Executive Briefings

Benefits and Drawbacks of Demand Management vs. Demand Sensing

One of the new buzzwords in the demand planning arena is demand sensing. Developed around 2003, demand sensing has slowly been grabbing the interest of the CPG, energy, food, beverage, and chemical industries. Often viewed as an alternative to demand management, demand sensing is anything but. Let's compare the two.

Benefits and Drawbacks of Demand Management vs. Demand Sensing

Purpose

The purpose of demand management is to devise a manufacturing plan aligning supply and demand. Demand management consists of coinciding processes that include: sales and operations planning, demand planning, statistical forecasting, sales plan reconciliation and key performance indicator monitoring.  These activities tie together to help demand planners achieve a realistic demand plan dictating how much product is manufactured and procured within an organization. In turn, operations use the demand plan as the main driver for annual operational planning within plants and distribution centers.

Demand sensing is a program usually sold by a third-party vendor that uses proprietary heuristics to improve the short-term results of demand management. Demand sensing is applied to the short-term forecast and automatically adjusts the consensus forecast within a 4- to 12-week time frame. The main purpose of demand sensing is to improve forecast accuracy slightly outside of lead time to help prevent unwanted product from reaching the inventory pipeline. Demand sensing also provides visibility to variations in supply and demand allowing an organization to prevent out-of-stock situations due to overselling product. Demand sensing does not generally incorporate the total supply chain, but focuses on select products and customers.

Data

The result of demand management is a consensus sales plan that is applied to an organization's wholly owned locations directly supplying a customer. The supplying location "backwards schedules" based on the sales demand due date to generate demand in the form of purchase orders or inventory at each level in the supply chain. Ultimately the plant locations receive a distribution plan which translates into a production plan. Therefore, demand management focuses on DC-to-DC or plant-to-DC shipments as a primary input to determine production needs across an organization.

Demand sensing tries to focus on real-time market data and trends. The most common input data used by demand sensing is point of sale data (POS) which is usually provided by the customer. Companies receiving POS data usually create a data repository which the demand sensing heuristics use to create and adjust the short-term forecast. In addition, demand sensing heuristics allow the user to consider multiple data inputs to determine the new forecast. For example, a demand sensing tool can look at net changes in POS data at a customer location, and then compare the impact of those changes on DC to customer shipments. This allows the forecast to be updated at the DC while considering order mandates like minimum order quantities, pallet dimensions, case quantities, etc.  A demand sensing tool can add these quantities back into the forecast changing the sales forecast to a ship forecast based on customer sales.  This allows the plants and distribution centers to see the true customer needs.

Because of the robustness of demand sensing, data accuracy is important. Since POS data generally resides at the customer, a certain amount of collaboration may be needed to obtain this information.  Since demand management is based purely on shipment history, this should be relatively accessible for any organization.

Focus

All tools have certain areas of applicability. Demand management encompasses gathering data around promotions, sales and operational challenges to devise a sales plan that can be executed by manufacturing. Demand management is generally applied across all items and locations in the supply chain. Many consider this a flaw with demand management because items with completely different characteristics - demand patterns, seasonality and spikes in demand - have the same statistical models and methodologies applied to them. Deciding the correct methodologies and models is purely at the discretion of the demand planner. Today, most fortune 1000 companies have established some form of demand management processes from stat forecasting to sales and operations planning meetings.

A differentiator is that demand sensing does not necessarily span all items. Since demand sensing adjusts short-term forecasts (immediate 4 to 12 weeks), the decision of whether or not to use demand sensing is based on production cycle times or lead time for distributors.  For example, it does not make sense to implement a demand sensing tool to adjust the first 12 weeks of forecast if the lead time to produce is 14 weeks.  In the example, adjusting the immediate forecast will do nothing to affect your production volume as production has already started.  This is why demand sensing is catching on quickly at CPG companies and other organizations with quick production cycle times.  Additionally, if the demand sensing tool is constantly increasing short-term forecast due to continuous over selling but production upside is minimal, implementing demand sensing may not make sense.

If an organization's demand management processes are not relatively mature, demand sensing may not be viable as demand sensing aims to increase forecast accuracy. Depending on demand planning process maturity, an organization may be able to achieve significant gains in forecast accuracy without the help of demand sensing.Therefore, demand sensing tools are best utilized when an organization's demand planning processes are relatively mature.

Forecast Horizons and Time Periods

Demand management is a forward looking methodology that reconciles monthly historical data to predict the future. To some degree, demand management follows the theory "history will repeat itself." Typically, demand management focuses on last month's forecast accuracy and market intelligence to make decisions about next month's plan.  This occurs in monthly buckets or fiscal periods for a rolling 12 to 18 months.

Demand sensing uses last week's sales data and market intelligence to adjust the forecast for the following week. This makes the forecast based on the most current market trends and sales data available. Demand sensing is for short-term use and applied to a rolling 4- to 12-week horizon only.  After the 12-week horizon the consensus forecast gathered through the demand management process would be used to dictate demand.

Summary

Demand sensing is a bolt-on tool that helps improve the results of demand management in the short-term forecast. In no way, is demand sensing a replacement for demand management or any of its processes. Demand management is aimed at providing an accurate manufacturing plan across all items within an organization by utilizing historical shipment data, while demand sensing aims at increasing short-term forecast accuracy across several key items or customers by utilizing a combination of shipment and point-of-sale data. Demand sensing is not for all organizations. Organizations may be able to achieve sufficient gains in forecast accuracy by improving robustness of existing tools and processes.  However, for organizations struggling to find additional methods to improve forecast accuracy, demand sensing is a viable option. Both demand management and demand sensing are effective tools. Each has their drawbacks and purpose making it imperative for organizations to put plenty of thought into selecting the right tool.

Source: Plan4Demand

Keywords: demand sensing, forecast accuracy, demand variability, demand volatility, supply chain solutions, demand planning, supply chain management

Purpose

The purpose of demand management is to devise a manufacturing plan aligning supply and demand. Demand management consists of coinciding processes that include: sales and operations planning, demand planning, statistical forecasting, sales plan reconciliation and key performance indicator monitoring.  These activities tie together to help demand planners achieve a realistic demand plan dictating how much product is manufactured and procured within an organization. In turn, operations use the demand plan as the main driver for annual operational planning within plants and distribution centers.

Demand sensing is a program usually sold by a third-party vendor that uses proprietary heuristics to improve the short-term results of demand management. Demand sensing is applied to the short-term forecast and automatically adjusts the consensus forecast within a 4- to 12-week time frame. The main purpose of demand sensing is to improve forecast accuracy slightly outside of lead time to help prevent unwanted product from reaching the inventory pipeline. Demand sensing also provides visibility to variations in supply and demand allowing an organization to prevent out-of-stock situations due to overselling product. Demand sensing does not generally incorporate the total supply chain, but focuses on select products and customers.

Data

The result of demand management is a consensus sales plan that is applied to an organization's wholly owned locations directly supplying a customer. The supplying location "backwards schedules" based on the sales demand due date to generate demand in the form of purchase orders or inventory at each level in the supply chain. Ultimately the plant locations receive a distribution plan which translates into a production plan. Therefore, demand management focuses on DC-to-DC or plant-to-DC shipments as a primary input to determine production needs across an organization.

Demand sensing tries to focus on real-time market data and trends. The most common input data used by demand sensing is point of sale data (POS) which is usually provided by the customer. Companies receiving POS data usually create a data repository which the demand sensing heuristics use to create and adjust the short-term forecast. In addition, demand sensing heuristics allow the user to consider multiple data inputs to determine the new forecast. For example, a demand sensing tool can look at net changes in POS data at a customer location, and then compare the impact of those changes on DC to customer shipments. This allows the forecast to be updated at the DC while considering order mandates like minimum order quantities, pallet dimensions, case quantities, etc.  A demand sensing tool can add these quantities back into the forecast changing the sales forecast to a ship forecast based on customer sales.  This allows the plants and distribution centers to see the true customer needs.

Because of the robustness of demand sensing, data accuracy is important. Since POS data generally resides at the customer, a certain amount of collaboration may be needed to obtain this information.  Since demand management is based purely on shipment history, this should be relatively accessible for any organization.

Focus

All tools have certain areas of applicability. Demand management encompasses gathering data around promotions, sales and operational challenges to devise a sales plan that can be executed by manufacturing. Demand management is generally applied across all items and locations in the supply chain. Many consider this a flaw with demand management because items with completely different characteristics - demand patterns, seasonality and spikes in demand - have the same statistical models and methodologies applied to them. Deciding the correct methodologies and models is purely at the discretion of the demand planner. Today, most fortune 1000 companies have established some form of demand management processes from stat forecasting to sales and operations planning meetings.

A differentiator is that demand sensing does not necessarily span all items. Since demand sensing adjusts short-term forecasts (immediate 4 to 12 weeks), the decision of whether or not to use demand sensing is based on production cycle times or lead time for distributors.  For example, it does not make sense to implement a demand sensing tool to adjust the first 12 weeks of forecast if the lead time to produce is 14 weeks.  In the example, adjusting the immediate forecast will do nothing to affect your production volume as production has already started.  This is why demand sensing is catching on quickly at CPG companies and other organizations with quick production cycle times.  Additionally, if the demand sensing tool is constantly increasing short-term forecast due to continuous over selling but production upside is minimal, implementing demand sensing may not make sense.

If an organization's demand management processes are not relatively mature, demand sensing may not be viable as demand sensing aims to increase forecast accuracy. Depending on demand planning process maturity, an organization may be able to achieve significant gains in forecast accuracy without the help of demand sensing.Therefore, demand sensing tools are best utilized when an organization's demand planning processes are relatively mature.

Forecast Horizons and Time Periods

Demand management is a forward looking methodology that reconciles monthly historical data to predict the future. To some degree, demand management follows the theory "history will repeat itself." Typically, demand management focuses on last month's forecast accuracy and market intelligence to make decisions about next month's plan.  This occurs in monthly buckets or fiscal periods for a rolling 12 to 18 months.

Demand sensing uses last week's sales data and market intelligence to adjust the forecast for the following week. This makes the forecast based on the most current market trends and sales data available. Demand sensing is for short-term use and applied to a rolling 4- to 12-week horizon only.  After the 12-week horizon the consensus forecast gathered through the demand management process would be used to dictate demand.

Summary

Demand sensing is a bolt-on tool that helps improve the results of demand management in the short-term forecast. In no way, is demand sensing a replacement for demand management or any of its processes. Demand management is aimed at providing an accurate manufacturing plan across all items within an organization by utilizing historical shipment data, while demand sensing aims at increasing short-term forecast accuracy across several key items or customers by utilizing a combination of shipment and point-of-sale data. Demand sensing is not for all organizations. Organizations may be able to achieve sufficient gains in forecast accuracy by improving robustness of existing tools and processes.  However, for organizations struggling to find additional methods to improve forecast accuracy, demand sensing is a viable option. Both demand management and demand sensing are effective tools. Each has their drawbacks and purpose making it imperative for organizations to put plenty of thought into selecting the right tool.

Source: Plan4Demand

Keywords: demand sensing, forecast accuracy, demand variability, demand volatility, supply chain solutions, demand planning, supply chain management

Benefits and Drawbacks of Demand Management vs. Demand Sensing