Win in the flat world
Effectively Managing Demand Variability in
CPG Industry
With focus on CPG manufacturer
– Venky Ramesh
White Paper
Abstract
Demand forecast serves as a key input for consumer-packaged goods (CPG)
manufacturers, driving their production and inventory deployment plans.
Higher quality forecasts invariably translate into higher customer service
levels, within planned costs. Among others, the two primary factors that
have a huge bearing on the quality of the forecast are demand uncertainty
and demand variability. While there are robust processes in place to deal
with demand uncertainty, there is an urgent need in the industry to tackle
the issue of demand variability.
This paper focuses on how CPG manufacturers can manage demand
variability by incorporating the voices of the customer, consumer, reason
and caution into the demand planning process
For more information, contact askus@infosys.com
July 2009
An introduction - Importance of managing uncertainty and variability in
demand
The terms demand uncertainty and demand variability are often used interchangeably. It is important to
note that demand uncertainty is different from demand variability. Demand uncertainty refers to a limited
knowledge about what is going to sell or the inability to predict demand. It is measured using the
forecast error. On the other hand, demand variability refers to the range of values for demand, which is
variable based on effort in marketing or promotions, seasonality, holidays, special events and other
extrinsic factors. Demand variability is measured through variability of the historical demand. The table
summarizes the differences between uncertainty and variability in demand:
Demand Uncertainty Demand Variability
General concept of doubt or The range of values for demand
Definition limited knowledge about what is
going to sell
Limited knowledge of the market Effort in marketing or promotions,
Cause and the environment seasonality, holidays, special
events and other extrinsic factors
Mean Absolute Percentage Error Variability of historical demand
Common metrics
(MAPE), Bias
Minimize forecast error Obtain insights into the demand
Planning objective pattern to match forecast curve as
close to the demand curve
Long-term impact: Inaccurate business and resource planning (production
capacity, manpower and financial resources).
Negative
repercussions Short-term impact: Inaccurate decisions related to production planning,
material planning and inventory deployment. Inaccurate safety stock
leading to higher inventory costs or customer service issues.
Given the negative repercussions that demand uncertainty and variability can have on supply chain
planning, it is imperative that CPG manufacturers take steps to avoid them. This need is underscored as
the business model of CPG manufacturers depends on their ability to fulfill customer orders from the
stock. CPG manufacturers manufacture products based on forecast and stock it at their distribution
centers (DC). This ensures that when customer orders come in, the products are available for shipment
from the DC inventory. This approach ties the manufacturer's service levels to their ability to produce a
quality forecast. To add to the challenge, retailers are placing orders with increasingly shrinking lead
times (24 - 72 hours or lesser). This gives manufacturers very little time to sense any change in forecast
or to respond appropriately within planned cost and promised service levels. In order to achieve high
service levels in the face of changing demand, it is crucial for manufacturers to undertake initiatives to
reduce uncertainty and variability in demand.
The CPG manufacturers have responded to the problem of demand uncertainty through investments in
forecasting solutions. This enabled them to predict with reasonable certainty what they will sell in near
future, in aggregate weekly terms. As their degree of confidence increased in their weekly forecasts, their
focus is now shifting towards managing the demand variability occurring during order fulfillment planning.
Fundamental factors causing demand variability
In order to manage demand variability, it is important to understand the key factors that cause fluctuations
in demand. These include:
• Demand shaping activities: Consumer packaged goods are characterized by low-value, frequently
purchased products. Purchase decisions are often made impulsively in the store. To entice shoppers
into buying their products, CPG manufacturers constantly undertake demand-shaping initiatives,
such as in-store promotions. While this results in a sales lift and subsequent increase in shipment
orders, the massive out-of-stocks that occur during promotion may result in order overstatement
post the promotion period. The level of uncertainty arising due to promotions often results in a
highly variable demand pattern, adding to the forecasting woes of the manufacturer.
• Demand latency: CPG manufacturers rely on retailers to sell their products to the consumers.
While this strategy lets the manufacturer and retailer focus on their key strengths, it also leaves
the manufacturer one level removed from the consumer. When the consumer picks up a product
at the store shelf, a demand signal goes out to the manufacturer in the form of a purchase order
(PO), all the way through the store backroom and retailer distribution center. Replenishment policies
at each of these stock keeping entities at the retailer may lead to a time lag of anywhere between
2-4 weeks for the consumer demand signal to translate in the form of a retail order to the
manufacturer. This could result in the manufacturer getting a distorted picture of demand signal
in terms of timing and quantity. Due to this demand latency, a small variation in demand at the
shelf ends up as a highly variable demand signal for the manufacturer.
How can the CPG manufacturers manage demand variability?
CPG manufacturers can effectively manage demand variability by implementing demand-planning
processes. This will not only help them reduce any demand variability arising from the factors discussed
above, but also manage any variability that will nevertheless continue to remain due to the inherent
nature of the business. We will discuss here some of the processes that leading CPG manufacturers
have implemented and expecting to see promising results.
• Collaboration with retail customers (Voice of customer): Promotions and other retail events generate
the greatest variations in demand. As consumer demand peaks, the out-of-stock rate nearly doubles.
Typically, once a year, the manufacturer and the retailer agree on an event calendar. However, if
there are coordination gaps between the two parties, as in failure to communicate a change in
the event timing, price points or ad positioning, then order volume and timing can go wrong,
causing swings in demand. Collaboration with customers on promotions and retail events largely
Voice of
caution
Voice of
Demand
Demand Voice of
customer reason
Variability
Variability
Voice of
consumer
helps in managing this variability. In this process, the manufacturer works with the retailer in
developing a collaboration strategy and a joint business plan for promotions. Then they work
together to determine the impact of planned events on consumer demand. As events occur,
promotional orders are placed, and delivery takes place. The manufacturer and retailer
collaboratively plan on the lift volumes, resulting in reduced out-of-stocks during promotions.
This in essence, results in reduced demand variability during and after a promotion.
There are several tools available in the market such as, JDA's Networks Collaborate and Oracle Demantra
that aid collaboration between manufacturers and retailers. These tools are part of their demand planning
suites and well integrated to provide a seamless demand planning solution.
• Integration of consumer demand into forecasting (Voice of consumer): Manufacturers constantly
strive to predict the changing preferences of consumers. However, their order or shipment forecasts
often lack visibility into the downstream demand driving them. Many companies have bridged
this gap by incorporating consumer demand into their forecasting process. There are several
variations to this process. Two such variations include:
> Using point-of-sale (POS) forecast to derive the order forecast in the short term for making
comparisons with the regular shipment or order forecast. Such comparisons help
identify exceptions and take necessary corrective actions, thus eliminating excess inventory
throughout the supply chain. POS inputs can be obtained from syndicated data sources such
as IRI, AC Nielson and Vision Chain. These inputs can then be run through a standard
DRP tool (such as JDA Fulfillment) configured with the manufacturer DC to retailer DC to store
supply chain model.
> Using advanced causal modeling techniques and processes, such as multi-tiered causal analysis
(MTCA) by SAS Institute Inc. This integrates POS, syndicated scanner data and other sell-in
data into the forecasting process to determine the effects of consumer demand on CPG
shipments. A causal model is applied to predict POS data, using all the significant causal
factors affecting consumer sales (such as retail price, media gross rating points (GRPs), in-
store merchandizing vehicles as well as competitive retail activities). A second causal model
is developed to forecast shipments using past POS data and the POS forecast as the main
explanatory factor. The model also takes into account the time lag between POS and shipments
along with other causal factors like forward buys and trade promotions.
Thus, CPG manufacturers can reduce demand latency by integrating the POS information into their
demand planning processes. This would result in reduced demand variability.
• Daily recalculation of operational forecast (Voice of reason): Many CPG manufacturers perform
forecasting at a weekly aggregated level, which is recalculated at a weekly frequency. However,
as in-store events are in-flight, there is a constant inflow of market information (such as customer
orders, shipments and POS data) containing crucial insights regarding what is selling in the
market and at what pace. Majority of the CPG manufacturers take more than two weeks to
sense demand and another two weeks to respond. Given short order lead times and high costs
of out-of-stocks, there is a lag in responding to changes in market activities. CPG manufacturers
are looking for ways to make sense of the in-flowing information and use it quickly to change
their near-term operational forecasts on a daily basis. Tools like Demand Sensing (Terra
Technology) and Dynamic Demand Response (JDA) have addressed this need for near-real
time forecasting. These tools perform a three-way matching of historical data, demand planning
forecasts and current market activities (such as order in-flows) to arrive at a daily forecast.
This forecast tends to be more accurate than the daily forecast from traditional planning systems
using rule-based consumption logic. Manufacturers can use this daily forecast for taking
decisions on inventory deployment and making changes to production plans. Thus, through
effective demand sensing and timely response, manufacturers can manage demand variability
without incurring unplanned costs.
• Calculation and optimization of safety stock using scientific principles (Voice of caution): Despite
having the best in-class demand planning processes, there will always remain some amount of
demand uncertainty. For the CPG manufacturer, there exists an average weekly forecast error of
45% at the DC level (measured as mean absolute percentage error of forecast with respect to
actual). Such residual uncertainty can be hedged against by having the right amount of safety
stock.
CPG manufacturers rely on safety stock to provide high service levels to customers. Service failure
costs are considerably higher than cost of excess inventory in the CPG industry. Therefore, there
is a tendency to err on the side of caution in carrying safety stocks. However, safety stock comes
at the cost of tying up the working capital and slowing down the inventory turns. The latter comes
with the risk of losing product freshness, an important consideration in the CPG industry.
Some mistakes that CPG manufacturers make while calculating their safety stocks are:
Tendency to perform Planners tend to look at safety stock needs at individual
local optimization echelons of the sourcing network, leading to overall sub-
optimization at the network level
Once safety stocks are calculated, no adjustments are made
Infrequent recalculation
unless there is an issue. Any changes to the demand profile
between reviews can result in excess inventory or shortages.
When there are any service failures, planners react by
Reactive approach
increasing the safety stock. This kind of reactive approach may
lead to increasing the network inventories, yet not completely
guaranteeing protection against future shortage risks.
In the absence of a system, planners tend to set safety stocks
Generalization
based on general SKU classification, such as ABC
classification. This approach has two associated risks. One is
that all SKUs within the same classification might end up
getting the same Safety Stock assigned. The other is that all
the SKUs within a given item network might be assigned the
same value, regardless of their position in the network.
One can avoid these mistakes by automating the safety stock calculation process using a multi-echelon
inventory optimization tool. Taking inputs in the form of historical forecast, historical demand, lead-
time, network structure, replenishment parameters, cost factors and desired customer service levels,
the tool can scientifically calculate safety stocks that are network optimized. Most of these tools have
the ability to perform what-if analysis allowing planners to build scenarios around the inputs. The planners
can use these scientifically calculated safety stocks in their planning or, if required, tweak them to apply
any additional business intelligence. Optiant Power Chain Suite, SmartOps, Toolsgroup's DPM, Terra
Technology's Inventory Optimization are some of the multi-echelon inventory optimization tools that are
available in the market.
A Last Word
By integrating the voices of the customer, consumer, reason and caution into
of the puzzle in the larger scheme of achieving the perfect order metrics.
Companies can seek to achieve the true benefit of increased demand visibility
only if it leverages this information effectively to develop a demand response
that delivers the perfect order in a profitable manner under all situations. The
company that is able to pull this off is most likely to win at the shelf in the
long term.
About the Author
Venky Ramesh is a Senior Consultant with the Supply Chain Management practice of Infosys. He
has over 8 years of experience in the CPG industry, wherein he has led several implementations of
supply chain planning packages and processes. He has experience working on tools like JDA Demand
Planning, JDA Fulfillment, JDA Dynamic Demand Response, Terra Technology Demand Sensing
and Optiant Inventory Optimization. Venky has a degree in Mechanical Engineering from Anna
University and an MBA from IIM, Ahmedabad.
Infosys Technologies Ltd. (NASDAQ: INFY) defines, designs and delivers IT-enabled business solutions that help Global
2000 companies win in a flat world. These solutions focus on providing strategic differentiation and operational superiority
to clients. Infosys creates these solutions for its clients by leveraging its domain and business expertise along with a
complete range of services.
With Infosys, clients are assured of a transparent business partner, world-class processes, speed of execution and the
power to stretch their IT budget by leveraging the Global Delivery Model that Infosys pioneered.
For more information, contact askus@infosys.com www.infosys.com
© 2009 Infosys Technologies Limited, Bangalore, India. Infosys believes the information in this publication is accurate as of its publication date; such information is subject to change without notice. Infosys acknowledges the proprietary
rights of the trademarks and product names of other companies mentioned in this document.
