The CPG industry is characterized by low involvement products bought and used frequently. Because of low switching costs for the shopper, brand loyalty is fickle. If the right product is not available at the shelf, the shopper may select a competitor's product.

To ensure shelf availability, the CPG manufacturer needs to effectively execute: 1) producing the right product at the right time, and 2) positioning the right product at the right distribution center within its distribution network at the right time. Subsequently, when the retailer's order is received, the product can be shipped to the retailer's DC for onward shipment to the right store for placement on the right shelf.

Thus, demand forecast plays a crucial role in ensuring shelf availability. In turn, the accuracy of demand forecast depends upon placing and pacing.

What are Placing and Pacing?

The consensus demand plan drives the CPG manufacturer's master production schedule as well as the positioning of products at various DCs. Consequently, fulfillment depends on the CPG manufacturer accurately pointing forecast to the right shipping location, which is called placing the forecast. Incorrect placing of forecast can lead to a disconnect between demand forecasting and fulfillment planning.

There is another potential disconnect between demand forecasting and fulfillment planning -- time granularity. Fulfillment planning is done at a more granular level (daily) than demand forecasting (weekly). Therefore, weekly demand forecasts require disaggregation to a daily level. The process of allocating weekly demand to a daily level is called pacing the forecast.

Placing and Pacing Challenges

Challenges with placing arise from assumptions regarding SKU-DC-customer sourcing relationships. When customer orders do not follow these assumptions, potential problems arise. Typically, CPG manufacturers re-route such orders to the appropriate DC via manual intervention. However, there are instances when such orders get fulfilled from an incorrect source location. For example, a specially priced order for direct plant shipment is incorrectly placed at a DC and shipped from there because of product availability. This could lead to potential out-of-stocks (OOS) at the DC, and the resulting inability to fulfill regular orders at the DC would negatively impact shelf availability.

A common challenge associated with pacing is for promotional orders. Promotional orders may not be received for the period (week) for which they were forecast. If not appropriately identified, promotional orders arriving earlier than anticipated lead to an overestimation of demand and could potentially result in OOS for short lead-time orders followed by excess inventory later. Conversely, promotional orders received later than anticipated result in holding excess inventory until the order is actually received. Either situation could prevent the CPG manufacturer from achieving the planned ROI from its promotional campaign -- a major issue since approximately 15 percent of CPG manufacturers' annual revenue is spent on trade promotions.

Traditional Approaches

Challenges with placing are typically addressed by one or a combination of: 1) better collaboration (CPFR, VMI) with strategic customers, 2) establishing SKU-DC-customer sourcing relationships to point forecast to the right source DC, and 3) having fulfillment planners manually monitor near-term customer orders and manually intervene to align with DC-specific fulfillment requirements.

Challenges with pacing are typically addressed by either an ad-hoc basis or a scientific basis. Ad-hoc basis uses "gut" feel to allocate weekly forecast to a daily level. Subsequently, a rule-based approach may be used to address/resolve any misalignment between pacing and actual customer orders as the week progresses. Scientific basis periodically analyzes historical demand data to identify day-of-the-week demand patterns that are used to allocate weekly forecast to a daily level. Subsequently, a rule-based approach is used to address/resolve any misalignment between pacing and actual customer orders as the week progresses.

Traditional approaches can capably handle products with relatively steady demand patterns. However, problems arise when facing certain scenarios that are common to the CPG industry.  One example is promotions where pacing is not aligned with the actual promotional order. In this scenario, the promotional order is received either earlier or later than the time period for which it was forecast. A second example is forward-buying situations where the customer places a single order for a quantity that would normally cover multiple orders across multiple time periods. Without close collaboration between fulfillment planner and demand planner to identify such situations and make necessary changes in the placing, this situation results in overestimation of demand.

This could possibly explain why The GMA 2008 Logistics Survey indicated a national average weekly forecast error (measured as Mean Absolute Percent Error) of 45 percent at the shipping location level. This level of inaccuracy is unacceptable for an extremely competitive industry with low margins. Therefore, better approaches are needed to address challenges with placing and pacing. 

Emerging Approaches

New approaches are emerging that recognize the demand dynamics with respect to placing and pacing. They work in conjunction with existing demand forecasting systems and replace traditional placing and pacing in the near-term period. Beyond this pre-determined near-term period, forecast from the traditional demand forecasting system is used. These emerging approaches can be broadly categorized as regenerative and adaptive.

The regenerative approach uses historical customer order and shipment information in conjunction with advanced pattern recognition and statistical techniques to develop a forecast model for the near-term period. This forecast model, after validation for accuracy and fine-tuning, uses real-time customer order and shipment information to identify market dynamics and regenerate forecasts for the near-term period. For this pre-determined near-term period, a new forecast generated at a daily granularity replaces the traditional forecast. Given its daily granularity, the regenerative approach also eliminates the traditional time granularity disconnect between demand forecast (weekly) and fulfillment planning (daily). Thus, it simultaneously addresses issues with placing and pacing.

The adaptive approach tackles potential disconnects between placing and pacing with respect to actual customer orders fundamentally differently than the regenerative approach. It senses changes in demand in the near-term period and resolves disconnects between pacing and actual customer orders by enabling customer-specific demand-type (e.g., base, promotion) forecast consumption. It adjusts and ensures that an under-forecast for a specific customer does not consume the forecast for other customers. An adaptive approach works within the confines of the traditional forecast and uses real-time customer-specific actual demand to adapt to market dynamics.

A point to note is that the traditional forecast continues to be an important input for both emerging approaches. Any inherent inaccuracy in the traditional forecast in the near-term impacts both emerging approaches. This impact, in the case of an over-forecast, is more pronounced for the adaptive approach since it operates within the confines of the traditional forecast.

Some Considerations

At a minimum, the emerging approaches require the following considerations for a successful implementation:

Business process/workflow: The emerging approaches address near-term demand forecasting in a fundamentally different way than done previously. The regenerative approach, especially, is the equivalent of a black-box approach once the initial near-term forecasting model is built, fine-tuned, and validated for accuracy on a weekly basis. The adaptive approach may require more user involvement in the initial set-up as well as for ongoing operations. Therefore, demand planning workflow requires modification to reflect the chosen approach. 

Change management: Identification/documentation of changes in work procedures and training in new work requirements are necessary to sustain process transformation. This is especially true for the adaptive approach since it may require more operational involvement from the user than the regenerative approach.

Data sourcing: Both emerging approaches are data intensive and some of the required data may reside on enterprise systems other than the ones used to support traditional approaches. Therefore, proper sourcing of data is a crucial dimension of a successful implementation. To add to the complexity, sometimes the data might not be available at the desired granularity. In such instances, new enterprise processes and systems may be required to ensure data availability at the desired level of granularity.

Data integration: Once all the data sources to support the emerging approaches have been identified, they need to be integrated seamlessly to provide relevant data on a pre-determined schedule. Additionally, the output from these approaches needs to be integrated with: 1) the fulfillment planning system, and 2) business intelligence system for operational analytics.

Other consideration: On achieving an appreciable improvement in forecast metrics, safety stock parameters need to be updated to deliver on the promise of these emerging approaches - improved customer service levels with reduced inventory levels.

Conclusion

Selection of a specific emerging approach depends on whether the CPG manufacturer wants to focus on placing and pacing near-term demand forecasts at just the DC-level (regenerative approach) or at a more granular customer-level (adaptive approach). In either case, CPG manufacturers would derive benefits over the traditional approach to placing and pacing near-term demand forecasts. CPG manufacturers who have adopted the new approaches have improved their customer service metrics through improved forecast accuracies and their operational metrics through resulting reduction in safety stock investments. 

Sunil Desai is a principal at Infosys Consulting, and Venkatesan Ramesh is senior consultant for enterprise solutions at Infosys Technologies Limited. Visit www.infosys.com.