Demand and order‐fulfillment planning: The impact of point‐of‐sale data, retailer orders and distribution center orders on forecast accuracy

• Published date: 01 July 2019
• Author: Funda Sahin,E. Powell Robinson,Arunachalam Narayanan
• DOI: http://doi.org/10.1002/joom.1026
• Date: 01 July 2019

TECHNICAL NOTE

Demand and order-fulfillment planning: The impact

of point-of-sale data, retailer orders and distribution

center orders on forecast accuracy

Arunachalam Narayanan | Funda Sahin | E. Powell Robinson

Department of Decision & Information

Sciences, C.T. Bauer College of Business,

University of Houston, Houston, Texas

Correspondence

Arunachalam Narayanan, Department of

Decision & Information Sciences,

C.T. Bauer College of Business, University

of Houston, 260E Melcher Hall, Houston,

TX 77204.

Email: anarayanan@bauer.uh.edu

Handling Editor: Suzanne de Treville

Abstract

Industry consultants claim that investing in systems that base forecasting on point-

of-sale (POS) data throughout the supply chain will improve forecast accuracy. We

explore what actually happens to forecast accuracy for demand and order-

fulfillment planning when the forecast demand signal is based on POS, retailer

orders, or distribution center (DC) orders. We begin by comparing the forecast

accuracy for different demand signals using daily demand and operating data from

a large consumer-products supply chain. We then extend the analysis by varying

the demand and supply-chain parameters to evaluate their impact on demand signal

performance. We find that POS data improve forecast accuracy for demand plan-

ning but not for order-fulfillment planning. These findings thus challenge

consulting-firm claims about the ability of POS-based forecasting systems to

improve forecast accuracy across all contexts.

KEYWORDS

forecasts, order-fulfillment planning, point of sale, retail

1|INTRODUCTION

System developers assert that forecast accuracy at distribution

centers (DCs) and suppliers improves when forecasts are based

on consumer point-of-sale (POS) data rather than on incoming

orders. JDA (2014), a major advanced-planning-system soft-

ware provider, reports that a majority of manufacturers con-

sider POS data sharing as central to improving supply-chain

planning and execution. The direct link between channel activi-

ties and consumer demand created when basing forecasting on

POS data has led some researchers to propose this approach as

integral to the demand-driven supply chain (Cooke, 2013). The

POS demand signal is argued to reflect the consumers' demand

pattern in near real time and to not be distorted by operational

factors such as order lot sizing and replenishment lead time

(Cooke, 2013; Grewal, Levy, & Kumar, 2009; Hammond,

1994; Lee, Padmanabhan, & Whang, 1997b and Zondag &

Ferrin, 2014). Enterprise application-integration software

extends use of the POS demand signal to external suppliers

(Ali, Babai, Boylan, & Syntetos, 2017; Ciancimino, Cannella,

Bruccoleri,&Framinan,2012andLebreton,Meyr,Rosi



c,

Seipl, & Wetterauer, 2015).

Given that retailers in the consumer packaged goods

(CPG) industry must combine high on-shelf availability with

strict inventory control (Fleischmann, Meyr, & Wagner,

2015), we expect to observe high usage of POS data for

forecasting if it results in improved forecast accuracy. How-

ever, this is not the case: Of the top performing 20% of

retailers, only 37% collaboratively share POS data with

channel partners, falling to 18% for the remaining 80% of

retailers (Ball, 2015). The Food Marketing Institute (FMI)

and the Grocery Manufacturers of America (GMA) report

Received: 20 September 2016 Revised: 13 February 2019 Accepted: 24 February 2019

DOI: 10.1002/joom.1026

468 © 2019 Association for Supply Chain Management, Inc. wileyonlinelibrary.com/journal/joom J Oper Manag. 2019;65:468–486.

that when POS data are shared, only 57% of suppliers use it

for forecasting because of differences in data formats and

the need to manage large amounts of data (FMI/GMA,

2014). Ali et al. (2017), Seifert (2003), Butner (2010),

Allred, Fawcett, Wallin, and Magnan (2011) and E2Open

(2013) report similar findings. Also, many companies lack

the ability to use POS data effectively when it is available to

them (Ball, 2015). Our discussions with several CPG execu-

tives revealed reluctance to implement POS data sharing:

Concerns cited stem in part from the costs involved in POS

data capture, cleansing, analyzing, implementing and

maintaining IT infrastructure, but also include uncertainty

about whether forecasting accuracy actually improves

enough to warrant the investment.

CPG supply chains have traditionally used incoming

orders from their immediate downstream channel members as

the forecasting demand signal. However, advanced planning

systems allow decision makers to choose between three

demand signals: retailer POS, retailer orders,

1

and DC orders.

Although other demand signals or combinations of demand

signals are possible, current industry practice applies a single

demand signal at a given facility (Lebreton et al., 2015).

Figure 1 illustrates the demand signals that are available at

each facility level in a retail supply chain.

We explore the impact on forecast accuracy when an

order-based demand signal is replaced by one based on POS

data. We consider forecasting for both demand and order-

fulfillment planning. The major characteristics of these two

planning activities are summarized in Table 1. As noted in the

table, a critical input for demand planning is the forecast of

consumer demand served by each facility, which we refer to

as the “demand-planning forecast.”Order-fulfillment planning

FIGURE 1 Alternative demand

signals in a CPG supply chain. The retailer

forecast is based on POS data. Information

sharing makes POS data available to the DC

and supplier and retailer orders available to

the supplier. Thus, the DC chooses either

retailer orders or POS data for the demand

signal. The supplier chooses from POS data,

retailer orders, or DC orders as its demand

signal. Demand signal type.

Consumer demand (POS data).

Retailer orders. DC orders

TABLE 1 Demand planning and order-fulfillment planning

comparison

Demand planning

Order-fulfillment

planning

Objective Match overall supply

with end-customer

demand at each

facility.

Ensure sufficient

inventory is on-hand to

fill incoming orders.

Planning

focus

Capacity planning,

production and

distribution

scheduling, aggregate

production planning,

warehouse staffing

and fleet planning to

meet expected

customer demand.

Establish inventory policy

parameters for safety

stock, reorder points,

and order up-to levels

to meet target in-stock

service levels.

Forecast

item

Consumer demand at

each facility

Incoming orders at each

facility

NARAYANAN ET AL.469