve of a price reduction as a way to close any gaps between forecasted demand and the AOP. Such an action could negatively affect the brand’s reputation, and potentially lower the consumer’s reference price for that product.

• New Product Introduction (NPI) timing. In some cases, NPI timing can either be delayed, or accelerated, to create a gap-closing strategy. For example, if an “old” product that is being replaced by a new product is seeing volume declines that are more rapid than originally expected, and if the new product is ready to introduce earlier than planned, closing a gap is possible by changing the timing of the new product introduction.

Of course, these short term–oriented gap-closing strategies will sometimes have a “rob Peter to pay Paul” effect. If a new product is introduced earlier than anticipated, then the demand might be shifted to an earlier period, but overall demand levels might not change. If immediate demand is increased through a price promotion, it might mean that either business customers or consumers will “load up” and not buy in future time periods. More long-term strategies for increasing demand up to the levels targeted in the AOP might include expanding to new markets, introducing new brands that might appeal to underserved markets, or using new overall marketing mix strategies designed to revive mature or declining markets.

Thus, the task of the demand forecaster, in preparation for the demand review, is to prepare the initial demand forecast (Phase I of Figure 8-1), identify gaps between forecasted demand levels and the targets articulated in the Annual Operating Plan, and identify possible gap-closing strategies that can be reviewed during the demand review itself (Phase II).

Phase III: Demand Review Meeting

As stated at the beginning of this chapter, you can think of the entire demand forecasting process as preparation for the demand review meeting. When viewed in the context of the entire DSI super-process, the demand review is typically the first major step. It is here that the demand side of the enterprise (sales and marketing in a manufacturing context and merchandising in a retailing context) passes along to the supply side of the enterprise its best guess about the level of demand that their efforts will generate over the upcoming planning period. At its best, the demand review is seen as a hard commitment on the part of the commercial team to deliver that stated level of demand to the company. This demand forecast then drives the supply team to finalize its plans for all the supply chain components (transportation, production, procurement, and so on) that are needed to support the level of demand to which the commercial team has committed. It also drives the financial team to acquire the capital needed to support this level of demand, and to report to the company’s owners, whether those are shareholders or outright owners, on the financial outcome they can expect. This is why it warrants all the work described throughout the bulk of this book!

The agenda for the demand review meeting should include a number of items. One should be a review of previous month’s performance and assumptions. This agenda item should serve only as the first step, and not as the focus of the meeting. One of the typical problems that I’ve observed in DSI processes is that the meetings are too focused on “How did we do last month?” rather than on “What decisions should we make now in anticipation of future demand?” Still, a review of past performance, along with a review of the status of documented assumptions from previous months, is an appropriate starting point for the demand review meeting.

Another agenda item is to review the initial forecast, by exception. Recall our discussion earlier in this chapter, where the point was made that this initial forecast will consist of an amalgamation of several inputs: the statistical forecast; the bottom-up forecast, which is created using micro-level market intelligence; the top-down forecast, which is created using macro-level market intelligence; and possibly customer-generated forecasts. In most companies, thousands or even tens of thousands of initial forecasts are created each month, depending on the appropriate forecasting level employed. Obviously, a demand planner cannot review each initial forecast, nor can these reports be discussed at the demand review. Two strategies are employed to manage this complexity. The first strategy is to establish exception rules, which drive the decision of whether, and how, to discuss the forecast at the demand review. Table 8-1 shows an example of such a set of exception rules. This table assumes that an “ABC” classification scheme is in place, where the most important products are classified as “A” products, mid-level products are classified as “B” level products, and low-importance products are classified as “C” level products.

Table 8-1. Example of Exception Rule for Demand Review

The MAPE figures that are included in Table 8-1 are used as illustrations—the actual threshold that a company should use is highly idiosyncratic to each company. The point is, though, to establish some sort of decision rule to drive the decision about which forecasts to discuss.

The second strategy used to manage this product complexity is to make forecasting decisions, not at the SKU level, but at a higher level of aggregation in the forecasting hierarchy. Recall from Chapter 3 the discussion of how SKU-level forecasting can often be problematic because of the excessive variability that is experienced at the SKU level. I made the point there that many companies forecast at the product family level, because discernible patterns often exist at that higher level of aggregation that don’t exist at lower levels. When companies take this approach, they often manage the complexity of thousands of SKUs by creating exception rules at the product family level, not at the SKU level. Then, all A level product families are discussed at the demand review, and B and C level product families are discussed only if they fail to meet a pre-determined accuracy threshold at the product family level.

You can find an example of some of these strategies in an article by Mentzer and Schroeter.¹ The company they worked with was Brake Parts, Inc., a manufacturer of aftermarket automobile brake systems and parts. The daunting task faced by this company’s forecasting team was a monthly workload of more than 600,000 SKUby-location forecasts. Obviously, something needed to be done to manage this, because no team of forecasters would be able to analyze 600,000 forecasts, and no demand review would consist of discussing and reaching consensus on 600,000 forecasts! Their solution was to utilize technology and rely on their statistical forecasting system to grind through all those forecasts, and use exception rules to identify for the demand forecasters those specific products that required human intervention and thought. They also managed their demand review by product family, and again used exception rules to drive their decision about which products to discuss at the demand review meeting. Their goal was to create a system that would effectively forecast demand for these products, and only require human beings to examine or discuss a maximum of 1,000 of the 600,000 products they forecasted each month.

¹ Mentzer, John T. and Jon Schroeter (1993), “Multiple Forecasting System at Brake Parts, Inc.,” Journal of Business Forecasting, (Fall), 5–9.

The next agenda item in the demand review is to discuss significant results from the portfolio and product review, focusing on high-impact new product introductions scheduled for the near-term, and decisions made concerning any significant SKU reductions and their effect on demand for other products. Following these discussions, the demand forecasting team should be prepared to present the results of their gap analysis. Articulation of the anticipated gaps, an analysis of the type of gap involved, a presentation of possible demand-side gap-closing strategies, and discussion among decision-capable participants should all occur. To return to a critical point that first introduced in Chapter 1: key, decision-capable representatives from the demand side of the enterprise should attend the demand review meeting, including product or brand marketing, sales, customer service, and key account management. This discussion of gap-closing strategies is the reason that “decision-capable” individuals must be present at the demand review. Even before demand-supply balancing takes place at the supply review stage, decisions need to be made at the demand review about which demand “levers” should be pulled to bring demand and supply into alignment. As noted in the best practices discussion in Chapter 7, Stage 2 companies might have a formalized DSI process in place, but they often fail to have decision makers in attendance at the key meetings. Without these decision-capable individuals in the room, the demand review often reverts to a discussion about “why we didn’t make our numbers last month.” Stage 3 and Stage 4 companies—those who are best in class—have the key players in attendance at all the critical meetings, where decisions can be made and communicated to all other relevant parties.

After discussing these gap closing strategies, and making the decisions, having those in attendance make a statement of consensus is critically important. As discussed in various places in this book, a spirit of consensus defines the optimal culture for DSI, and such a statement of consensus at the demand review meeting ensures that all participants have “bought in” to the decisions that have been reached. I’ve attended formal demand review meetings where the accepted “protocol” is that at the end of the meeting, the meeting chair literally points to each person in the room, and asks for a verbal statement of support that the numbers that have been discussed are the numbers the group will commit to, and that the gap-closing decisions that have been reached have the support of the group. Sometimes, meeting attendees will not be comfortable voicing that support, and further discussion results. But by the end of the meeting, all important players have gone on record in support of the group’s decisions.

The output, then, of the demand review meeting is the consensus forecast of demand, and the agreed-upon gap-closing strategies. But that’s not enough. Another, equally important output of the demand review is a clear statement of the assumptions that underlie the forecast, and any risks and opportunities associated with the forecast. Chapter 5 covered these assumptions in some detail, and the collection and interpretation of market intelligence forms the basis of these internal, and external, assumptions.

Conclusions

This brings us to the end. The consensus demand forecast that comes out of that demand review now goes off to inform the rest of the demand/supply integration (DSI) process. It goes to the supply review meeting, where the supply side of the business will match it up against their capacity forecast, balance total forecasted demand with total forecasted supply, and identify issues that need to be resolved at higher levels in the firm. It then goes to the reconciliation meeting, where the financial community of the firm gets actively involved, dollarizes all the decisions made at earlier meetings, and resolves any issues that can be resolved. Finally, it goes to the executive DSI meeting, where the firm’s leadership team makes sure that the plans that have been agreed upon to capture the identified demand are in alignment with the goals and strategic direction of the enterprise.

And then, you do it all over again.

As I conclude, allow me to make some summary comments, all of which have been made elsewhere in this book, but which deserve one more mention at its conclusion. These represent random neuron firings, and are in no particular order of priority.

• Because a forecast is a guess about the future, it will always be wrong. The challenge is to make it the least wrong that it can be.

• No one buys stock in a company because that company is good at forecasting. Forecasting is only important, or interesting, or worth the effort, if it leads to good business decisions that serve customers, enhance revenue, and reduce costs.

• Statistical forecasting is a necessary, but insufficient, step in a good demand forecasting process. Remember, if you only look in the rear-view mirror, you are likely to get hit by a truck.

• Sales and marketing or merchandising must participate. Period.

• Senior executives must buy into DSI as a way to run the company, and put their money where their mouths are. Without executive support, both financial and emotional, DSI will fail. Period.

• An organization’s culture is far more important to the success of DSI, and good demand forecasting, than any process flowchart or any piece of technology.

And with that, I now conclude. I hope that all of your forecasts are accurate, and that all of your businesses are successful.