Goals and Objectives

• A formalized demand planning process that is tactical with a focus on creating a shipment-based (supply-based) forecast that is not seen as a strategic planning process reflective of true demand by the commercial business teams.
• A short-term unit volume forecast is developed using the shipments (supply history) and/or sales orders (replenishment history). The planning horizon is based on an operational (one- to three-month frozen) planning view.
• S&OP process (bridge) is designed to close and explain the gaps between the current view of demand and the annual operating plan (budget) with the goal of setting sales targets to close the gaps.

People

• Demand planners’ primary role is the management of information and data with the goal of creating a final shipment forecast (supply plan).
• Data cleansing and information management take up 80 percent or more of the demand planners’ time during each planning cycle.
• Demand planners report upstream to the operations planning organization, with ownership and accountability falling primarily on the demand management team.
• Demand planners use system-generated basic time series methods (e.g., moving averaging and non-seasonal exponential smoothing)—a black box approach—to create the statistical baseline forecasts.
• Little if any investment is made in updating demand planners statistical skills.
• There is little use and little knowledge of downstream data.

Process

• Simple statistical methods, primarily moving averaging and nonseasonal exponential smoothing, are used to create a statistical baseline forecast. Promotions and outliers are cleansed from the historical demand history (shipments and/or sales orders), and adjusted and passed to the commercial teams for manual input.
• Little if any ownership or accountability for commercial overrides to the promoted volumes, which are manually layered back to the baseline forecast volume by the demand planners.
• The collaboration planning process is politically charged with minimal accountability for those who make manual adjustments to the statistical baseline forecasts.
• S&OP process is vertically aligned and tactical in nature with a focus on creating a three month frozen supply plan. No perceived benefits for the commercial teams to participate in the process. It is a top-down process with a focus on OP (operations planning) with no integration of downstream data, or accountability for financial plugs designed to close the gaps between the financial plan (budget) and current demand forecast.

Analytics

• Lights-out black box approach using simple statistical methods (e.g., moving averaging and nonseasonal exponential smoothing) to create the baseline demand forecasts. No use of predictive analytics.
• Some descriptive analytics are used for reporting purposes.
• The primary performance metric is MAPE focused on the aggregate level of the business hierarchy. Little if any attention to the lower-level mix MAPEs.
• Not using FVA/lean forecasting to monitor and track value-added and non-value-added activities in the process.

Technology

• Excel spreadsheets are the dominant tool for capturing and analyzing forecast information. They also are the resources for sharing information and data across the organization.
• Manually cleansing demand history into baseline and promoted volumes to accommodate shortcomings of the analytics methods deployed in their ERP demand management technology.
• Many times companies attempted to buy their way into demand planning competency by purchasing demand planning software before their processes, data, and organizations are developed and ready to become demand-driven. In other words, they purchase demand planning solutions and then twist and bend them to fit their existing process. This not only sub-optimizes the technology, but also the process, with poor results. This results in failure and expensive software not being used.
• Demand planning is based on ERP (enterprise resource planning) transactional data. ERP systems capture numerous amounts of detailed data, and the large volume of transactions complicates and slows down the process. Data downloads often take a very long time, and developing meaningful information from hundreds of thousands or more transactions is very tedious, complicated, and prone to errors.
• No real centralized enterprise data repository or warehouse to store information and data. Information and data are fragmented across the organization in data marts supporting local applications. As a result, unable to explore big data to uncover actionable insights using predictive analytics.
• The biggest challenge is converting massive amounts of data into actionable (or usable) information that the organization can absorb, understand, and use effectively to make better-informed decisions.

Goals and Objectives

• There is executive-level sponsorship required to assure adaption and sustainability of the new process design. The initial executive level support for demand planning usually comes from the senior supply chain leadership team who sponsors, supports, and becomes the champion of the new demand planning process. The sponsor has a vested interest in the outcome of each demand plan update, and interacts with its business cross-functional peers (sales, marketing, and finance) to improve decision making process to drive revenue growth and profit. However, that does not mean a C-level manager from sales and/or marketing should not be given the opportunity to champion the new generation demand management process, as it is not just about reducing costs. In fact, given that downstream data, sales tactics, and marketing strategies play a key role in the process, it makes more sense to have a C-level manager from the commercial side of the business champion the future state.
• Preferably, a committee/board made up of C-level managers represented by sales, marketing, finance, and operations should be established as the overseer of the process by which the stakeholder (champion) reports. Any challenges related to change management and other related activities requiring approvals (i.e., hiring, training, technology purchases, and others) will require sign-off by the oversight committee/board. This cross-sectional board over time will better understand the value and benefits creating accountability and ownership of the process.
• The unconstrained consumer demand forecast is generated using downstream POS/syndicated scanner data (where appropriate), sales orders, and shipment history supported by internal consumer/trade insights. The demand planning process relies on historical customer/consumer demand in developing both the short-term tactical and long-term strategic plans. Most companies usually leverage only the historical data from customer sales orders and shipments in order to get a view of historical demand patterns. Emerging collaboration with sales and marketing teams provides valuable consumer insights for future demand forecasts. So, it is imperative that information regarding sales promotions, pricing, and other related factors is identified to sense and shape future demand, including customer sales orders and/or shipments where POS/syndicated scanner data are not available or do not make sense.
• The main objective is to generate a unit volume unconstrained demand plan as an input into the S&OP/IBP process. Since one of the main uses for demand forecasts is supply planning, the forecast should reflect unit volumes. Also, translation from units to revenue is more accurate than the reverse due to pricing changes. Therefore, it is recommended to always create the demand plan in units, then convert it to revenue. The final demand plan is developed at the SKU/location level, and the resulting data are uploaded to upstream operations planning systems.
• Demand planning is a tactical and strategic supply chain process that can no longer be compiled in a vacuum upstream in the operations planning area. The demand forecasts of the next generation demand management process provide valuable input to sales and marketing to support demand generation as well as operational execution. If dollars, profit margins, and other related versions of the future demand forecast are required, it is necessary to invest in new demand-driven technology that can convert the unit volume forecast into those unit measures on the fly through a conversion table to support all planning requirements. However, it is critical that the initial demand forecast be derived in units, then converted to dollars, profit, pallets, and others to maintain continuity and accuracy across the organization. There should always be one version of the truth (most up-to-date unconstrained demand forecast) that all participating departments can view on demand.

People

• Dedicated demand planners supported by demand analysts who are embedded in the commercial organization. Companies are investing in centers of forecasting excellence at the corporate level, and are staffing those centers of excellence with demand analysts, not demand planners. Demand analysts are responsible for creating the statistical baseline forecasts for all the regions/divisions. Once those statistical forecasts are tested and validated, working closely with the regional/divisional demand planners, the forecasts are turned over to the demand planners, who work closely with the local commercial business teams to make adjustments based on local sales and marketing activities, such as sales promotions and other related marketing programs.

  The skill sets of these newly created demand analyst positions are different in that those individuals have advanced statistical skills and strong business acumen. They also have strong collaboration skills as they work closely with the regional demand planners. The regional demand planners are in a true planner role, rather than statistical forecasters working with the local commercial business teams to make adjustments to the statistical forecasts based on local sales/marketing activities (i.e., sales promotions, pricing actions, and others).

• Companies are realizing the value of analytics and the positive impact it has on demand planning. Companies are hiring dedicated demand analysts to provide more structure and analytics to the demand planning organization. The ratio of demand analysts to the number of demand planners depends on the size and complexity of the business. It is very common for global companies to assign a lead demand analyst at the global level to support each demand planning team at the regional level. The new role of demand planners is more of a coordinating role supporting demand planning, rather than manipulation of the statistical baseline forecast.
• The demand planning role is moving from its traditional supply chain origins to the sales and marketing organizations. The demand planner roles were initially established under the supply chain organization, but now companies are slowing realizing that demand planners need to be closer to the customer/consumer in order to improve forecast accuracy. Reporting into the supply chain puts demand planners too far upstream, removed from the customer, to understand those program vehicles that influence demand. Working downstream in the process supporting the commercial business, who are responsible for demand generation, demand planners are better positioned to translate those sales/marketing programs into viable inputs into the statistical unconstrained demand forecast. These new organizational changes allow demand planners to collaborate closely with the commercial team. Companies are finding a strong correlation between a centralized demand analyst structure supporting a decentralize regional demand planning organizational tends to improves forecast accuracy.

Process

• Formalizing the demand planning processes. With dedicated demand planning resources trained and in place, the demand-planning process becomes standardized across the entire supply chain including sales and marketing. The best practices are propagated across the different demand channels. New-product life cycle management planning plays a key role and needs to be included in the next generation demand management process.
• Statistical methods are deployed to create the unconstrained demand forecast. A holistic approach to statistical forecasting requires more advanced methods (e.g., ARIMA, ARIMAX, multiple regression) to develop holistic statistical baseline forecasts that are reflective of the sales and marketing tactics and strategies along with traditional time series methods (exponential smoothing) to address the different product segments (slow moving, new products, fast moving, and steady state). Although the most common platform for demand forecasting is Excel spreadsheet applications, they are not scalable, nor do they have the depth and breadth of predictive analytics to forecast bottom-up, middle-out, or top-down at the SKU/location level of granularity. This requires an enterprise data warehouse with large-scale automatic forecasting technology that can take advantage of information that resides in the company's demand signal repository.
• Tactical and strategic forecasts are required to support short- and long-term supply chain requirements. Future state companies are extending their planning horizons past the comfort of customer order lead times. This is why companies extend their demand forecasts out to 18 months to execute against the strategic plan. However, they also require a 6- to 13-week forecast for tactical executional purposes. It is still common, however, for the fiscal year-end period to form a threshold beyond which the quality of forecast detail begins to diminish.
• Cross-functional collaboration is a key part of the process. Internal cross-functional communication and collaboration take place between the customer-facing teams (sales, marketing), providing domain knowledge into the demand planning process. They review the results of the statistical unconstrained demand forecasts and provide insights into any significant upcoming marketing events, whether it is internal trade promotions or changes to regular customer/consumer demand patterns. At this stage, the customer-facing teams use the statistical model output to run what-if scenarios at the middle-out forecast levels to test and validate tactical and strategic programs to determine which programs drive not only incremental unit volume but also growth and profitability. Collaborating across the commercial organization and working closely with demand analysts, the cross-functional commercial teams as well as finance finalize the unconstrained demand forecast.
• Unconstrained demand forecasts and financial plans are compared to identify gaps. The unconstrained demand forecasts (demand response) are compared to the financial plan that is usually expressed in unit volumes, and the financial plan, which is usually expressed in revenue terms to identify gaps. In order to make a fair comparison, it is especially important to have the capability to express both the unconstrained demand forecast and financial plan in both units and revenue. Once those gaps are identified and additional programs are validated and funded the future unconstrained demand forecast then becomes an input into the S&OP/IBP processes. Although the unconstrained demand forecast and financial plan are independently developed to ensure that the demand forecast and financial plan are directionally in sync, they are compared with each other during the S&OP process where upside/downside risks are documented, and a final supply plan is created based on supply constraints. Many companies refer to the final supply plan as either the shipment forecast or constrained demand forecast. It is actually a supply plan that is derived from the unconstrained demand forecast.

Analytics

• More advanced analytics methods will be integrated into the demand planning process. Demand analysts (or data scientists) will be hired who have advanced statistical knowledge and can apply more advanced models like ARIMA, ARIMAX, and multiple regression using causal factors (i.e., price, intervention variables to capture sales promotion lifts and correct for outliers, along with other causal factors).
• Holistic modeling will replace data cleansing. Companies will holistically model the baseline trend, seasonality, correct for outliers, and model the effects of price and sales promotions using ARIMAX and multiple regression models without cleansing the historical data. Furthermore, they will do it automatically up/down a company's business hierarchy for hundreds of thousands of products by geography, market, channel, brand, product group, product, SKU, demand point, and key account (customer). This will require investment in new demand-driven technology.

• Segmentation of the products will be required to determine appropriate methods to deploy across the company portfolio. When forecasting on a large scale, demand analysts will consider segmenting brands and products based on their value to the company, the availability of data and information, and forecastability. When segmenting demand to determine what methods are appropriate for brand, product group, and product efforts, demand analysts will focus on four key areas:

  1. Low value, low forecastability
  2. Low value, high forecastability
  3. High value, low forecastability
  4. High value, high forecastability

• When evaluating forecast data, companies will look at two key factors: (1) value to the company and (2) forecastability. This conceptual design can be taken further to consider the company's product portfolio as falling into four quadrants:

  1. Slow-moving products with fragmented data across targeted markets and consumers.
  2. New products with little historical sales data (revolutionary new products) or with similarities with existing products (evolutionary products or line extensions). Also, short life-cycle products like fashion jeans, which normally have six-month life cycles.
  3. Fast-moving products that are highly correlated to sales and marketing causal factors, requiring the collection of causal data and information.
  4. Steady-state products with long stable historical demand with distinct time series patterns.
• The company will begin to segment its products to determine how forecastable they are and what methods they should apply, given the strategies surrounding each brand based on market dynamics associated with consumer buying habits, competitive activities, and others.
• Using downstream data to improve forecast accuracy. Consumption based modeling using the MTCA process will be used to link downstream and upstream data using advanced quantitative methods. Consumption-based modeling is designed to integrate statistical analysis with downstream (POS and/or syndicated scanner) and upstream (shipment) data to analyze the business from a holistic supply chain perspective. This process provides both brand management and demand planners with the opportunity to make better and more actionable decisions from multiple data sources (i.e., POS/syndicated scanner, internal company, and external market data).

• In the future, measuring forecast performance is viewed as critical to improving the overall efficiency and value of the demand forecasting process. There are two distinct purposes for measuring forecast accuracy:

  1. Measure how well predicting the actual occurrence or outcome, and
  2. Comparing different statistical models to determine which one fits (models) the demand history of a product and best predicts the future outcome.
• The methods (e.g., MAE, MPE, MAPE, and WAPE) used to calculate forecast error are interchangeable for measuring the performance of a statistical model as well as the accuracy of the prediction.
• The primary purpose for measuring forecast accuracy is not only to measure how accurately the actual occurrence was predicted, but also to understand why the outcome occurred. Only by documenting the design, specifications, and assumptions that went into the forecast will companies begin to learn the dynamics associated with the item(s) they are trying to predict. Forecast measurement will be a learning process, not just a tool to evaluate performance. You cannot improve forecast accuracy unless you measure it. You must establish a benchmark by measuring current forecast performance before you can establish a target for improvement. However, tracking forecast error alone is not the solution. Instead of only asking the question, “What is this month's forecast error?” companies will also need to ask, “Why has forecast error been tracking so high (or low) and is the process improving?” These questions will become practice for those companies who successfully transition to the next generation demand management.
• The results in any single month may be due purely to randomness. Companies will not jump to conclusions or even spend time trying to explain a single period's variation. Rather, they will be reviewing the performance of the process over time and determining whether they are reducing error. Ongoing documentation of the specifics that went into each forecast will be viewed as more important to improving forecast performance.
• FVA will become the most important approach the company can take to measure and evaluate the demand forecasting process. FVA will become the most critical performance metric for evaluating the performance of each step and each participant in the forecasting process. FVA is consistent with a lean approach identifying and eliminating process waste, or non-value adding activities that should be eliminated from the process. Non-value adding resources will be redirected to more productive activities that add value to the company. The FVA performance metrics are a proven way to identify waste in the forecasting process, thus improving efficiency and reducing cycle time. By identifying and eliminating the non-value adding activities, FVA provides a means and justification for streamlining the forecasting process, thereby making the forecast more accurate.

Technology

• Next generation demand planning technology will offer an integrated suite of forecasting, analysis, visualization, reporting, and optimization workbenches built on a common data model along with data integration capabilities. The technology will provide users of all types (i.e., sales, marketing, finance, and demand planning) the information they need through a set of common control towers, dashboards, and scorecards as well as dynamic performance reports.
  • Performance metrics including FVA, MAPE/WAPE, customer service, out-of-stocks (OOS), days of supply will be monitored, tracked, and reported daily, as well as sell through metrics to determine how well customer/consumer sales and marketing tactics are performing.
• The technology will be big data ready. It will include the ability to extract value from nontraditional data sources (e.g., unstructured text from social media), which requires the combination of rich analytics and data management.
  • Included will be data integration applications/tools combined with parallel and grid processing technology to access and analyze big data.
  • These solutions will have the ability to access demand signal repositories where internal transactional data, POS data, syndicated scanner data, as well as other internal/external information reside.
• Large-scale automatic hierarchical forecasting engines with expanded model repositories with a complete array of advanced forecasting methods (i.e., ES, ARIMA, ARIMAX, multiple regression, and others) to model and forecast all products across a company's portfolio. They will integrate consumer demand (pull), model it, and forecast it automatically using data access tools and predictive analytics. The depth and breadth of statistical capabilities will allow demand analysts to model and predict incremental lifts of sales volumes (demand sensing) associated with sales promotions, marketing events, and other related activities, as well as automatically identify and correct for irregular events (outliers) that affect demand. They will provide faster integrated simulation and scenario planning capabilities (demand shaping) that allow demand analysts and planners to test various scenarios using model parameter estimates to determine the impact on the future forecast up/down companies' business hierarchies for hundreds-of-thousands of products. Included will be top-down, bottom-up, and middle-out reconciliation.
• New product forecasting capabilities that combine analogies with sound judgment providing an objective basis for predicting new product demand using as-like or surrogate products based on a product profile. The process will help validate user judgment and allow for the elimination of outliers to produce a better historical set of data for the new product launch.
  • Included will be the capability to integrate unstructured data (i.e., social media, IoT, and others) with structured data supporting the new product launch using sentiment analysis to capture consumer preferences providing critical real-time information to adjust the supply chain inventories, capture future enhancements for sustainability, and monitor sales and marketing tactics.
• Collaborative planning capabilities providing workflow reconciliation up/down the business hierarchy creating averaged or weighted consensus forecasts based on past performance and future projections. Assessment routines can be performed against financial KPIs to determine the impact on revenue management:
  • Configurable workflow approval process with an easy to use Excel interface for users.
  • Planners can easily override forecasts at any level of the product hierarchy and instantaneously see the overall impact across geographies, markets, channels, brands, products, and SKUs down to key customers and demand points.
  • Utilizes FVA to track and measure touch points in the process to identify added value and nonvalue added adjustments.

Goals and Objectives

• Provide adequate support for demand planning. One of the core challenges for companies moving from current state to future is inadequate executive sponsorship and support for demand planning. Business executives may not have experienced the benefits of a more mature demand planning process, and may not understand how to leverage the process in support of the company's business strategy. The end result, senior level managers view the demand planning process as an operations planning process that is required to manage lower level product mix, and such, inadequately supports the development of a demand plan. In fact, the unconstrained demand forecast and resulting demand plan should be viewed as the current market conditions. Also, demand analysts working with sales, marketing, and finance can sense demand signals and shape future demand to determine the appropriate programing to close gaps between the business plan and current market conditions.
• The demand plan aligns with the annual financial plan. The annual financial plan is used as the ongoing operating business plan instead of using it to drive business objectives. Current state companies use the business plan as the locked-in overriding demand plan, regardless of current market conditions. Any demand planning activities are expected to always produce a demand plan that aligns with the annual financial plan. It is very common to close any gaps that develop during the year with unspecified stretch goals (sales targets), or financial plugs. Senior management uses those stretch goals to make adjustments to the current demand plan in hopes that by some miraculous event the sales team will close the gaps, putting the company back on plan.
• Operations planning focuses on short-term execution of supply and replenishment. The demand plans that are produced supposedly reflecting future demand are seen as unreliable by the supply chain function and they are reluctant to make decisions within the tactical time horizon using such a demand plan. As a result, the plans are primarily used for avoiding short-term stock-outs and maximizing revenue.

People

• Demand planners are managers of information and data. Demand planners spend over 80 percent of their time managing information and cleansing data, rather than focusing on using analytics to model and forecast future demand. Many barely understand how to apply predictive analytics to historical demand to measure the effects of sales promotions, correct for outliers, and identify other related causal factors that influence demand. Most have very little knowledge of downstream data, and almost none understand how to integrate downstream (POS/syndicated scanner) data with upstream data (sales orders/shipments). They are essentially supply planners. Demand planners will need to transition from managers of data to demand analysts, with a focus on predicative analytics in the future state.
• There has been little if any investment in training demand planners in the area of predictive analytics. Many companies feel that the primary role of demand planners is to manage the demand planning process, which requires strong communications skills, rather than analytics skills. This has led to a process that is based solely on judgment with very little if any applied analytics. As mentioned in prior chapters the demand planning discipline has digressed over the past decade to managing information, cleansing data, and collaborating across the organization. Although strong communications and collaboration skills are important, analytics are as important if not more important.
• There are no dedicated demand analysts. No resources are assigned for producing a statistical demand forecast other than the system-generated lights out forecast. Demand planning is seen as a minor task of managing data and information that is assigned to a spreadsheet jockey (typically production planners, buyers/planners, or others) in each function that requires a demand forecast.

Process

• Demand plans are developed in silos across the supply chain. Different functions need a projection of future requirements. Without an agreed to demand plan for the business, the functional groups are using plans that they develop independently using the historical volumes that are available and they have supported.
• The finance function plays a dominant role in the demand planning process. The finance function dominates the demand planning process. Their directive is to hold to the financial plan at all costs. In fact, they practice a philosophy called hold-n-roll—hold to the annual plan and roll the miss forward into the next period. In addition, the plans are expressed in financial terms, and there is no effective conversion from revenue to unit volumes.
• The operations planning team has too much influence in the decision-making process. In many current-state companies the operations planning function tends to have a lot of influence in the decision-making process. They are supply-driven, operating in a push environment where the manufacturing plants set the production rates according to their absorption and utilization objectives. They are inside-out focused, pushing product into the channels of distribution based on optimizing production efficiencies, rather than building demand based on consumer preferences.
  • The results are the wrong mix of products, requiring sales and marketing to run discount programs (i.e., sales promotions, temporary price reductions on shelf, and other related merchandising) to pull products through the channels of distribution. This reduces profit margins, thus reducing revenue and profit, not to mention lost market share.
• There is limited or no use of downstream data to uncover consumer behaviors that influence demand. Current-state companies lack the analytic knowledge to use downstream data to provide sales and marketing with insights regarding consumer/customer demand and apply those insights in a meaningful way to show the effects across the supply chain. Current-state companies focus all their forecasting resources toward predicting historical shipment (supply) volumes. Those shipment forecasts are used as the basis for calculating the future supply requirements. However, this can be quite detrimental to supporting seasonal demand and calculating the lifts associated with sales promotions, not to mention predicting new product launches.

• The sales & operations planning (S&OP) process has failed to deliver the benefits due to vertical alignment. Most current state companies are actually doing OP (operations planning), rather than S&OP. The process is vertically aligned, managed by a demand planning team that reports into the operations planning organization with a focus on creating a supply plan (shipment forecast). The VP/director of supply chain management along with finance makes the final decisions as to the supply plan. This creates a supply-driven environment with little attention to downstream data, demand generation, and revenue and profitability. The key discussions are based on reducing inventory costs, customer service, and on-time delivery.

  Meanwhile, the commercial business (sales/marketing) is not in attendance. The S, which stands for sales and marketing, perceives no real benefits from attending, as their core responsibilities are demand generation and increasing market share, revenue, and profit. They rely almost exclusively on downstream data (POS/syndicated scanner) to manage the business. Plus, they have virtually no accountability for demand forecast accuracy.

• There is no alignment between demand and supply. There is limited or no capability to produce a truely integrated demand plan that can be synchronized across the supply chain. As a result, companies tend to have unbalanced inventories that are based solely on historical supply (shipments), rather than true demand.

Analytics

• There is limited capability to improve demand forecast performance. This is because of the lack of analytics skills and statistical methods available in the company's first generation ERP systems. Demand forecasting and planning are seen more as an art than a science, with multiple touch points in the process where adjustments are made completely based on judgment. There are no performance metrics capable of using accuracy measurements to improve process efficiency.
• There is no standard forecast accuracy measurement process other than MAPE. The forecast accuracy is measured on an ad hoc basis, and the calculation methods vary between the businesses, regions, and functions.

Technology

• There is a lack of standardization and flexibility within companies' first-generation ERP solutions. The demand management modules lack advanced statistical methods (e.g., ARIMA, ARIMAX, multiple regression), and were not designed to integrate downstream data into the process. They lack the capabilities to sense demand signals, shape future demand, and link demand to supply using consumption based modeling. This is due to the inflexibility of their technology for gathering source data requiring cross-referencing and mapping, restructuring, and scrubbing before it can be used for analysis and planning. This is also due to the lack of advanced predictive analytical methods that can do holistic modeling. As a result, demand planners spend 80 percent of their time managing data and analytics. This leads to islands of Excel spreadsheets driven by individual departmental purposes creating different views of the same data.
• Several systems contain data that could be valuable for improving the demand planning process. These systems (e.g., ERP systems, data warehouses, data marts, and business intelligence (BI) systems) may be globally dispersed into regional data centers or perhaps even local servers, with no single group or function having a clear picture of what information is available and where.
• Processing information is slow and fragmented across the organization, making it difficult to utilize standard templates. Reporting is Excel based, with varying degrees of detail and virtually no drilldown capabilities. The entire process is manual with many people burning the midnight oil to gather all the appropriate information and enter it into their ERP systems for processing.

As companies start developing the future roadmap to address the gaps identified, some interdependencies must be taken into consideration. Moving from current state to future state will require improvements in the demand planning process, as well as the resulting demand forecasts alone will not produce significant business benefits. The demand planning efforts must be closely aligned with the following adjacent areas:

• Demand planning: The demand planning process must be able to integrate downstream data as well as consumer information into the demand forecast. Demand planning process improvement must be coordinated not only with supply planning requirements but also those commercial requirements so that the demand forecast can be translated into information that is relevant and usable across the entire supply chain. In this definition, supply chain includes sales and marketing. It is strongly recommended that the demand management transition be placed in parallel to sales and marketing tactical and strategic initiatives. Transferring accountability and ownership to the commercial organization will be essential to fully engage the commercial organization. This is a radical change in the process that will require a champion to lead the change management process. Corporate behavior changes by the commercial, supply chain, finance, and executive management organizations will take time and require small changes to gain adoption. Over time, the changes need to become part of the new corporate culture to assure sustainability.
• Pretechnology roadmap: As companies migrate from current state to future state with their demand management initiatives, one of the first requirements that arises is the need for new demand planning technology. In reality, companies who are in current state of their demand planning process are often not ready to define the specific requirements for an integrated technology solution, nor are they able to effectively use advanced planning analytics. The best course of action is for the demand planning continuous improvement activities to focus on the process design by assessing the current state and blueprinting the future state. During the interim, investment in statistical skills training, and implementing FVA will begin to improve forecast accuracy and add efficiency to the process. Data integration and investment in an enterprise data warehouse along with the necessary technology (i.e., servers, grid processing, in memory processing, and others) is necessary before investing in a new demand management solution. These are necessary prework efforts that are required not only to identify specific data sources and technology gaps in the demand planning process but also to collaborate with the IT team to find the right technology foundation layer to support solution upgrades or acquisitions.
• Alignment of performance metrics and objectives. One of the core reasons for siloed planning across the organization is due to independently set objectives created by each department involved in the process. When the functional objectives across the supply chain are set in silos for different purposes, or intent, it causes conflict and political posturing. Due to the conflicting performance targets, and because each group creates its own bias forecasts in order to meet those different objectives, no single plan will enable all the participating groups to achieve their goals. In support of the demand planning initiative, the business and finance leaders must plan for horizontal cross-functional metrics alignment that will support a one-consumer-based forecast from which all plans can be created. Shipment forecasts are supply plans, not a forecast.
• Horizontal alignment of the S&OP process. Not only does there need to be horizontally aligned performance metrics, but the inclusion of the commercial teams (sales and marketing). According to the demand management book written by Oliver Wight (Crum, Palmatier 2003), companies cannot synchronize demand and supply without sales and marketing participation. Integrating downstream data, embedding demand analysts in the commercial organization, and introducing consumption based modeling will not only create accountability and ownership of the unconstrained demand forecast, but will provide rationale for the sales and marketing organizations to participate in the S&OP process. Although demand planning is a foundational part of the S&OP process it can only be effective if aligned with sales and marketing and tasked to create an unconstrained demand forecast reflective of sales and marketing tactics and strategies, not with the intent to create a supply plan. The supply plan should be created by the operations planning team (e.g., supply planners). As companies move to future state the demand planning process needs to be more aligned to sales and marketing, not operations planning. Ideally, the demand planning team should report into a neutral organization in order to be unbiased to properly support the horizontal S&OP process roadmap. The unconstrained demand forecast is constrained through the S&OP process to create the final sales, marketing, and supply plans. Even more so, as companies transition from S&OP to IBP (integrated business planning).

Goals and Objectives

• It is very important to develop a clear strategy and vision for the goals of the demand planning process. The overall demand management vision does not need to be an exact plan, but rather, a strategy of how the end-state demand planning process can add value to the overall business strategy. As you develop the demand planning strategy, include all the relevant changes in people, process, analytics, and technology required to form an overall demand planning vision for your business.
• Develop a migration path for demand planning that demonstrates a clear roadmap from current state to future state with milestones, risks, and benefits. Once leadership commitment is confirmed to support and sponsor the development of the new demand planning process, start developing a formal project plan outlining the timing, work, and responsibilities with milestones. In the plan, set the overall end objective to future state across all four dimensions of demand planning with clearly defined goals and objectives. Gain executive and business leaders' support building a business case to show the value of the new demand planning process. While forecast and demand planning errors and disconnects affect most business metrics, the best place to start building the business case benefits is around demand generation, revenue, and profit improvements, not just improved customer service levels, and inventory costs reductions, waste, and working capital.
• Take a phased approach starting with a small subsector of your business to pilot the new process, analytics, and technology to show a quick win. Identify a product line or business unit that is most manageable because of the size, limited complexity, cooperative leadership, and/or because it has the best potential for a quick win. Furthermore, start with high revenue (fast-moving) products that are considered growth products where there is sufficient internal data, POS/syndicated scanner data coverage, and causal factors. Those products not only are high volume, but are also fairly stable with high value.

People

• Establish the benefits and secure approval to hire demand analysts (data scientists) to support demand sensing and shaping activities to complement the demand planners. The business case for these new analytics roles should emphasize the advantage of enhanced statistical forecast accuracy due to the introduction of causal information, as well as the integration of consumption based modeling in support of demand generation, sales tactics, and marketing strategies. This strategy has been shown to improve inventory performance and customer service levels, thus lowering costs and improving revenue and profit.
• Create more structured demand analyst and demand planner roles and responsibilities across the organization. Place more emphasis on creating unconstrained demand forecasts and write new job descriptions for the two roles. Also, invest in analytics training for all the demand planners.
• If no one is doing demand analysts activities, then consider recruiting and hiring several experienced professional demand analysts with strong statistical skills. The average ratio based on recent experience is one demand analyst per four demand planners. Also, consider creating a COE (center of excellence) organization, to which demand analysts would report.

Process

• Design a structured demand planning process that includes the commercial side of the business that encourages accountability and ownership. Develop a process roadmap with detailed descriptions of the process steps, workflow, required inputs, roles and responsibilities, and expected outcomes. Clearly outline how demand planning and the S&OP/IBP process(es) integration are aligned with defined roles and responsibilities.

• Determine the maximum and minimum data requirements. This is both to get the process started and to not underestimate data collection, processing, and storage capabilities. However, scope your technology requirements to the maximum data requirements, so as not to underestimate the size of the server(s).

  This happens quite often with initial implementations. Once the data requirements are known, find out whether the required data are currently available. If available, then find out where the data are located, and if not, where they can be found. This is also key to data integration, Extract, Transform, and Load requirements, and sizing of the data model.

• Identify the A products (SKUs) for the first phase of the new demand planning process. For the purposes of the initial demand planning process setup, the A items are those that make up roughly 20 percent of the product portfolio, but 80 percent of the revenue. As you demonstrate, the demand planning capabilities of the A items add SKUs to the planning process over phases that are digestible for the organization.
• Develop a demand forecast for the chosen businesses. Where appropriate, use POS/syndicated scanner data, linking it to sales orders (or shipments) as they are closer to true demand. In either case, it makes sense to develop forecasts by geography, market, channel, brand, product group, product, and SKU levels in a hierarchical format. Hierarchical forecasts tend to be more accurate and easier to manage with the right enabling technology. The best place to start may be with sales orders (or shipments) alone, but with the intentions of including causal factors to measure promotion lift factors, correct for outliers, and forecast future sales orders. Comparing it to actual sales orders (or shipments) as a starting point could be a good way to defuse resistance to change from the operations planning group, but begin the transition to consumption-based forecasting where data are available and make sense. This will begin to draw in the commercial team to the demand planning process, as well as provide rationale to participate in the S&OP/IBP process(es).
• Start engaging the sales, brand/product management, and/or other customer facing teams in the process. Once you have the initial unconstrained demand forecast developed, highlight a manageable number of key brands/products and causal factors. Then, encourage the sales and marketing teams to review the list and provide feedback on the chosen products and causal factors to be integrated into the consumption-based forecasts. It is important to emphasize the benefits of improved demand planning in support of demand generation to the sales and marketing teams. As demand forecast accuracy improves, the entire supply chain (including commercial) will be able to better support sales promotions, adhere to customer delivery commitments, plan for new business, and, due to increased customer collaboration, improve the company's credibility with the customer base.

Analytics

• Consider segmenting the company product portfolio into four categories. These categories are (1) slow-moving products, (2) new products, (3) fast-moving products, and (4) steady-state products. Start by using time series methods (e.g., exponential smoothing—seasonal/nonseasonal, ARIMA—seasonal/nonseasonal models) to model and predict future demand for the steady-state products. Then, introduce ARIMAX (ARIMA with intervention and causal factors) to model the fast moving growth products capturing the effects of sales promotions, price, and other causal factors. Finally, address the slow-moving new products. Show small analytics wins (successes) to gain confidence and trust in the analytics.

• Use performance metrics to monitor, track, and improve forecast accuracy and process efficiency. Develop a standard for forecast error measurement. The commonly used forecast accuracy measurement is MAPE, but also consider WAPE (weighted absolute percentage error across product groups). Clearly define where the data comes from, how the metrics are to be calculated, what are the lag times (one- to three-month frozen horizon), how to determine the appropriate lag times, the level of hierarchy for the measurement, and whether there are any specifics to the inclusion or exclusion of special SKUs.

  Introduce FVA as soon as possible to improve not only forecast accuracy but also efficiency in the demand planning process. Eliminate or reduce the non-value-added touch points, and continue to include the value-added touch points.

Technology

• Before a company can consider investing in new technology, they need to formalize the demand planning process, identify and source the data requirements, and finalize technology requirements. It is important that the technology has the capabilities, functionality, and proper workflow to support and enable the future state. The biggest mistake companies make is purchasing the technology before the process and workflow activities have been documented, assessed, and formalized. This also includes identifying the demand planning roles and responsibilities, including skill-set requirements.

  In many cases, companies go ahead and purchase the technology solution before they have completed the process and workflow requirements without identifying the data requirements and how to source the data. Before the technology can be implemented, there must be a single view of all the data required to support the process. This includes not only structured and unstructured data. The data must be quality ready to assure optimal technology performance. Also, the enterprise data warehouse (or demand signal repository), including supporting data marts, need to be optimized to support both descriptive and predictive analytics. In many cases, the data are not in a high-quality state and are almost always optimized for descriptive analytics (reporting), not predictive analytics.

• It is important to develop standard templates for data gathering and for presenting the demand planning results, which includes performance metrics. It is recommended to allow flexibility with the tools and files used in analysis and planning, as long as there is adherence to the data source and master data. Ensure that your demand planning source data include all demand markets, channels, brands, product groups, products, and SKUs. In addition, identify and document the systems that contain the demand information for all regions and all customer types.
• Start working with the IT organization to establish a governance model for the demand planning master data. The demand planning process requires quality data for products, customers, and internal locations.
• Delay making a big technology investment until you are able to clearly articulate the requirements for your technology solution. Once the technology requirements are documented it is recommended to write a request for information (RFI) from three to five potential software vendors. This provides the opportunity to review the capabilities and functionality across several software vendors and validate your requirements.
• Upon review of the RFIs select one to three software vendors to complete an RFP (request for proposal) to evaluate and choose the appropriate technology solution. This will require a demo of each solution, and possibly a POV (proof-of-value) using a subset of your data (select one or two markets, brands, product groups, products, and SKUs). This will include a business case with the results of the POV along with a formal proposal from each software vendor.