Customer Demand

Planning at Lucent
Technologies
A Case Study in Continuous Improvement
through Sales Forecast Auditing

Mark A. Moon
John T. Mentzer
Dwight E. Thomas, Jr.

One of the critical drivers of supply chain success is effective
customer demand planning. A company must be able to accu-
rately forecast its business opportunities, and then effectively
plan throughout its supply chain to exploit those opportunities.
Unfortunately, many companies fail to understand the impor-
Address correspondence to John T. Mentzer, Department of Marketing,
Logistics and Transportation, University of Tennessee, 310 Stokely
Management Center, Knoxville, TN 37996-0530. Tel: (423) 974-1652; fax:
974-1932. E-mail: Jmentzer@utk.edu
tance of effectively forecasting customer demand. This article
presents a case study of one company–Lucent Technologies–that
is committed to continuously improving its ability to effectively
forecast customer demand. Lucent recently demonstrated that
commitment by conducting a sales forecasting audit, performed
by external experts, that has had a significant impact on its sales
forecasting process and performance. The case study presents
the process for conducting a sales forecasting audit, the benefits
Lucent has realized from the process, the challenges they con-
tinue to face, and the factors that can be applied to other
organizations. © 2000 Elsevier Science. All rights reserved.

Industrial Marketing Management 29, 19–26 (2000)

© 2000 Elsevier Science Inc. All rights reserved. 0019-8501/00/$–see front matter
655 Avenue of the Americas, New York, NY 10010 PII S0019-8501(99)00108-X
 Customer demand planning is essential for
supply chain coordination.
INTRODUCTION
One of the critical drivers of supply chain success is
effective customer demand planning. A company must
be able to accurately forecast its business opportunities,
and then effectively plan throughout its supply chain so it
can exploit those opportunities [1].
Unfortunately, many organizations fail to recognize
the top- and bottom-line effects of poor customer demand
planning. From a top-line perspective, companies cannot
take advantage of demand for their products and services
if, first, they fail to accurately forecast that demand and,
second, they fail to develop and implement appropriate
plans to build the required supply chain capacity to fulfill
that demand. From a bottom-line perspective, effective
customer demand planning allows a company to mini-
mize its supply chain costs by minimizing inventory, pur-
chasing, logistics, and production costs.
One large industrial company that has recognized the
leverage available from effective customer demand plan-
ning is Lucent Technologies. The purpose of this paper is
to describe the process improvements that Lucent Tech-
nologies has undertaken over the last several years, as
well as the challenges that they continue to face. This
company’s experiences help to point out principles of ef-
fective customer demand planning that will allow other
organizations to improve their management of this criti-
cal business process. It will also describe an audit process
that has helped Lucent Technologies identify opportuni-
MARK A. MOON is a faculty member in the Department of
Marketing, Logistics & Transportation, the University of
Tennessee, Knoxville, Tennessee.
JOHN T. MENTZER is the Harry J. & Vivienne R. Bruce
Excellence Chair of Business Policy in the Department of
Marketing, Logistics & Transportation, the University of
Tennessee, Knoxville, Tennessee.
DWIGHT E. THOMAS, JR. is a senior logistics executive at
Lucent Technologies.
20
ties for improvement, and provided a platform upon
which fundamental organizational change can occur.
COMPANY BACKGROUND
Lucent Technologies was formed in 1995 when AT&T
divided into three major businesses. What used to be sev-
eral manufacturing units within AT&T, including Net-
work Systems, combined with Bell Labs to become a
new company called Lucent Technologies. Lucent Tech-
nologies primary product families now consist of Sys-
tems for Network Operators (revenues for the 12-month
period ending September 30, 1998 of $18.8 billion);
Business Communication Systems ($8.1 billion); and
Microelectronic Products ($3.0 billion). As a result of its
recent merger with Ascend Communications, Lucent is
now one of the leading global suppliers of data network-
ing systems. Other major product lines include switching
and access systems, wireless communications systems,
optical networking systems, and fiber optic systems.
CUSTOMER DEMAND PLANNING
Lucent Technologies has established an organization,
as well as a core business process, called Customer De-
mand Planning (CDP). The CDP process is defined as a
business planning process enabling sales teams (and cus-
tomers) to develop product/application demand forecasts
as inputs to inventory and production planning, revenue
planning, and services planning processes.
This definition indicates that Lucent Technologies has
an enlightened view of the separate, yet interconnected,
roles of forecasting and business planning. Many compa-
nies use the terms forecasting and planning almost inter-
changeably, but Lucent recognizes that they are separate,
and sequential, processes. Forecasting is seen at Lucent
as the process of developing the most probable view of
what the level of future demand will be, given a set of as-
sumptions about technology, competitors, pricing, mar-
keting expenditures, and sales efforts. Planning, on the
 Understanding customer requirements
is important for success in
competitive markets.
other hand, is seen at Lucent as the process of making
management commitments that will allow the company
to efficiently and effectively respond to the demand fore-
cast. Forecasts at Lucent precede plans: it is not possible
to make a management commitment, for example, to
production scheduling or to purchasing and inventory
levels (supply chain/business plans), until a forecast is
developed that gives a reasonably accurate, most proba-
ble view of what demand will be over the forecasting ho-
rizon.
Lucent regards CDP as critically important to its busi-
ness because of the large number of internal “customers”
that must be satisfied by providing accurate, credible
forecasts. Following is a list of the various internal cus-
tomer groups or business processes that cannot function
effectively without the forecasts that are provided via the
CDP process:
• Finished goods inventory planning
• Production requirements planning
• Commitments to external customers and to internal Lu-
cent sales teams
• Production reservation scheduling
• Allocation of limited supply products
• Supplier coordination
• Strategic planning gap analysis
• Engineering/installation/distribution planning
• Global product demand forecast data repository
• Allocation of inventory
• Revenue forecasting and planning.
The CDP organization at Lucent sees itself as an inter-
functional process planning and management organiza-
tion which oversees the collection of data from different
organizational functions, such as sales, marketing, and
operations, and then provides information that facilitates
effective decision-making back to other organizational
functions, such as logistics, finance, and manufacturing.
The CDP process has been deployed globally at Lu-
cent since 1993. Since late 1998, the CDP process
counted over 2,000 different individuals from 33 coun-
tries and 170 U.S. zip codes as users. Input is provided by
79 sales teams worldwide, as well as direct sales input by
29 key external customers and 15 marketing organiza-
tions. The CDP process supports production planning
and scheduling at 20 factories or joint ventures around
the world. There are over 16,000 monthly forecasts gen-
erated for over 2,000 forecast lines, using a single global
schedule, data repository, and set of procedures. Signifi-
cant CDP system training has been implemented to sup-
port this forecasting organization, consisting of 14 tele-
training classes, as well as traditional classroom and
video techniques. In addition, there is significant user
support available, including extensive documentation, as
well as a Web site and hotline designed to answer ques-
tions and resolve problems.
The CDP forecasting system focuses on forecasts in units,
which can then be converted to revenue forecasts using
historical average price data. Those who provide input to
the CDP system have the ability to enter forecasts using a
number of different forecast hierarchy levels, including prod-
uct family, individual product by customer, project, or to-
tal application, a predefined combination of several differ-
ent products. The normal time horizon for the CDP
forecasting process is a rolling 12-month horizon. The CDP
system provides a direct feed to the production schedul-
ing systems at Lucent’s manufacturing sites. In fact, one
feature of the CDP system is that the factory is required
to respond to all forecasts with supply commitments that
can then be passed along to customers. These commit-
ments are also used by sales teams to schedule reserva-
tions, which are advance orders for long lead projects.
The CDP system is designed to be “user friendly” for
those people in the sales organization at Lucent who in-
put information into the forecast. There is an emphasis on
21
 Management commitment is not possible
without an accurate forecast.
graphical tools, and the user interface is designed to pro-
vide a “point-and-click” environment. In order to provide
salespeople inputting information into the forecast with
the most recent information, 3 years of global customer
demand history, plus year-to-date customer demand, is
included in the system. This history is updated weekly. In
addition, known future demand, which takes the form of
long-term contract business for which there are not yet
specific orders, is detailed for 12 months into the future.
Also, order-level information is available for the previous
3 months and for all future-committed orders. The system
also provides the capability for sales teams to enter sub-
jective information, such as forecast risk and significant
upside or downside variables.
While the CDP system is very well engineered, it did
not just happen. Accuracy measures were not originally
at a level that the company felt was acceptable, and this
led Lucent to undertake a sales forecasting audit in a con-
tinuous improvement effort to evolve the CDP process.
THE SALES FORECASTING AUDIT
In late 1996, management at Lucent came to the real-
ization that forecast accuracy was at unacceptably low
levels. Management had established overall targets of
75% to 85% accuracy across all major product lines and
businesses [2], yet accuracy numbers were consistently
in the 60% range. In addition to this, a major internal op-
erations re-engineering program was underway, one
which created an environment that was conducive to
changing a number of key business processes. Finally,
changes in management resulted in new executive own-
ership of the forecasting process. Together, these factors
created an atmosphere where a comprehensive audit of
the entire forecasting process could be supported across
different levels of the organization.
To help identify opportunities for improvement, a
team of academic experts, who in the past had performed
similar audits for over 30 companies [3], was commis-
sioned to perform this audit. This choice was critical for
two reasons. First, because the audit team was not a part
22
of Lucent, those involved in forecasting could honestly
report their problems and concerns, without fear of retri-
bution. Second, because the audit team brought with
them more than two decades of experience in sales fore-
casting [4], and because of the perspective of the other
companies they had audited, new ideas and approaches
that might have been outside the normal thought pro-
cesses at Lucent could be considered.
The process for conducting the forecast audit at Lucent
began in June 1997, with a 2-day meeting between the
audit team and representatives from the Lucent’s CDP
organization. At this orientation meeting, the audit team
was given a detailed description of how the current fore-
casting systems and processes were intended to work.
Detailed documentation of the CDP process was pro-
vided to the team for analysis and evaluation. Over the
next month, 21 teleconference interview sessions took
place with 64 different process stakeholders. Efforts were
made to include a wide range of individuals from differ-
ent organizations in this data collection process. The au-
dit team interviewed personnel who were forecast pro-
cess owners; who contributed input to the forecasts–such
as sales and marketing personnel; and who were users, or
customers, of the forecasts–such as manufacturing and
logistics personnel. It was truly a global data collection
effort, with personnel from China, Europe, India, and
Japan, as well as participants from the United States, in-
cluded in the interviews. For each interview, two mem-
bers from the audit team were present to facilitate inter-
rater reliability. All interviews were audio taped to allow
for clarification and to resolve disagreements at the time
of data analysis. Following collection of the data, the au-
dit team members met to reach agreement on how Lucent
compared to other companies in their database, and to de-
velop a set of recommendations that would help Lucent
improve.
The findings from the audit could be summarized into
four major themes. First, the audit team found significant
“islands of analysis” at Lucent. These “islands of analy-
sis” can be thought of as separate information systems
and processes that are not interconnected–physically,
 Many groups cannot function without an
accurate forecast.
electronically, or conceptually. Second, it was found that
performance was not adequately measured and, as a re-
sult, there was a lack of performance evaluation. In other
words, there were no rewards to anyone at Lucent for do-
ing a good job of forecasting, nor were there any negative
consequences for poor performance. Third, the forecast-
ing process, which on the surface appeared to be quite
rigorous, was, in fact, based almost entirely on qualita-
tive, or judgment-based, techniques. There was minimal
use of statistical tools to uncover patterns of historical de-
mand and project those patterns into the future. Finally,
the audit team determined that there was a limited com-
mitment to sales forecasting as a business function. This
limited commitment was manifested in a number of
ways, including: lack of training for those who provided
inputs to forecasts; limited financial commitment to sys-
tem resources; inadequate policy directives which could
have provided uniform guidance to forecasters world-
wide; and limited use of performance metrics along with
insufficient opportunities for reward and recognition for
forecasting excellence.
In terms of specific recommendations, the audit team
presented Lucent with the following four major catego-
ries of recommendations:
1. Training. While members of the forecasting group at
Lucent were adequately trained in how to manage Lu-
cent’s CDP system, there was inadequate training in
the role of forecasting in the organization, as well as
the opportunities that were available to employ statis-
tical analysis to improve forecasting accuracy. In ad-
dition, the sales teams at Lucent provided the bulk of
forecast input, and while those individuals had re-
ceived training on how to interact with the system,
they had received no training in how to forecast. The
audit team recommended enhanced training programs
to address both of these deficiencies.
2. Reward and recognition. The audit team recom-
mended that Lucent institute a program of rewards
and recognition that would be directly linked to fore-
cast performance. It was recommended that this pro-
gram go beyond the forecasting group, and include
members of the sales organization, whose input was
so critical to the process. The audit team pointed out
that this reward and recognition program needed to be
significant–there needed to be enough at stake that
forecasting would “get on people’s radar screens.”
While compensation is an ideal way to reward excel-
lence in forecasting, there are also opportunities to
employ recognition events and programs as a way to
get people to pay attention to this critical process.
3. Metrics. It was pointed out in the audit that Lucent did
not have adequate performance metrics in place,
which is clearly a critical requirement of a reward and
recognition program. Unless performance can be
measured, it cannot be adequately recognized and re-
warded. The audit team emphasized in their report
that the contribution of every individual who interacts
with the forecast must be measured, and that feedback
mechanisms must be put into place so that all those in-
dividuals know whether their contributions to the
forecast have been positive or negative.
4. Process. The audit team recommended that Lucent
employ a knowledgeable core group that would use
statistical tools, such as time-series and regression
models, to develop an initial forecast view. These ini-
tial statistical forecasts could then be distributed to
those knowledgeable individuals, such as marketing
directors and sales teams, for their adjustment. It was
recommended that this core group be positioned in the
marketing organization, with enough “clout” and visi-
bility to be able to truly impact the forecasting behav-
iors of others in the organization.
IMPLEMENTATION OF THE AUDIT
RECOMMENDATIONS
Within each of the four broad areas of recommenda-
tions outlined above, Lucent has had some successes and
some challenges. Following is a “progress report” on
how Lucent has dealt with each of the recommended
23
 Developing a supply chain forecast is a
multiple-step process.
changes, approximately 14 months after the audit recom-
mendations were delivered.
Training
Following the recommendation given by the audit
team, Lucent has significantly enhanced the training op-
portunities available to its employees. Of most interest is
a series of training videos aimed specifically at the sales
teams located around the world. These videos are de-
signed to give salespeople an appreciation of the impor-
tance of forecasting in the effective operation of the busi-
ness, as well as to provide skills in how to adjust
statistical forecasts, given the salespeople’s knowledge of
current market and customer conditions. Several hundred
copies of these tapes have been distributed, and the vid-
eos are also available for viewing on Lucent’s Intranet.
Reward and recognition
This area has proven to be the most challenging to Lu-
cent. Management of the forecasting group has not been
fully successful at “selling” higher level management on
the notion of salary or other compensations linked to
forecasting performance. The message that has been de-
livered to forecasting management is that such linkages
would be inconsistent with corporate compensation di-
rections. In spite of this, some progress has been made to
develop broad recognition programs, which have had
some success at getting the attention of members of the
sales organization. While change has not been easy in
this area, there has, in recent months, been some renewed
interest in revisiting the reward element of this recom-
mendation.
Metrics
The audit team’s report and recommendations were
successful at giving an impetus to more widespread use
of accuracy metrics. Soon after the completion of the au-
dit, an extensive software enhancement was completed,
allowing calculation of these metrics in a multitude of
24
ways, including by an individual forecaster. Thus, be-
cause these metrics are now available, the forecasting
group has the tools in place to be able to implement the
reward and recognition programs discussed above.
Process
As for organizational changes, it is often the case in
large, global organizations that changes in executive
leadership and corporate reorganizations can come fre-
quently, and these can have an impact on the speed with
which process change can take place. This impact has oc-
curred at Lucent and, as a result of such leadership
changes and reorganizations, some of the organizational
process recommendations have been implemented and
some have yet to be addressed. For example, Lucent has
just implemented an extension of the CDP process that
provides a new unit planning capability to its annual bud-
geting process. This new capability, Strategic Volume
Planning (SVP), supports capacity planning by the engi-
neering, provisioning, and installation resource units
within the company. This was the first time that re-
sources across the company were planned from a com-
mon baseline. This process also establishes the capability
to monitor the monthly most probable forecast view
against the budget as the year progresses. Lucent has es-
tablished a “gap analysis” Web site capable of doing this.
However, Lucent is still struggling to properly engage its
various marketing organizations in the forecasting pro-
cess. This has proven to be difficult, in part because of
frequent internal restructuring, as well as the indepen-
dence given to each of its business groups.
RESULTS
Despite the challenges with which the company has
been faced to implement the recommended changes, Lu-
cent has achieved some significant improvement follow-
ing the sales forecast audit. Specifically, while forecast
accuracy was in the upper 50% to lower 60% range in
 Forecast accuracy can be
dramatically improved.
early 1997, it was consistently in the 80% to 85% range
in late 1998.
In addition, there has been broad-based improvement
across the various sales teams. Before the audit, a fore-
casting package had been implemented, allowing the fi-
nal forecast to be a combination of time-series, regres-
sion, and qualitative techniques. However, there was
wide variance from one sales team to another in terms of
the quality of their qualitative forecasts. A year after
completion of the audit, that variance has been reduced,
and accuracy has improved across the board. Signifi-
cantly, Lucent met its on-time customer delivery objec-
tives for 1998, met its sales growth objectives of 20%
plus per year, and brought inventory levels more under
control. Each of these corporate accomplishments points
directly to improved forecasting performance as an im-
portant contributor.
CONCLUSIONS
The experience of going through a sales forecasting
audit has been extremely beneficial for Lucent. Consider-
able effort went into development of the CDP system and
procedures prior to the audit, but the CDP team was re-
ceptive in learning from the audit that forecasting prac-
tices at Lucent had room for improvement. There was
tremendous value in obtaining the insights from experts
who had not “grown up” in the Lucent environment and
who were not influenced by previous experience with the
CDP process. These insights provided the company the
opportunity to rethink fundamental process design and
deployment. The audit also provided a platform for af-
fecting organizational change. Regardless of whether Lu-
cent management agreed with various findings or recom-
mendations put forward by the audit team, these findings
and recommendations have provided, and continue to
provide, a framework for continuous improvement.
There are several learnings from the Lucent experience
that can be used at other companies. One of the most im-
portant insights that came out of the audit process was
that companies must be prepared to confront “clashes”
with corporate culture and entrenched ways of doing
business. Because of these inevitable conflicts, agents of
change must, above all, be patient and tenacious. It will
not always be possible, for various cultural or political
reasons, to immediately accomplish everything necessary
to become world-class in forecasting. Rather, it is impor-
tant to win small victories and demonstrate success,
thereby building the momentum that may be needed to
make more substantive changes. This is the road of con-
tinuous improvement on which Lucent has begun to
travel in their efforts to become world-class in this criti-
cal business function.
Discussion of other learnings can be organized around
the four focus areas that emerged from the audit: training;
reward and recognition; metrics; and process. First, in terms
of training, the experience at Lucent has reinforced the
importance of extending forecasting training beyond the
forecasting department. Because they rely heavily on the in-
put from their sales organization, Lucent felt that it was
necessary to deliver forecasting training to its sales teams.
Every company should identify those individuals whose
input is critical to sales forecast accuracy, and then de-
velop and deliver training programs that help those indi-
viduals understand both how to forecast and why the ac-
curacy of forecasts is so important to the organization.
Second, on the subject of rewards and recognition, Lu-
cent discovered that this area was the most difficult to ad-
dress. While there was no disagreement at Lucent regard-
ing the axiom “what gets measured gets rewarded and
what gets rewarded gets done” [4], changing the way that
people are compensated may be the most politically
charged issue the company has faced. While it may be
politically charged, there is no better way to influence
people’s behavior relative to the forecasting process than
through a well-designed reward and recognition program
for excellence in forecasting.
Third, in terms of metrics, an important learning that
emerged from the audit process at Lucent was that individ-
ual behavior cannot be influenced unless a process is in
place to measure individual contribution to forecast accu-
racy. These metrics should capture how accurate the fore-
25
cast would have been without an individual’s contribution,
and how accurate the forecast was with that individual’s
contribution. Such measures are not only important to en-
able the reward/recognition efforts discussed above. These
measures are also important in deciding whether an individ-
ual should provide input to a particular forecast at all.
Finally, in terms of process, Lucent has learned that
obtaining perspectives from multiple sources will lead to
better forecasts. For example, they learned that asking
salespeople to forecast at the customer level, over a rela-
tively short time horizon, made forecasts more accurate.
They also learned that asking marketing personnel to
forecast at the product-line level over long time horizons
made the forecast better. In other words, people with in-
sights about future demand patterns should be asked to
provide input to forecasts, but only at the level and time
horizon at which they normally operate.
An additional source that Lucent found to have im-
proved forecasts was history. Following the audit, Lucent
began using sophisticated time-series and regression
tools to identify patterns in historical demand data, and
this too helped to improve their forecasts. All companies
can enhance their forecast accuracy if they use a combi-
nation of quantitative (statistical) techniques to find pat-
26
terns in historical demand, with qualitative (judgment)
techniques to adjust those statistical forecasts.
In summary, the forecast audit process at Lucent was
successful in two critical ways. First, after implementing a
number of the recommendations that emerged from the au-
dit, Lucent improved their forecast accuracy dramatically.
This improvement has led to reductions in operations and
marketing costs and improved levels of customer service.
Second, the audit has provided Lucent with a framework
for continuous improvement. While this continuous im-
provement process is always challenging and occasionally
painful, the rewards have been clearly worth the cost.
REFERENCES
1. Kahn, K. B., and Mentzer, J. T.: EDI and EDI Alliances: Implications for
the Sales Forecasting Function. Journal of Marketing Theory and Practice
4(2), 72–78 (1996).
2. Mentzer, J. T., and Kahn, K. B.: Forecasting Technique Familiarity, Satisfac-
tion, Usage, and Application. Journal of Forecasting 14(5), 465–476 (1995).
3. Moon, M. A., Mentzer, J. T., Smith, C. D., and Garver, M. S.: Seven Keys
to Better Forecasting. Business Horizons 41(September–0ctober), 44–52
(1998).
4. Mentzer, J. T., and Bienstock, C. C.: Sales Forecasting Management. Sage
Publications, Thousand Oaks, California, 1998.