Forecasting and Demand Management

Forecasting
And
Demand Management
Strategy
And
Development
A concise writing on the actions needed to begin the
implementation of world class forecasting and demand
management program.

Copyright 2000 by ECRU Technologies, Inc.

Forecasting and Demand Management

Copyright 2000 by ECRU Technologies, Inc.

Forecasting and Demand Management

Forecasting
And
Demand Management
Strategy
And
Development
J. Martin
R. Roth

ECRU Technologies Publisher

A portion of the
KNOWLEDGE TRANSFER SERIES
from the WORLD OF LEARNING.
Copyright 2000 by ECRU Technologies, Inc.

Forecasting and Demand Management

Copyright 2000 by ECRU Technologies, Inc.

All rights reserved. No part of this publication may be reproduced, distributed, or transmitted
in any form or by any means, including photocopying, recording, or other electronic or
mechanical methods, without the prior written permission of the publisher, except in the case
of brief quotations embodied in critical reviews and certain other noncommercial uses
permitted by copyright law. For permission requests, write to the publisher, addressed
Attention: Permissions Coordinator, at the address below.

ECRU Technologies, Inc.

1475 Terminal Way, Suite E
Reno, NV 89502-3225
Ordering Information
Orders by individuals and organizations. ECRU Technologies, Inc. publications are available
through bookstores or can be ordered direct from the publisher at the ECRU Technologies,
Inc. address above or by email to 103700.676@compuserve.com.

Library of Congress Cataloging-in-Publication Data

Martin, J./Roth, R.
Forecasting and Demand Management Strategy and Development: A concise writing
on the actions needed to begin the implementation of world class forecasting and demand
management program / J. Martin / R. Roth. 1st ed.
ISBN 1-931186-02-2 (papercover)
1. Science
2. Organization 3. Theory
Title

4. Systems

5. Behavior

I.

First Edition
First Printing September 2000

The information included in this book is further amplified when used in

conjunction with other books from ECRU Technologies, Inc. Other titles are:
Supply Chain Management Direction Strategy and Supply Chain Management
Development Strategy.

Copyright 2000 by ECRU Technologies, Inc.

Forecasting and Demand Management

Contents

INTRODUCTION........................................................................................................... 9

WHAT DO WE TRY TO ACCOMPLISH?............................................................... 11

2.1
2.2
2.3
2.4
2.5
2.6
2.7

DETERMINE FORECAST-ABILITY OF END PRODUCTS ................................................ 12

INSTALL A FORECASTING TOOL AND PROVIDE TRAINING IN ITS USE ........................ 13
DEVELOP DEMAND MANAGEMENT PROCESS OUTLINE ............................................ 13
IDENTIFY NECESSARY SUPPORT REQUIREMENTS ..................................................... 13
DETERMINE KEY METRICS ...................................................................................... 14
DETERMINE TIME / RESOURCES TO IMPLEMENT THE PROCESS................................. 16
DOCUMENT DETAILS IN PROJECT DOCUMENT ...................................................... 16

PROCESS DEFINITION:............................................................................................ 18
THE ORGANIZATION VIEW: ................................................................................................. 18
THE FORECASTING SCOPE.................................................................................................... 20

FORECASTING SUPPORT REQUIREMENTS ...................................................... 21

4.1
4.2
4.3

IT PLATFORM ASSESSMENT ..................................................................................... 21

INFORMATION SYSTEMS AND SERVICES INTEGRATION ............................................ 22
BUSINESS LOGISTICS AND PROCESSES ..................................................................... 23

PROCESS OUTLINE................................................................................................... 25
5.1
UNDERSTANDING THE FORECASTING PROCESS:....................................................... 25
5.2
BACKGROUND: WHAT FORECASTING IS .............................................................. 25
5.2.1
Statistical Forecast.......................................................................................... 26
5.2.2
Demand Management of the Statistical Forecast ....................................... 26
5.2.3
The Forecasting Process................................................................................. 27
5.3
DEFINING WHAT TO FORECAST THE SKU....................................................... 28
5.4
DISCIPLINE AND RESPONSIBILITY ............................................................................ 30
5.5
METRICS MEASURES OF PERFORMANCE ............................................................... 31
5.5.1
Forecast Metrics ............................................................................................. 31
5.5.2
Performance Metrics....................................................................................... 32
5.6
OTHER IMPLICATIONS .............................................................................................. 32

FORECASTINGS ROLE IN CORPORATE FUNCTIONS ................................... 33

6.1
6.2
6.3
6.4
6.5
6.6
6.7
6.8

MARKET & PRODUCT VIEW..................................................................................... 33

STRATEGIC / CORPORATE VIEW ............................................................................... 35
TIME / SEQUENCE VIEW ........................................................................................... 36
TACTICAL / BUSINESS VIEW .................................................................................... 38
OPERATIONAL / PROCESS VIEW ............................................................................... 40
FORECASTING PROCEDURES VIEW ........................................................................... 40
ROLES, RESPONSIBILITIES, AND METRICS VIEW ...................................................... 42
SUMMARY ................................................................................................................ 43

ROLES & RESPONSIBILITIES DEVELOPMENT ................................................ 44

METRICS DEVELOPMENT PROCESS................................................................... 45

DEMAND MANAGEMENT DRIVERS..................................................................... 47

9.1
9.2

TERMINOLOGY: DRIVERS, ASSUMPTIONS, DEMAND MANAGEMENT ...................... 47

TYPES OF DRIVERS................................................................................................... 49

Copyright 2000 by ECRU Technologies, Inc.

Forecasting and Demand Management

9.2.1
External Drivers.............................................................................................. 49
9.2.2
Internal Drivers............................................................................................... 50
9.2.3
Sector Drivers ................................................................................................. 53
9.3
DRIVER IDENTIFICATION .......................................................................................... 55
9.4
DRIVER MANAGEMENT ............................................................................................ 57
9.5
APPLYING DEMAND MANAGEMENT RESULTS ......................................................... 62
10

THE FORECASTING & DEMAND MANAGEMENT CYCLE ......................... 64

10.1 DOWNLOAD AND CYCLE START............................................................................... 65

10.2 FORECAST REVIEW & DEMAND MANAGEMENT ANALYSIS ........................................ 66
10.3 CHANGE CONSOLIDATION, FORECAST ADOPTION, UPLOAD ....................................... 67

Copyright 2000 by ECRU Technologies, Inc.

Forecasting and Demand Management

The Authors

J. Martin is an international consultant providing services in

supply chain management, e-commerce, organization change
management, and ERP system project management. He has
worked with a variety of industries ranging from heavy equipment
manufacture, transportation, paper, automobile, healthcare, and
electronic components. He has spoken before many industry
groups which include AT&T and IBM His educational
background includes a B. S. in operations management, M. S. in
computer science and a Ph.D. in psychology. He can be reached
via email at 103700.676@compuserve.com.
R. Roth is founder and president of Systems Services International
Ltd., providing business management and technology integration
services since 1975. Clients are in many industries construction;
automotive; communications; space and defense; manufacturing;
governments; medical devices. R. Roth and his team have proven
their multi-cultural experience and sensitivity in Europe, the
Middle East, South East Asia, North East Asia, and North
America. R. Roth is a seasoned economist with extensive
experience in information technology implementation. He can be
reached via email at RolfRoth@compuserve.com

Copyright 2000 by ECRU Technologies, Inc.

Forecasting and Demand Management

Copyright 2000 by ECRU Technologies, Inc.

Forecasting and Demand Management

Introduction

Accurate forecasting can play a decisive role in an organizations

planning, budgeting and performance monitoring process. The
meanings of forecast" are:
(1) the statistical forecast derived from historic data,
(2) the demand-managed forecast incorporating specific
assumptions into the statistical forecast, and
(3) the process used to develop and control the two former.

Statistical formulae are used to derive the forecast model, baseline,

trend and seasonality of the units as well as predict the likely future
quantities of the units. Historic data that has been accumulated
under past conditions of influence factors (Drivers) such as market
development, promotions, or other company actions are used. If
the past conditions continue to prevail, and if no changes are made
to influence the future development, then the forecasted quantities
are likely to occur. If the actions described are being taken in the
described quantity/value and time and if no other influences will
take effect, then the statistical forecast plus the changes resulting
from the described assumptions [heuristics] will likely occur.

Copyright 2000 by ECRU Technologies, Inc.

Forecasting and Demand Management

In order to determine whether or not the assumed effect actually

takes place we must:
(1) measure the actual quantities/values observed (i.e. the
new historic data), (
(2) compare the actual data to the demand-managed forecast
data to determine the difference if any,
(3) analyze, in the event of a significant difference between
actual and forecast, the underlying assumptions to
determine whether they were correct.
At completion of the analysis, the resulting new assumptions (more
optimistic/pessimistic, different actions, etc.) must be applied to
the then current statistical forecast.

This process should be performed at least monthly, to achieve a

reasonable measure of actual performance early enough to
implement any corrective action. Companies may have thousands
of SKUs (Stock Keeping Units) subject to independent demand.
Clearly, forecasting all of them would require a tremendous effort.
Consequently, a selection process must be put in place to forecast
only meaningful data. Criteria for the selection could be any SKU
defined for forecasting purposes as any measurable unit that can be
extracted and/or composed from historic data.

Copyright 2000 by ECRU Technologies, Inc.

10

Forecasting and Demand Management

The forecast process will result in improved business forecasts

accuracy if, and only if, the organization implements the process in
a disciplined fashion. Assumptions must be noted in detail; their
correctness must be analyzed; corresponding changes must be
defined.

This necessitates two implementation tools:

(1) a description of the forecast process (from timing through
data extraction, analysis, note taking, to applying changes),
and
(2) assignment of responsibility for each step in the process.
Typically, companies develop roles and responsibilities for the
process, and incorporate them into the position descriptions of the
responsible functions.

What do we try to accomplish?

This book outlines, at a high level, the forecast and demand

management process. It is intended as a comprehensive
summary, and as a guideline for developing your own work-inprogress to be modified and amended as you progress with the
implementation.

Copyright 2000 by ECRU Technologies, Inc.

11

Forecasting and Demand Management

Note that the forms referenced in this book, and shown as

examples, easily can be created in electronic format. This will
enable company-specific modifications to be made, and the forms
can be used as a blueprint for the development of automated
recording and tracking facilities.

This book is not a primer on statistical forecasting, forecasting

methodologies, or related theories. Numerous publications are
available that cover these subjects.

2.1 Determine Forecast-ability of End Products

As a first cut it is necessary to determine which independent

demand (the end products as sold to customers) are in effect
forecast-able with applicable forecast model, trend, seasonality
etc., and determination of A, B, etc. classification. This provides
the base for focus on the most important products. Spending 80%
of your effort on the top 20% of your products produces
measurable results, quickly!

Copyright 2000 by ECRU Technologies, Inc.

12

Forecasting and Demand Management

2.2 Install a Forecasting Tool and provide Training in its use

Forecasting is best performed using automated tools. Ideally,

several forecasting applications available in the market, should be
tested and evaluated. After installation, the system can be used for
forecast processing (both initial statistical forecasting and
subsequent demand management), and as a training tool for
personnel.

2.3 Develop Demand Management Process Outline

The forecasting and demand management process starts with

processing the historic data, and ends with the preparation of
forecast data (including demand management input) for the next
period.

This book comprises the process outline. It can be used to develop

your own action plan.

2.4 Identify necessary Support Requirements

Requirements covering organization functions directly involved in

the forecast and demand management process (i.e. departments,
managers, etc.), as well as such requirements that need to be
addressed but are outside the scope of the topic, are identified.
Copyright 2000 by ECRU Technologies, Inc.

13

Forecasting and Demand Management

Directly involved functions and their roles are described in this

document. Others need to be identified and documented as part of
the development process. Note that the organization should
address these requirements reasonably urgently, as the quality and
timing of many forecasting and demand management activities are
directly influenced by them.

2.5 Determine Key Metrics

Metrics measure the difference between established targets

and actual performance against these targets. This necessitates that
both target and performance are quantifiable in the same unit of
measure. Example:

a) TARGET:

Improve Sales, month-over-month, by

10%

b) PERFORMANCE:

Improved sales by $ 100, $110, $ 121,

etc.

Does this imply the target has been met? No since the
performance unit of measure (dollars) is different from the target
unit of measure (percent). Under these conditions, seemingly
conclusive data is meaningless or worse, misleading. The fault in
the example could be corrected in either of two ways:

Copyright 2000 by ECRU Technologies, Inc.

14

Forecasting and Demand Management

a1) TARGET:

Improve Sales, month-over-month, by

10% BASED ON $ 1000 start

b1) PERFORMANCE:

Improved sales by $ 100, $110, $ 121,

etc.

OR
a2) TARGET:

Improve Sales, month-over-month, by

10%

b2) PERFORMANCE:

calculate based on following actual data:

Sales in

Month1

Month2

Month3

etc.

Sales$

1000

1100

1210

etc.

Sales%

100%

+10%

+10%

etc.

Consequently, a system of metrics must be developed if the

company is to have accurate and meaningful performance
measurement capability.

Ironically, forecasting is the business process that least lends itself

to performance measurements. The reason: forecasting is a
scientifically calculated prediction of future sales UNDER
CONDITION THAT the assumptions and resulting actions that
were effective in the past and are reflected in the data history, will
REMAIN UNCHANGED. Yet it is the purpose of the demand
management process, to lead to assumption corrections and
different action conclusions. Paradoxically, the forecasting and
demand management process is most successful if actual
performance continually exceeds forecast, either with positive
(such as for sales) or negative (such as for costs) trends.
Copyright 2000 by ECRU Technologies, Inc.

15

Forecasting and Demand Management

Initially, the success of the forecasting process, itself, usually is

measured by determining whether actual sales for the forecast
period were equal to the forecasted sales for that period (or within
a pre-determined percentage-band around the forecast, e.g. +/5%). In the long term, this inadequate measure should be replaced
by metrics aimed at business functions leading to, and resulting
from, the forecasting/demand management process.

2.6 Determine Time / Resources to implement the Process

Once the Action Plan is drafted, for planning, training and

implementation actions resulting or identified in this project phase,
a high-level estimate of required efforts and likely time frames can
be developed.

2.7 Document details in Project Document

Throughout your project, issues will be identified that are out of

scope of the topic but will require addressing in the short term.
SKU (Stock Keeping Unit) planning and related Information
Services requirements are a good example. All project-related
information should be documented in clearly identifiable separate
binders, such as:

Copyright 2000 by ECRU Technologies, Inc.

16

Forecasting and Demand Management

a) The Forecasting and Demand Management Process

Implementation Manual
b) The Forecasting System Implementation Manual
c) The Forecasting System User Manual
d) The Training Plan (which might be combined with process
scripts)
e) The Action Plan

Copyright 2000 by ECRU Technologies, Inc.

17

Forecasting and Demand Management

Process Definition:

Forecasting and corresponding performance monitoring are an

integral part of most organizations management functions. A
high-level indication of such functions can be represented as
follows:

3.1 The Organization View:

External Events / Trends
Demand Management
Channels, Relationships,
Promotions, Events

Related Topics
Product Planning
Groups & Families, Form,
Fit, Function, Cost

Concurrent
Design

Package
Implementation

Laboratory / Prototyping

Transition Management
Sales Force / Automation

Product Structure,
Configuratiog, Pricingn,
Variants, Std. Costing

Items / Kits / Assemblies

Continuous Updates
ALL Areas

Bills of Materials

Concurrent
Design

Bid / Contract Contracts Management

Routings/Ass'y Instruct.
/ Process Formul.
Constraint Planning

Customers
(& theirCustomers
Customers)

CRP - Capacity
Requirements
Planning

RCCP - Rough Cut

Capacity Planning

Joint Specifications &

Development

Performance Metrics
Performance Metrics

PRP - Project
Requirements
Planning

Order Fulfillment
Order Fulfillment

DRP - Distribution
Requirements
Planning

MRP - Materials
Requirements
Planning

Purchase Orders /
Contracts

(& their Customers)

Sales Orders / Entry &

Billing

FCP - Finite Capacity /

FFC Finite Forward
Scheduling

MPS - Master
Production
Scheduling
Concurrent
Design

Project Implementation

Inventory / Service
Policy

EDM - Engineering Data

Management
Design, Specifications,
Revisions

Documentation

Package Evaluation &

Blueprint for Change

Forecasting / Modeling
Product X, Quantity Y,
Delivery Z

Suppliers / Vendors
(&Suppliers
their Suppliers)
/ Vendors

Production Orders

Distributors / VMI

Replenishment /
Distribution Centers,
Vendor Managed
Inventory

(& their Suppliers)

Distribution / Outbound
Logistics
(VMI?)

Maintenance - preventive
and remedial

Inbound Logistics

QA & Staging
Labeling, Packaging

Parts / Subassemblies
Management

Process &
Production Management
& WIP Control

Claims and Refunds

Claims and Refunds
Cash Management

Supply Chain Mgmt

Credit Management
Finance - GL, AP, AR

Payroll / Human Resources

Budget, Planning,
Costing
Metrics &
Measurements

Asset Management

Budget - P/L &

Performane
Targets

Standards &
Metrics

Actuals / Cost
Accounting

Hours Accounting

Completion Reporting

Backflushing

Data Provision & Acquisition

Materials Issue
Damaged and Returned
Products

Damaged and Returned

Products
Product Inventory
Management

SFC - Shop Floor Control

Line, Cell, Job, Repetitive,
Process

Customer Service,
Technical
Services,
Customer
Service,
Maintenance
Technical Services,
Maintenance

Training & Help

Desk

EDI & e-commerce

Applications
Management

Hardware

Data Base
Administration

Network /
Security
Management

System Software

Legacy
Management

Utilities/Monitors

Rollout Mgmt

Y2K Review

Note that forecasting and related functions (such as order

fulfillment, customer satisfaction, etc.) ultimately relate to virtually
Copyright 2000 by ECRU Technologies, Inc.

18

Forecasting and Demand Management

all functions in the company. Accurate forecast data can drive

production and financial planning, logistics and distribution
management, service and parts management.

For the topic, the scope is limited to forecasting proper. This

includes the forecasting process, and the application of demand
drivers. It excludes any of the uses of forecast data, such as for
MRP (Materials Requirements Planning), production and
purchasing planning, etc. Forecasting can be considered a standalone function concerning methodology, training, and process
execution. However, without translating the forecast results into
usable manufacturing planning data, and without monitoring the
accuracy of periodic forecasts and making corresponding policy
adjustments, the efforts expended on forecasting would be wasted.

Forecasting proper includes the activities in the periodic cycle

from accepting historic demand data through statistical
forecasting and demand management to providing forecast and
demand-managed data to other systems and functions. This
process is graphically represented as the High-Level Forecast
Process Flow: Note that the following explanations and
descriptions are based on this representation of the forecast and
demand management cycle.

Copyright 2000 by ECRU Technologies, Inc.

19

Forecasting and Demand Management

3.2

The Forecasting Scope

Set-up and Initialization
Process (one-time)

Create Operating Environment for Forecasting

Tools, on the Target Platform

Cyclical Forecasting
Process (monthly)

Perform initial Data Setup and Forecast

Calculation

Start-Up Data
Systems and Applications
that store / provide data
Historic Data
Extractions
(monthly data)

Applic. 1
X
Applic. 2
Y
Applic. n
Z

Flat
File X
1 of n

"History" may be a single file, or

menu selection of multiple files

Flat
File Y
2 of n

History File
Common Format suitable for
Forecast Processing

Flat
File Z
3 of n
Periodic: Actual
Data Reports

Historic Data,
Definitions,
Calendars
extracted from
Applications Files

Multiple
Regression
Analysis
(occasional)
Driver 1,
Location 1

"Driver" data used

to manage the
forecast process

Driver X,
Location 2

Forecasting
SYSTEM (1)

Cycle as needed
Statistical
Statistical
Forecast
Forecast
(Snapshot
- IF
(Snapshot
- IF
nothing
is done
nothing
then...)is done
then...)

Apply "Driver
Updates" to the
forecast data =
Demand
Management

Keep Record of
Keep Record of
Assumptions and
Assumptions and
Bases
Bases - update as
needed

Driver Z,
Location n

Use of forecasted data in other systems

and applications

Sales Input

Forecasting
SYSTEM (2)

Operations
Input

Managed
Managed
Forecast
Forecast
(Snapshot
-IF
(Snapshot
-IF
Demand
Mgmt
Demand
Mgmt
applied,
then
..)
applied, then ..)

Flat
File X
1 of n
Flat
File Y
2 of n

Forecast
Output Format
for MRP,
Reporting etc.

Flat
File Z
3 of n

MONITOR for
Delta (FC/Actual)

Generate "Flat
Files" with Forecast
Data for source
systems and
applications

Forecast Scope

High-Level Forecast Process Flow - HL-FC-Flow.vsd - 03/06Jan00 - RR

Copyright 2000 by ECRU Technologies, Inc.

20

Forecasting and Demand Management

Forecasting Support Requirements

Forecasting and Demand Management are business functions

which require extensive cooperation between functions and tasks.
The forecast frequency itself (which depends, in turn, on the nature
of the business) usually is monthly. However, all related activities
must be performed and completed within a very short time frame.
If the forecast is not completed within at most two or three days
after month-end closing, the forecast intelligence could become
useless as not enough time would remain in the period for sales
and operations to take forecast-inspired actions.

To operate in this compressed time fence, three specific business

areas must be addressed.
(1) IT Platform Assessment
(2) Information Systems and Services Integration
(3) Business Logistics and Processes

4.1 IT Platform Assessment

Most of the forecast-related data can be collected from existing

systems, supplied from front-end input prepared by sales persons
and other functional heads, and distributed via existing information
systems. Therefore, the information technology platform should
be analyzed to determine its suitability as a support platform.

Copyright 2000 by ECRU Technologies, Inc.

21

Forecasting and Demand Management

Specifically, this involves:

Inventory of and Access Mechanisms to Hosts, Servers,
Networks, PCs
Capabilities and Use of Extranet, Intranet, Internet, Virtual
Private Networks, etc.
Host and Server Interfaces capabilities and functions, for
data feeds and use
Interim Data Structures (data warehouses, data marts, etc.)
to facilitate focused data management
Development of Platform Architecture document with
Training and Development Plan

4.2 Information Systems and Services Integration

Forecasting is a tool consisting of limited and specific-purpose

software applications that perform only one set of functions:
assessing historic information, classifying items, determining their
forecast-ability, and forecasting anticipated future quantities (of
units, dollars, etc.) based on supplied history.

Specifically, the forecasting tool set does not address issues such
as how the SKU (the item or part) is defined, how the data is
massaged during extraction from the host data bases, how it is
summarized, and how transitions from old to new SKUs will
be accomplished.
Copyright 2000 by ECRU Technologies, Inc.

22

Forecasting and Demand Management

For all these capabilities, the organization relies on its Information

Services department or provider. Specifically, this involves:
Design and implementation of pre-processing capabilities
Design and implementation of post-processing capabilities
Split-and-Bypass-and-Merge (Windowing) requirements
design and implementation
Cutoff, Transfer and Balance Management between host
applications and forecasting
Generational File Maintenance for Simulation, Restart,
Backup and Restore Purposes
Support and Analytical Software such as Multiple
Regression Analysis, automated Data Feeds (Duns, Dodge,
etc.)
On-line Documentation Support Capabilities for instant
access to assumptions, action decisions, etc.
Development of IS Support Structure Architecture document
with Training and Development Plan

4.3 Business Logistics and Processes

Forecasting has a very intimate relationship with other business

functions. For example, the independent demand quantity forecast
for any one product and period, will depend not only on historic
data but also on the companys decisions such as customer service
Copyright 2000 by ECRU Technologies, Inc.

23

Forecasting and Demand Management

levels, safety stock requirements, distribution points and supply

logistics, etc.

To link these functions to the forecasting and demand management

process, it is necessary to define their contributions to the process,
through business logistics mapping. Typically, this involves:
Development of Business Functions Matrix with focus on
FC/DM
Organization Skills Assessment (Forecasting, Statistics,
Data Analysis etc.) and Training Development
Roles and Responsibilities Development and Position
Description Updates to reflect contributions
Updating of Compensation / Incentives programs to reflect
weight of FC/DM and related functions
External Interfaces Considerations such as reporting (SEC),
compliance (documentation), etc.
Automating Integration of FC/DM into the Business Flow
Development of to-be business process flows with
Training and Development Plan

Without these support processes firmly developed and in place, it

will be difficult, though not impossible, to maintain the discipline
of preparing monthly forecasts, reliably, with reproducible analysis
and assumption options.

Copyright 2000 by ECRU Technologies, Inc.

24

Forecasting and Demand Management

These support functions are excluded from scope of this book they should be included, at a high level, in the Action Plan.

Process Outline

5.1 Understanding the Forecasting Process:

The term forecasting often is mis-used as a set of statistical

formulae, as a sales predictions process, or as an educated guess
as to what might happen. Properly used, forecasting can play a
decisive role in the organizations planning, budgeting and
performance monitoring process. Therefore, an outline of what
forecasting really is, is required.

5.2 Background:

What forecasting is

For the purpose of this book, we distinguish three specific

meanings of forecast. They are:
(1) the statistical forecast derived from historic data,
(2) the demand-managed forecast incorporating specific
assumptions into the statistical forecast, and
(3) the process used to develop and control the two former.

Copyright 2000 by ECRU Technologies, Inc.

25

Forecasting and Demand Management

5.2.1 Statistical Forecast

Statistical formulae are used to derive, from quantitative history of

independent demand for units, the forecast model, baseline,
trend and seasonality of the units, as well as a prediction of the
likely future quantities of the units. Historic data has accumulated
under past conditions of influence factors (or Drivers) such as
market development, promotions, or other company actions.
Consequently, the statistical forecast derived from this data, has
a narrow meaning:

IF the past conditions continue to prevail, AND IF no changes

are made to influence the future development, THEN the forecast
quantities are likely to occur.

5.2.2 Demand Management of the Statistical Forecast

To make the forecast more meaningful as a predictor of the

future, decisions must be made about the factors that influence
demand, the strength of these factors, and their likely impact on
future demand. For example: a targeted promotion estimated to
cost $ X for a specific market, will result in a Y% one-time sales
increase in that market, for a period of four months, commencing
one month after the start of the promotion. In practical terms: the
forecast quantity/value for the future periods 2, 3 and 4 would have

Copyright 2000 by ECRU Technologies, Inc.

26

Forecasting and Demand Management

to be increased by Y%. The resulting demand-managed forecast

has the following meaning:

IF the actions described are being taken, in the described

quantity/value and time, AND IF no other influences will take
effect, THEN the statistical forecast plus the changes resulting
from the described assumptions, likely will occur.

5.2.3 The Forecasting Process

In order to determine whether or not the assumed effect actually

takes place, the following steps are required, at least at the end of
period 4 (in the above example):

a) Measuring the actual quantities/values observed (i.e. the new

historic data)

b) Comparing the actual data to the demand-managed forecast data

to determine the difference if any

c) In the event of a significant difference between actual and

forecast, analyzing the underlying assumptions to determine
whether they were correct. (Note that several assessments
must be made to determine, inter alia, whether there may have
been a systemic error in the assumption, whether the assumed

Copyright 2000 by ECRU Technologies, Inc.

27

Forecasting and Demand Management

effects where weaker, stronger or in a different time frame than

expected, etc.)

d) At completion of the analysis, the conclusions drawn must be

noted (with the original assumptions), changes about the
assumptions must be noted, and the resulting new assumptions
(more optimistic/pessimistic, different actions, etc.) must be
applied to the then current statistical forecast.

This process should be performed at least monthly, to achieve a

reasonable measure of actual performance early enough to
implement corrective action if necessary.

Note that the repetitive period refinement of assumptions forces

the organization to continually improve its analysis and
conclusions process, thereby increasing its knowledge of
customers, markets, the competition and other influence factors
(drivers).

5.3 Defining what to forecast the SKU

In many instances, companies have thousands of SKUs (Stock

Keeping Units) subject to independent demand (by customers in
the market). Clearly, forecasting all of them would require a
tremendous effort. In most cases, the result would not be

Copyright 2000 by ECRU Technologies, Inc.

28

Forecasting and Demand Management

significantly different than when only 25% (or some other

measure) of the SKUs would have been forecast. Consequently, a
selection process must be put in place to forecast only
meaningful data. Criteria for the selection could be:

a) SKUs that have the greatest effect on results. Typically,

about 10% of all SKUs account for more than 35% of sales,
with another 15% accounting for an additional 25% of sales.
Forecasting these 25% of all SKUs (often described as A and B
items) will result in a meaningful forecast of 60% of sales.

b) SKUs that allow the best possible assessment of influence

factors. In many instances, relatively few large customers
account for a very large proportion of all sales. Therefore,
seeking to influence demand by these few customers will have a
large effect on anticipated sales. Similarly, a few product
models out of thousands of end items, components and parts,
are much easier to control and forecast, than the total
population of thousands of SKUs.

Note that the definition of SKU changes with the desired degree
of control. Forecasting software will allow determination of A, B,
etc. status, forecast-ability, forecast model, for any measurable
unit. Therefore, the desired SKUs must be carefully designed and
corresponding data must be extracted from the historic data files.
Consequently, a stock keeping unit for forecast purposes is vastly
different than an SKU for production and inventory purposes:
Copyright 2000 by ECRU Technologies, Inc.

29

Forecasting and Demand Management

An SKU defined for forecasting purposes, can be any measurable

unit that can be extracted and/or composed from historic data.

Note that forecasts for SKUs that are too narrowly defined, can
become meaningless because the number of items becomes to
small, for meaningful statistical analysis.

To enable tracking of actual performance against forecast for these

constructed SKUs, the same cycle of assumptions,
measurements, corrections (the demand management) must be
applied as for real SKUs.

5.4 Discipline and Responsibility

The forecast process as described, will result in improved business

forecasts, IF and only if the organization implements the process in
a disciplined fashion. Assumptions must be noted in detail; their
correctness must be analyzed; corresponding changes must be
defined.

This necessitates two implementation tools: (1) a description of

the forecast process (from timing through data extraction, analysis,
note taking, to applying changes), and (2) assignment of
responsibility for each step in the process.

Copyright 2000 by ECRU Technologies, Inc.

30

Forecasting and Demand Management

Typically, companies develop roles and responsibilities for the

process, and incorporate them into the position descriptions of the
responsible functions.

5.5 Metrics Measures of Performance

Determining responsibility is not, by itself, sufficient to ensure the

discipline of the process. To do so, a set of performance
measurements are required. They fall into several basic categories:

5.5.1 Forecast Metrics

The quality of the forecast process itself can be measured by

constantly monitoring how close the observed actual data track
to the forecast data. The better the process, the smaller the
difference between them. Note that the forecast merely reflects the
organizations past history and its best judgment on how to
influence future demand. In and by itself, the forecast cannot be
right or wrong (unless underlying statistical formulae are
incorrectly applied).

If external influence factors are used to modify the statistical

forecast, additional statistical evaluations will be required to ensure
that these factors are correctly applied. For example, to determine
the effect of an interest rate rise on an interest-sensitive

Copyright 2000 by ECRU Technologies, Inc.

31

Forecasting and Demand Management

organization, a multiple regression analysis might be necessary to

determine the degree and the time of the influence.

5.5.2 Performance Metrics

In many instances, forecasting can be accurate as measured by

forecast metrics, yet not reflect severe problems. For example,
actual total shipment volume can be exactly as forecast, while
many of the shipments could have been very late. In this instance,
customer dissatisfaction eventually would result in sales reductions
event though other sales and marketing assumptions might be
adequate.

Consequently, regular business metrics must be employed as part

of the forecasting process. Such metrics can be operationally
relevant (late shipments, partial shipments, excess damage, out-ofstock situations etc.), reflective of market changes (new
competition, un-anticipated seasonal influences, etc.)

5.6 Other Implications

It is both probably and likely, that the forecast process as a

methodology, a discipline, and a set of usable results might
influence corporate decision-making more than any other
application of business knowledge. Customer Satisfaction the
ultimate measurement of success will be as good as the

Copyright 2000 by ECRU Technologies, Inc.

32

Forecasting and Demand Management

organizations capability to anticipate the customers requirements

and expectations.

Forecasting is the process used to define and quantify these

expectations.
6

Forecastings Role in Corporate Functions

A top-down view of the corporation reveals a progressively

more detailed focus from the perception of the market, to the
metrics to be tracked for measuring performance in that market.
These views can be represented as follows:

6.1 Market & Product View

Copyright 2000 by ECRU Technologies, Inc.

33

Forecasting and Demand Management

The first view indicates how the companys products/services must

compete in the market. Typically, production and distribution
Product "Make"

Product "Value"

Make to Stock

Suppliers

Hours / Days

MtS / POS Repl.

Make-to-Stock
Assemble-to-Order
Days/Months

cents / dollars

Order
Delivery &
Replace't

Forecast Horizon

MtS / AtO

'00s / '000s

Forecast Horizon

EtO

9 - 60+ months

Markets

continuous
replenishment

FCH

EtO / DBO

Competition

Customers

Life Cycle

one-of-a-kind

'00,000s,millions/billions

Compliance

Regulation

Production and Distribution Characteristics

Design Drivers

Industry Drivers

Market Drivers

Manufacturing & Materials Lead Time

Long Lead Time
Short Delivery Time
Reliable Fabrication / Assembly Processes
Poor Process Capability
High Reliability
Continuous Asset / Tools Utilization
High Set-up / Tooling Costs
Immediate Consumption
Regular Demand
Irregular Demand
High Throughput
"Replacement Lifetime " Use
"Forever" Use
1000's Suppliers, Millions Customers
Market-specific
Customer-specific
Low Inventory Levels
One Customer - One Contractor Thousands Customers - Dozens Suppliers
"State" Risk Reduction

Forecast & Demand Management

- Market & Product View

Models(2).vsd - 18Feb98 - RR

characteristics as well as forecast time fences are a direct result

of the market environment.

Note that the forecast horizon the time available to forecast

likely demand, and to supply that demand, becomes ever more
Copyright 2000 by ECRU Technologies, Inc.

34

Forecasting and Demand Management

short as the corporation moves from an engineered-to-order

environment, to a make-to-stock operation.

6.2 Strategic / Corporate View

At the next level, the corporation identifies the business planning,

and sales and operations planning required to successfully operate
in the market.

The Tactical / Business View then defines how customers and

business partners become the focus of tactical planning, while the
Operational / Task View defines the production and operational
requirements.

Copyright 2000 by ECRU Technologies, Inc.

35

Forecasting and Demand Management

Strategic / Corporate View

FC & DM

Business Planning
Sales and Operations Planning (S&OP)

Tactical / Business View

Customer Relationship Management (CRM)
Supply Chain Management (SCM)

Operational / Task View

ERP

Master Scheduling
Materials Requirements Planning (MRP / II)
Manufacturing Execution (MES)
Logistics and Service

Forecast & Demand Management

- Strategic / Corporate View

Models(3).vsd - 18Feb98 - RR

Forecasting and demand management apply the time fences

around these views.

6.3 Time / Sequence View

Copyright 2000 by ECRU Technologies, Inc.

36

Forecasting and Demand Management

The forecast horizon is long-to-medium-term attempting to

predict the likely independent demand for the companys products
(services) based on business, and on sales and operations planning
(i.e. based on the Corporate View!). Note that this horizon shrinks
the closer the company comes to make-to-stock operations.

Projection

Strategic / Corporate View

Sales and Operations Planning (S&OP)

Forecast "Horizon"

FC & DM

Business Planning

long term

Tactical / Business View

Operational / Task View

ERP

Orders

near term
tomorrow

Supply Chain Management (SCM)

TIME HORIZON

Customer Relationship Management (CRM)

detailed

Master Scheduling

today
history

Logistics and Service

Shipped

Manufacturing Execution (MES)

WIP

Materials Requirements Planning (MRP / II)

Forecasting & Demand Management

- Time/Sequence View
Models(4).vsd - 18Feb98 - RR

The accuracy of the forecast especially in a short-term delivery

environment cannot be proved using actual orders. Ideally,

Copyright 2000 by ECRU Technologies, Inc.

37

Forecasting and Demand Management

orders booked should be equal to the independent demand that

has been forecast.

In reality, the periodic forecast is a snapshot that predicts what

will happen, based on past history, IF the company does not
change its underlying policies and procedures which are reflected
in that history.

Using the forecast result as an indicator, the company needs to

determine which factors or drivers it needs to change, in order to
improve its likely result, compared to the initial forecast.

These influence factors are identified and applied, as part of the

tactical / business operations.

6.4 Tactical / Business View

The forecasting process together with customer relationship

management commences the periodic (usually monthly) tactical
review process. It aims at quantifying the desired results in
dollars and high-level product units, and determines which
contributions are necessary to achieve these results.

Copyright 2000 by ECRU Technologies, Inc.

38

Forecasting and Demand Management

MIS - Platform and Applications Support

Production
Management

Marketing &
Sales

Sales Force Commitment &

Participation

Senior /
Finance Mgmt

Customers
Products

Forecasting Process
Oriented towards Improving
s the Bottom Line
i.e. Product Dollars and Quantities

Markets
Competitors
Governments

DPAS

Seasons

CRM and Forecasting Systems

Overlapping Interests / Functions possible and likely
Bills of
Materials
Routings
Contracts

MRP II /
RCCP etc.
MIS / Nets
DBMS etc.

Engineering and Drawing Systems - Production Planning

Overlapping Drivers possible and likely

Production and "Back-Office", Logistics Systems

Capacity
Planning / Scheduling Process
Oriented towards Operational Efficiencies
i.e. Production Costs and Units

Equipment
Labor
Raw Materials
Governments

Manufacturing
Management

Purchasing /
Logistics

Distribution /
Logistics

Maintenance
& Service

Seasons

Forecasting & Demand Management

- Tactical / Business View

Models(6).vsd - 18Feb98 - RR

This view makes it obvious that forecasting alone cannot deliver

results. Forecasting is customer and market oriented. Production,
back-office and logistics are parts oriented.

The gap between the two spheres must be bridged by a functioning

engineering and data management system. It allows rapid
Copyright 2000 by ECRU Technologies, Inc.

39

Forecasting and Demand Management

translation of high-level unit forecast, into detailed parts

requirements.

6.5 Operational / Process View

The periodic nature of forecasting and the short time available to

satisfy the predicted independent demand necessitate that the
forecasting and demand management process can be performed
speedily, with as little effort as possible. The process needed to
meet these requirements is represented below.

Note that the process is bound between extraction and provision

of historic data (last periods), and the uploading of forecast data
for MRP and reporting purposes.

6.6 Forecasting Procedures View

At the most detailed level, the procedures view (relative to

forecasting and demand management) identifies the steps required
to implement the process, and the tasks that must be performed
regularly and repetitively.

Copyright 2000 by ECRU Technologies, Inc.

40

Forecasting and Demand Management

Initial Definitions & Periodic Maintenance

Platform

MIS

Business
Logistics

Startup

Forecast & Demand Management - Cycle Start

Business &
Finance

Marketing
&
Economics

Sales &
Customer
Relations

Operations
& Logistics

Consolidation - Assumptions, Contributions

Demand Management & Forecast Evaluation
Forecast Acceptance & Targets Update
Forecasting & Demand Management: Cycle End
Forecasting & Demand Management

- Task / Procedures View

Models(5).vsd - 18Feb98 - RR

The start of the forecasting and demand management cycle

produced the statistical forecast the calculated amount of
independent demand that would result from historic sales data
provided the company would not change underlying policies and
procedures.

Various business functions must assess these quantities and their

timing, and determine which factors likely would change the
quantities. These factors and their specific values are called
drivers. They exist in business and finance management
Copyright 2000 by ECRU Technologies, Inc.

41

Forecasting and Demand Management

(interest rate sensitivity), marketing and economics (housing starts,

restrictive regulations, whether), sales and customer relations
(promotions, sales events, floor space support), operations and
logistics (distribution locations and distances, expediting costs,
materials lead times), and in any other corporate function.

Successful and reliable forecasting can be achieved IF these

drivers are vigorously tracked, analyzed, and used to shape
corporate policy and practice (service levels, inventory, etc.).
The ability to measure success, depends on two corporate
objectives: (1) setting targets for each function that identify and
quantify what is required of the function in order to meet the
companys targets, and (2) establishing data collection, calculation
and reporting capabilities that allow evaluation of the functions
actual performance against their respective targets. Usually, the
two objectives are addressed through the development of roles
and responsibilities, and of metrics. This is the most detailed
and common view of corporate operations:

6.7 Roles, Responsibilities, and Metrics View

Copyright 2000 by ECRU Technologies, Inc.

42

Forecasting and Demand Management

From Corporate Objectives

through Roles and Responsibilities
Development, to Accountability and
Performance Measurements
in performing Business Processes
Corporate Performance

President

Objectives

Defines

Negotiate
Agree

Objectives

Defines

Negotiate
Agree

Customers

Team
Leader

Expectations

Negotiate
Agree

Team
Member

My Team

Objectives

Fulfillment

Defines

Derives

Consistent &
Dependable
Linkage

Tasks

Derives

CSF - Critical
Derives
Success factors

Accountability
Development and
Tasks
Performance
Measurements

Suggest, Direct,
Guide

Suggest, Direct,
Guide

Objectives

Roles &
Responsibilities
Development

Tasks

Suggest, Direct,
Guide

Suggest, Direct,
Guide

Link

Defines

CSF - Critical
Success factors

Derives

Link

Defines

Derives

Suggest, Direct,
Guide

Suggest, Direct,
Guide

Link

Vice
President

CSF - Critical
Success factors

Derives

CSF - Critical
Success factors

Require

Require

Objective 1

CSF 1.1

Derives

Event

Require

Tasks
Consistent &
Dependable
Linkage

Task 1.1.1
Task 1.1.2

Task 1.1.1
Business
Processes

Task 1.1.3

Task n.1. 2

Objective n

Position
Descriptions and
Performance
Targets

CSF n.1
Task n.1.1
Task n.1.2

Etc.
Objective On

Close

Other Teams'
Other Teams'
Objectives,
Objectives,
CSFs,
Tasks
CSFs, Tasks

CSF On.1
Task On.1.1
TaskOn.1.3

6.8 Summary

Forecasting is a tool. It calculates likely outcomes based on

performance history. Business functions must determine how to
sustain or change the predicted outcomes, through active
management of the drivers. Setting specific performance targets,
and measuring each functions actual performance against target,
enables success control and remedial action.

Copyright 2000 by ECRU Technologies, Inc.

43

Forecasting and Demand Management

Roles & Responsibilities Development

Forecasting is a tool to calculate likely outcomes, based on past

performance, under the assumption that "nothing changes" that
all policies, procedures, market environments, customer relations,
etc. remain static. In reality, this is highly unlikely. Consequently,
each business function must be assigned specific targets. The
targets, when met, would enable the company to achieve its
objectives for results.

The development of roles and responsibilities is a four-step

process.

(1) For each corporate level, establish quantifiable objectives

(2) For each objective, establish its critical success factors
(CSF)
(3) The CSF of the higher level, become the objectives of the
respective lower level
(4) Operational tasks (procedures) are developed to ensure that
CSFs can be achieved.

The development process is out of scope for the purposes of this

topic.

Copyright 2000 by ECRU Technologies, Inc.

44

Forecasting and Demand Management

Metrics Development Process

Performance Measurement is at the core of the organizations

determination whether or not its plans have met with success. The
assessment process involves four steps:

1. Setting Performance Targets for each business function

in the organization
2. Collecting data about actual performance, for all business
functions
3. Calculating actual performance based on collected data,
versus targets
4. Acknowledging success / non-success of target
achievement

The process of collection data for performance measurement

usually is referred to as metrics. Metrics measure the
difference between established targets and actual performance
against these targets. This necessitates that both target and
performance are quantifiable in the same unit of measure.

Initially, only the accuracy of the forecast is established as a

metric. It indicates that this specific metric (were actual sales
within x % of the forecast for the period?) is subject to company

Copyright 2000 by ECRU Technologies, Inc.

45

Forecasting and Demand Management

action on the preceding periods forecast and, therefore, very

subjective and prone to change.

To develop meaningful metrics for the organization measures that

allow assessment of performance even in the short term requires
a development and implementation program to be put in place.

Development of the metrics and performance measurement process

is out of scope of this topic.

Copyright 2000 by ECRU Technologies, Inc.

46

Forecasting and Demand Management

Demand Management Drivers

Statistical forecasts are a calculation of expected independent

demand, based on historic performance (sales). Policies and
procedures that were in effect in the past, are reflected in the
historic data.

9.1 Terminology: Drivers, Assumptions, Demand Management

For example, an earlier decision to heavily promote a specific

product, could be reflected in increased monthly sales quantities of
the product, since the promotion. Alternatively, the initial increase
could have been followed by a decline to traditional sales levels.
In the first instance, the promotion would have led to a permanent
gain, while in the second instance, the data would indicate the
specific promotion had a time-limited effect.

The statistical forecast would project a higher trend in the first

case, and a seasonality spike in the second. Whether or not this
projection would be accurate (better: whether actual sales would
by equal or similar to projected sales), depends entirely on the
analysis of the data, and on the conclusions the company draws
from the analysis.

Copyright 2000 by ECRU Technologies, Inc.

47

Forecasting and Demand Management

The process of analysis and conclusion can be described as

demand management. Technically, the company attempts to
influence (manage) demand by creating favorable conditions.
These conditions are aimed at specific factors, the drivers.
Conclusions drawn from prior factor analysis and assessment were
assumptions at the time the conclusions were drawn. Since
then, actual sales history indicates whether or not these
assumptions were correct.

Any relevant new conclusions must be recorded, as well as specific

action decisions taken to better influence the factor(s). To achieve
measurable results, actions and expected results must be quantified
in terms of effort and time. For example: The most recent
product promotion has resulted in sustained sales increases, where
for each dollar spent, we have seen an increase of 150 dollars.
Analysis indicates that we need to increase, permanently, our
promotion budget for this product by $ x (the budget target) in
order to sustain improving sales every month by y % (the result
target), recognizing that the improvement effect will be noticeable
approximately 2 months after the initial promotion (the time
target).

Copyright 2000 by ECRU Technologies, Inc.

48

Forecasting and Demand Management

The resulting actions would be reflected in these ways:

(1) the budget would show a promotion cost increase of $ x

every month,
(2) the sales forecast would be increased by y% every
month, but
(3) with a time delay of two months between the cost
increase and the sales increase

9.2 Types of Drivers

Specific business functions need to be responsible for identifying

and quantifying drivers on a regular basis. Which functions roles
and responsibilities will be so affected, depends on how the
drivers are classified, based on the companys past experience.
Some classification options are:

9.2.1 External Drivers

These are drivers that are beyond the companys control; they are
encountered in the many areas such as (with examples):
customer (vacation shutdown, short order lead time)
market (development permits, housing starts)
Copyright 2000 by ECRU Technologies, Inc.

49

Forecasting and Demand Management

competition (aggressive promotions / pricing for brand

builders
government action (mortgage rate deductibility)
fiscal / monetary policy (interest rate change)
licensing requirements (gas installation inspections)
engineering standards etc. (new UL requirements)
whether (seasonal influences, snow predictable,
unpredictable)
etc.
Some of these drivers occur infrequently (such as interest rate
reductions / increases, changes in engineering/UL specifications,
etc. ), while others occur quite often (such as snow falls could
delay construction starts, our customer shuts down for vacation
could drop some demand or simply postpone it, etc.)

9.2.2 Internal Drivers

These are those that influence our demand because of actions we

take, internally. They can be found in many areas such as (with
negative examples):
Accounting (credit approvals take too long)
Distribution (too little stock in our distribution
warehouses)
Copyright 2000 by ECRU Technologies, Inc.

50

Forecasting and Demand Management

Customer Service / Administration (installer calls kept on

hold, slow inquiry response)
Engineering (design not friendly for install ability,
insufficient specs, etc.)
Facilities (shipping dock too small, inadequate handling
equipment
Finance / Treasury (pre-determined price curve / price
points not flexible enough)
Forecasting (do not consistently analyze last periods
assumptions)
Human Resources (slow data entry causes backup of
orders)
Information Systems and Technology (no quick order
status inquiry)
Inventories (commodities safety stock too small for
ensured acceptable cycle time)
Logistics (competition for LTL space in regions increases
shipping costs)
Maintenance (equipment/tools breakdowns lengthen
average cycle time)
Marketing (we dont promote to brand builders for
profitability)
Materials (steel too thin excessive freight damage)
Management (too many special pricing requests real
exceptions take too long)

Copyright 2000 by ECRU Technologies, Inc.

51

Forecasting and Demand Management

Manufacturing (capacity constraints, scheduling, labor

availability etc.)
Overhead (too many salaried heads, expensive but
ineffective controls on quality etc.)
Systems / Processes / Procedures (retroactive price break
application too time consuming)
Production / Shop Floor (scrap and efficiency not
controlled)
Projects (effects of special projects subdivision
development promotion? not reflected in production
scheduling / materials purchases)
Purchasing (long lead times, pricing, quality assurance on
receiving)
Quality (high warranty costs)
Sales (incentives promote volume regardless of
profitability)

Some of these drivers require quick action from period to period

(such as responding to competitive promotions), others may
necessitate internal improvements in order to maintain or expand
demand (customer service / installation support, credit policies and
approvals etc.); still others may require investments to improve
profitability (inventory levels for immediate availability allow
profitable pricing).

Copyright 2000 by ECRU Technologies, Inc.

52

Forecasting and Demand Management

9.2.3 Sector Drivers

Other classifications can be made by sector or function. Some

examples follow.

9.2.3.1 Business and Finance Drivers

Existing Markets
New Markets
New Customers
New Product Potential
Interest Rates
Capacity and Inventory Financing Constraints
Production and Distribution Facilities Planning
Service Levels
Product Replacements / Phase outs
9.2.3.2 Economic and Market Drivers
Political Action
Fiscal / Monetary Policies
Economic Conditions
Competitive Environments / Imports
Distribution Structures
Market Outlooks
Industry Trends
Copyright 2000 by ECRU Technologies, Inc.

53

Forecasting and Demand Management

Construction Starts
Whether
9.2.3.3 Customer Relationship and Sales Drivers
Existing Customers / Sales Profiles
Existing Sales Channels, Markets
Demand Potential Existing Products
Demand Potential New Products
Customer Relationships and Data Exchange
Promotion Campaigns
Co-packaging and co-marketing
9.2.3.4 Operations Drivers
Capacity Constraints short / long-term
Materials Acquisition Trends and Logistics
Labor Markets
Distribution Logistics Time and Cost
Manufacturing Scheduling
Assembly Scheduling
Inventory Policies

Copyright 2000 by ECRU Technologies, Inc.

54

Forecasting and Demand Management

9.3 Driver Identification

The company needs to identify the business functions that are

responsible for the specific drivers by type and sector. The roles
and responsibilities for the respective positions need to be updated
to reflect driver-related objectives, and compensation and
incentives programs need to be adjusted to reward early change
detection and assumption management

Initially, the top 10 drivers in each area should be identified, as a

starting point. This allows the company to concentrate on the most
important focus areas right away. Over time, additional drivers
should be identified and included in the demand management
process.

It is particularly important that all functions cooperate in driver

identification and subsequent assumption management.
Forecasting is not a sales function, and cannot be used successfully
without input from many sources.

The forecast manager (the person or function assigned to plan

and manage the periodic/ monthly forecasting and demand
management process) must ensure that each identified function
participates in the analysis and assessment process.

Copyright 2000 by ECRU Technologies, Inc.

55

Forecasting and Demand Management

The combined input from all functions will be consolidated in a

single forecast change for each affected product.

Copyright 2000 by ECRU Technologies, Inc.

56

Forecasting and Demand Management

9.4 Driver Management

To facilitate identification and management of specific drivers, a

standard format should be used that would allow tracking of each
driver, by applicable product, customer, market, etc., and support
generational maintenance so that each forecasting cycle would
have access to preceding assumptions and resulting action
conclusions.

To facilitate the process, a manual work sheet is provided as an

example. It can be used for all drivers until such time as
Information Services can support on-line documentation.

Copyright 2000 by ECRU Technologies, Inc.

57

Forecasting and Demand Management

Drivers and Assumption Work Sheet

Subject:
Internal:

Ref.ID:

Briefly describe the "subject" to which this particular assumption / note applies:
ID.

Identification / Description

ID.

Identification / Description

Customer:
Model:
Product:
Part:
Service:
External:
Market:
Industry:
Competitor:
Comp.Product:
Government:
Date & Period:

Identify the specific date and / or period to which this assumption / note applies:
Date:

Week:

Period: from date:

Month:

Year:

to date:

Background:

Briefly describe the "environment" of the event or fact or development etc. dealt with:

Assumption / Note:

Briefly describe the conclusions drawn / actions to be taken etc., as a result:

Anticipated Effect:

How will the the the event or fact or development likely impact on us? (describe)

Positive:
Neutral:
Negative:
Information Only:
Calculation Model
Process:

Quantification:
Month & Year

How will "factors" be measured, and what calculations are used to arrive at :quantification"
Determine Variables and their units of measure
Determine Result Values and their units of measure
Determine Calculation Formulae
using these variables as input,
and producing Results as output
Identify Sources for each variable (time series, report field, responsible function, etc.)
Identify Use of each Result (in forecast, other functional actions, etc.)
Quantify the effect that we need to ancticipate, for the stated "Subject":
Quantity

In/Decrease

+/-

expand as needed

Follow-up:

units or %

Dollars

In/Decrease

+/-

other Comments:

value or %

transfer these changes to the Change Form for the respective SKU

"Accuracy, Dependability, etc." of Assumption as observed from ACTUAL development:

Date:

Completed by

Name:

Position:

Copyright 2000 by ECRU Technologies, Inc.

Date:

58

Forecasting and Demand Management

The following example illustrates the possible use of such an

assumption work sheet for many different but related or
associated drivers.

For each applicable (important) driver, the company needs to

identify which events can occur, and what would be the time and
quantity and price impacts of each event on the demand. Initially,
the aim should be to concentrate on the top 10 drivers that will
have the greatest impact on performance.

The intended result is the eventual ability to answer the question:

IF event X happens, what will be the effect on my
demand (sales) ?
Which action can we take to enforce positive / mitigate
negative effects ?
How much will it cost us to take that action ?
How much more (sales volume, profitability, cost
reduction etc.) do we expect as a result?

Copyright 2000 by ECRU Technologies, Inc.

59

Forecasting and Demand Management

An example of a forecast cycle involving market share would

be:

a) take expected housing starts (per state) as an objective

basis
b) develop an explosion for each type of start (i.e. for single,
multi family, commercial housing, how many units of our
models do we expect to result from each start
c) multiply the starts with the expected units of models
d) (Differentiate on the basis of quality and price based on
state, etc.)
e) this is the possible market size by month and state,
expressed in units of our models (with added average net
prices if required)
f) define the share of this market that the company wants to
attain, in the respective next forecast period (month, quarter,
etc.)

The result is a forecast market size

g) now derive a statistical forecast from historic data and

scrub it to remove known errors
h) note assumptions (drivers and their values) that would
increase, decrease the demand volume, change the
profitability in a market or for a model; estimate the effect of

Copyright 2000 by ECRU Technologies, Inc.

60

Forecasting and Demand Management

these assumptions (by market, customer, brand, month etc.

as appropriate) and update the assumptions accordingly
i) apply the assumptions (with estimates as noted) to the
scrubbed statistical forecast
j) re-generate the forecast (in units, and apply average net
prices)
k) measure our forecast quantity and value against the size of
the potential market and note the resulting share (%)

The result is a forecast market share

l) For the period, track the actual sales (units with average net
prices) with required / desired degree of detail (by product
and customer per state and region etc.) and calculate actual
units and dollars

The result is an actual sales report for the respective period;

assuming that the forecast market size has not changed in or
prior to the period (refer to assumptions!), the company can
calculate its actual market share (sales versus forecast market
size)

m) measure the actual market share against the forecast market

share and note the difference if any
n) for differences, analyze assumptions and determine likeliest
cause(s) of difference(s)
o) refine assumptions to arrive at better forecast estimates.
Copyright 2000 by ECRU Technologies, Inc.

61

Forecasting and Demand Management

9.5 Applying Demand Management Results

Once all inputs have been obtained, and their respective

quantitative impact on the statistical forecast has been determined,
the cumulative effect of the anticipated changes, on each affected
product, needs to be calculated.

Again, on-line facilities should be made available to track

assumption-based changes on a generational basis.

In the absence of an automated tool, the following work sheet may

be used to consolidate the various inputs for a product:
Forecasting - Demand Management Data Changes
Effective for Forecasting beginning with "Month 1" as month/year:

_________________

This form is intended to capture all desired changes in the forecast of an SKU, as indicated in one or more Assumptions for the SKU
The form assumes "no integration" with front-/back-end processes, and allows you to apply changes in a controlled fashion.
1)
2)
3)
4)
5)

Note that the definition of "SKU" may change depending on the forecast data you use (e.g. parts, customers, models, combinations, etc.)
Prepare one of these forms, for each SKU for which you plan to apply demand management (e.g. make changes resulting from assumptions)
For each assumption affecting this SKU, enter one "line of data changes", with cross-reference to the assumption form or file (for later tracking)
The desired changes must be shown as 'modifying quantities' for the applicable month(s) - note that "month 1" is the next applicable month
Change (or 'modifying" quantitites can be shown as absolute changes (e.g. +100 or -50), or as percent of the original forecast quantity (e.g. +/- 7%)
Note that your specific "front-end" and "back-end" processes may determine how to enter changes for subsequent automatic calculations.
6) In the absence of front and back-end integration, add up all individual changes, per month (lines 1 through n) in line B, then add/subtract this change total from line A.
7) Add/subtract the total changes (line B) to/from the original forecast quantity (line A) to arrive at the new, desired forecast quantity (line C)
Seq.

SKU #

SKU Description

Assumpt. Xref.

Mo.1

Mo.2

Mo.3

Mo.4

Mo.5

Mo.6

Mo.7

Mo.8

Mo.9

Mo.10

A DPAS Forecast Quantity before Changes:

Changes resulting from Assumptions (1 line per A.), below

1
2
3
etc.

n
B Total Changes for this SKU (total for 1 through n):
C Resulting new Forecast(s) for this SKU (C=A+B)

Copyright 2000 by ECRU Technologies, Inc.

62

Mo.11

Mo.12

Forecasting and Demand Management

Line C the Resulting new Forecast(s) for this SKU will be

entered into the forecast system.

These values will replace the calculated statistical forecast

values and be used for subsequent processing.

Copyright 2000 by ECRU Technologies, Inc.

63

Forecasting Strategy and Development

10 The Forecasting & Demand Management Cycle

On a periodic usually monthly basis, the forecasting and

demand management process is performed with least loss of time.
A forecast that takes too long to produce, becomes useless as a
predictive and tracking tool.

There are three phases in the periodic cycle. They should be

integrated in terms of required activities and their timing into
the companys normal month-end closing schedule (or such
other periodic calendar event as may be decided). The three
phases are:

Copyright 2000 by ECRU Technologies, Inc.

64

Forecasting Strategy and Development

10.1 Download and Cycle Start

Periodic Forecasting & Demand
Management Cycle
All non-electronic
All non-electronic
f
documents
documents - P-1
PARTINFO.
PARTINFO.
All electronic
TXT
TXT
Files P -1

FC/DM MANAGER:
FC/DM ba
MANAGER:
Ensure complete
ckup copy of
bathe
ckup
copy of
allEnsure
FC/DMcomplete
data as at
end-ofall FC/DM
data
the
end-ofthe-la
st-cycle
daas
te at
(FC
files,
the-la st-cycle
da document
te (FC files,
a ssumtion,
cha nge
etc.)
a ssumtion, cha nge document etc.)

Develop
Procedures
Generational
Maintenance of
Forecast and
Demand
Management Data

FROM:
_______
Sched. DATE/TIME _______
Act. DATE/TIME: _______
File Names received:
1:
______________
2:
______________
3:
______________

Ensure Cut-off
Management
IS: CLOSE Month
Extract Reference Data Updates
Extract Shipment Data
Format as required
Transmit to Forecast Manager

PARTINFO.
TXT P-1
USETRAN.
TXT P-1

Performance Targets - all R&R

Metrics from "FC systems"
Metrics from "other systems"
Develop PM
Ensure Perform. Data
Process
collection and Metrics
Calculations
All non-electronic
Non - / Electronic
f
documents
Support Informat.
PARTINFO.
PARTINFO.
DPAS
Master
TXT
and Ref.TXT
Files
Develop HR
Process

Roles &
Responsibilities

Organiz.
Charts

Forecasting - Download & Cycle Start

(DPASFCTG.vsd (12) - 01Mar00 - RR)

Develop Scripts
FC/DM MANAGER:
FC/DM
receive
/ loadMANAGER:
P -1 data
receive / load P -1 data

FC/DM MANAGER:
FC/DM MANAGER:
produce "statistical
forecast" for
produce "statistical
periof P; forecast" for
periof
produce Performa
nceP;Metrics for
produce Performa
nce Metrics for
Period P-1
Period P-1

etc.
Perform.Targets
P-1
Perform.Metrics
P-1
Stats. Forecast P

Forecast(s) for
Period P-1
FC/DM MANAGER:
Prepare FC/DM
Cycle PMANAGER:
Forecasting &
Prepare
Cycle P Forecasting
Demand
Management
Material for&
Demand Management
Material for
Distribution
, in agreed format
Distribution , in agreed format

Use to regenerate in the

event of system failures, and
for simulation purposes
requiring old / re-start
information

Actual Sales for

Period P-1
Mktg / Sales
Drivers/Assumpt's
used in P-1 FC
Oper's Drivers /
Assumptions used
in P-1 FC

Calculated based on updated

history ASSUMING all prior
Assumption remain
unchanged
Performance Targets &
Metrics - may be at different
periods
Forecast Reports:
1:
__________
n:
__________
Sales Reports:
1:
__________
n:
__________
Assumtion/Change Copies
1:
__________
n:
__________
etc.

FC/DM MANAGER:
FC/DM
MANAGER:Chart /
From updated
Organization
DISTRIBUTION LIST
To "Responsible
From
updated
Organization
List,
prepa
re Distribution
List forChart /
Attachment List
Positions" for
List,red
prepa
re Distribution
List for
prepa
ma teria
ls, a nd distribute
Expected Actions
Analysis, Updates,
prepa
ma teria
ls, a
nd expectations,
distribute
in red
agreed
format
-set
Expected Return Dates
Response
agreed
format -set dates
expectations,
and in
return
requirements
Develop WEB ?
and return requirements dates
Develop Calendar
distribution ?
IF simulation desired:
re-load backup files, adjust
using DPAS Menu
Selections, re-start process

12

These activities must be performed as soon as the financial month

is closed. All referenced information must be produced as quickly
as possibly and supplied to all parties involved in the forecasting
and demand management process.

Copyright 2000 by ECRU Technologies, Inc.

65

Forecasting Strategy and Development

When these parties receive their documents, the second phase of

the periodic cycle takes place.

10.2 Forecast Review & Demand Management Analysis

Perform.Targets
P-1
Perform.Metrics
P-1
May require re-issue
of correctd FC

Stats. Forecast P

Forecast(s) for
Period P-1
Actual Sales for
Period P-1
Mktg / Sales
Drivers/Assumpt's
used in P-1 FC
Oper's Drivers /
Assumptions used
in P-1 FC

Periodic Forecasting & Demand

Management Cycle
Receiving Process may
differ, depending on agreedupon distribution method /
formats

RESPONSIBE ROLES:
RESPONSIBE
ROLES:
Receive
"Foreca sting
a nd Dema nd
ReceivePackage"
"Foreca sting
Management
and a nd Dema nd
Management
Package" and
ensure completeness
ensure completeness
Develop
Guidelines
Change Form for
Statist. FC Correct.
P

RESPONSIBLE ROLES:
RESPONSIBLE
ROLES:
review
sta tistica l foreca
st
review
sta tistica
foreca
ma ke
correction
cha lnges
a sst
ma ke correction
needed cha nges a s
needed
provide Change
Form to FC Mgr
provide Change Form to FC Mgr

To correct obvious errors and

extraneous influences, such
as
"spike" in demand
"exceptional" demand
etc.
to Forecast
Manager for
Forecast updating,
and re-issue with
instant turn-around

Business & Finance

Driver Front-End
Intelligence
Reports

Economy & Market

Customers

Supporting
Systems

Operations
Develop
Procedures

RESONSIBLE ROLES:
ROLES:
collect RESONSIBLE
"driver intelligence"
as
collect "driver
intelligence"
a ppropria
te for specific
Rolea s
a ppropria
te for for
specific
Ana
lyze "Drivers"
ImpaRole
ct
Ana lyze "Drivers" for Impa ct

Develop
Analytical
Methods

RESONSIBLE ROLES:
RESONSIBLE
review/analyze
FC P-1ROLES:
/ Sales P-1
review/analyze
FC P-1
review
and/ Sales
updateP-1
P-1
review
and update P-1
Assumptions for
Intelligence
Assumptions
for Intelligence
qua ntify
resulting cha
nges
qua ntify
resulting
nges
prepare
Change
forcha
P Forecast
prepare Change for P Forecast
RESONSIBLE ROLES:
RESONSIBLE
ROLES:
Metrics
a nd Performa
nce
a ndAssessment;
Performa nce
MeaMetrics
surement
Mea
Assessment;
suggest
Tasurement
rget Cha nges,
Metrics
suggest Ta
rget Cha nges, Metrics
Changes
Changes

Assumpt. Updates
for FC P
Changes for
Assumpt. Intellig.

Show:
a) changes to assumptions
b) quantitative effect of
changes
c) expected FC changes
d) "net" FC changes

Metrics Changes suggested

RESONSIBLE ROLES:
RESONSIBLE
Implement
AssumptionROLES:
Changes
Implement Assumption
Changes
in P
in P

Results Transmittal
to FC Manager using agreed
methods to deadlines

13

Forecasting - Forecast Review & Demand Management Analysis

(DPASFCTG.vsd (13) - 01Mar00 - RR)

Lastly, when all assumption updates and resulting forecasting

value changes have been completed, the third and last phase
consolidating all suggested changes takes place.

Copyright 2000 by ECRU Technologies, Inc.

66

Forecasting Strategy and Development

10.3 Change Consolidation, Forecast Adoption, Upload

Change Form for

Statist. FC Correct.
P
Assumpt. Updates
for FC P
Changes for
Assumpt. Intellig.
Metrics Changes suggested

involves ALL
Responsible Roles
and Senior
Management - for
ACCEPTANCE

Periodic Forecasting & Demand

Management Cycle

FORECAST MANAGER:
FORECAST
MANAGER:
Receive
Assumption
Upda tes,
Receive
Updact),
tes,
cosolida
tedAssumption
Cha nges (Impa
cosolida
tedTa
Cha
nges (Impa
ct),
suggested
rget/Metric
Cha
nges
suggested Ta rget/Metric Cha nges

may involve
"instant" regeneration of
Statistical Forecast
for corrections

FORECAST MANAGER:
FORECAST
ensure
complete MANAGER:
and timely
ensurefrom
complete
and timely
receiving
a ll Responsible
receiving from
Rolesa ll Responsible
Roles

FORECAST MANAGER:
FORECAST
consolida
te a ll MANAGER:
cha nges,
te a llUpda
cha nges,
va lidaconsolida
te Assumption
tes &
va
lida te Assumption
resulting
Qua ntificaUpda
tionstes &
resulting
ntifica tions
("Sa nityQua
Check")
("Sa nity Check")
FORECAST MANAGER:
FORECAST
MANAGER:
enter
FC Changes,
enter
run Eva
luaFC
tionChanges,
Reports
run Eva
lua tion Reports
generate
/ document
/ enter
generate
/
document
/ enter
a dditiona l cha nges a s needed
a dditiona
l cha
ngesstaPs needed
genera te
Foreca
genera te Foreca st P
FORECAST MANAGER:
FORECAST
MANAGER:
ensure
ma intena
nce of
ensure
ma intena nce
of tes
genera tiona
l Assumption
Upda
genera
l Assumption
a nd tiona
Cha nge
Inputs, forUpda tes
a nd Cha
nge Inputs,
"reproducible"
re-genera
tionfor
as
"reproducible"
re-genera tion a s
needed
needed
FORECAST MANAGER:
FORECAST
ensure
forwardingMANAGER:
of "Target and
ensurechanges
forwarding
of of
"Target and
Metrics"
- and
Metrics"
changes
- and of
suggested
performance
suggested
performance
mea
surement
cha nges
mea surement cha nges

Validated
Assumption
Updates
Validated FC
Changes

Forecast Reports:
1:
__________
n:
__________
Sales Reports:
1:
__________
n:
__________
Assumtion/Change Copies
1:
__________
n:
__________
etc.

Probably requires
"codification" of "tribal
knowledge" as to what
constitutes "sanity"

Ensure distribution accordsing

to all Responsible Roles
Forecast P

Forecast Files
for
UPLOADING
Host MRP &
Host Reporting

IS: Update HOST Files

perform Host MRP analysis
perform Host reporting as defined

14

Forecasting - Change Consolidation & Forecast Adoption

(DPASFCTG.vsd (14) - 01Mar00 - RR)

Throughout the periodic process, the procedures involving demand

management the review and updating of assumption that were
used to produce last periods forecast, an assessment of current and
anticipated changes and preparation of resulting new assumptions,
quantification of volume and timing changes, etc. play a much
more dominant role than the use of the forecasting tool set itself.

Copyright 2000 by ECRU Technologies, Inc.

67

Forecasting Strategy and Development

Copyright 2000 by ECRU Technologies, Inc.

68

Forecasting Strategy and Development

Copyright 2000 by ECRU Technologies, Inc.

69

Forecasting Strategy and Development

Copyright 2000 by ECRU Technologies, Inc.

70