You are on page 1of 15

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/3077076

Rescuing Troubled Software Projects by Team Transformation: A Case Study


With an ERP Project

Article  in  IEEE Transactions on Engineering Management · March 2008


DOI: 10.1109/TEM.2007.912933 · Source: IEEE Xplore

CITATIONS READS
46 4,377

2 authors, including:

Keith C.C. Chan


The Hong Kong Polytechnic University
329 PUBLICATIONS   5,923 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Evolutionary Algorithm for Graphs View project

Multivariate Time Series Analysis View project

All content following this page was uploaded by Keith C.C. Chan on 15 January 2015.

The user has requested enhancement of the downloaded file.


IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT, VOL. 55, NO. 1, FEBRUARY 2008 171

Rescuing Troubled Software Projects by Team


Transformation: A Case Study With an ERP Project
Kim Man Lui and Keith C. C. Chan

Abstract—Many software projects fail, whether failure is mea- if to date it has met just one objective—but does not necessarily
sured in terms of budget, schedule, or some other requirement. mean it will be abandoned. It could still be partially implemented
The causes of such failures are many, but are not always easily in terms of the functions that the system is expected to achieve.
recognized. This is not the least due to the human dimension of
corporate activities, as spurious or misdiagnosed issues in Enter- If a software project is both overbudget and overschedule, we
prise Resource Planning (ERP) projects can take on a life of their do not generally consider it successful. A Conference Board
own and become a magnet for company politics. This paper reports Survey [2] reported that 117 companies that attempted ERP im-
an industrial case in which the senior management attempted to plementation found that, on average, the ERP implementation
deal with a troubled ERP implementation (SAP R/3) in an interna- costs went 25% overbudget. Postmortem reviews, risk manage-
tional fast moving consumer goods (FMCG) company during 2001
and 2002. This paper reflects this dimension as it uses original ment, and critical success factors give us guidance on control
emails and PowerPoint slides to recount a number of representa- actions and preventive actions so that we may stay out of trou-
tive episodes in a troubled but ultimately successful project. At the ble. However, once projects run into trouble, these remedies
heart of this success is the realization that whereas it can be diffi- and analyses are of little help because corrective actions differ
cult and time-consuming to do root-cause analyses, it is relatively from others in a way that we are responding to consequences
simple to identify problem owners. In this case, the senior manage-
ment without IT backgrounds turned around a failing project by of the past events and their influences. This paper considers a
reorganizing the team structure according to process areas so that recent, (2001–2002) real-life case in which an international beer
issues in each process area had one problem owner. We summarize company attempted to implement SAP R/3. The project ran out
the management’s actions into a troubleshooting framework, and of control. It was behind the original schedule and the overall
in addition, suggest three actions for rescuing troubled projects: progress was slow, pausing (e.g., not timely feedback), and even
keep the project manager but narrow down the manager’s scope of
responsibility to one or two process areas; assign the right people moving backward (e.g., rework). The top management had to
to be responsible for other process areas; and have the General take timely action before the project went overbudget.
Manager chair the ERP meetings. With headquarters in Europe and selling in more than a 100
Index Terms—IT project failures, risk management, software countries, KuDrink (not its real name), is a market-leading inter-
project management, team management. national beverage brand. KuDrink Hong Kong Ltd. developed
its core information system on an IBM A/S400 in the early
nineties to automate its core business processes such as sales,
I. INTRODUCTION distribution, and billing. It was a batch-based application in
ANY top executives with previously successful leader- which the database was updated at night. This meant that the
M ship roles in various sizable operations such as merger
and acquisition projects are baffled by an Enterprise Resource
last overnight processing provided the latest source of infor-
mation. At that time, competitive pressures were not so fierce,
Planning (ERP) project. So often, the implementation problems with fewer competitors or outlets, so this application satisfied
not only involve unfamiliar technical implications but also in- KuDrink’s requirements. In the year 2000, KuDrink’s IT strat-
volve organizational changes and internal conflicts. When an egy for both Hong Kong and China was reevaluated and the
ERP project is slipping behind schedule and team members are decision was made to implement SAP R/3 in Hong Kong by
putting a lot of effort into rework, its budget is likely to overrun. 2002.
There are a number of approaches that managers may typically Like many ERP projects, the KuDrink project did not go
use to respond to these issues, such as traffic light reporting, exactly as planned. Internally, there were conflicts among the
in which a project status is reported as “Green,” “Yellow,” or KuDrink departments; externally, there were disagreements be-
“Red” according to whether the project has met three basic tween the KuDrink users and the supporting vendor. The project
project objectives: within budget, on schedule, and achieving plan was revised four times. Worse yet, the project manager put
functions/performance as planned [1]. A project gets a red light it to the vote whether to “go live.” There were problems, but
it was difficult to identify their root causes and dissatisfaction
was evident even at the level of the Board of Directors. This pa-
Manuscript received March 1, 2006; revised September 1, 2006. Review of per reports on the troubled KuDrink ERP project of 2001–2002
this manuscript was arranged by Department Editor R.T. Keller.
The authors are with the Department of Computing, The Hong and discusses how the top management saved the IT project,
Kong Polytechnic University, Hunghom, Hong Kong (e-mail: cskmlui achieving its expected goal within budget.
@comp.polyu.edu.hk; cskcchan@comp.polyu.edu.hk). This paper is organized as follows. Section II reviews the lit-
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. erature on ERP implementation failures, team composition, and
Digital Object Identifier 10.1109/TEM.2007.912933 issues of learning from IT experience. Section III addresses our

0018-9391/$25.00 © 2008 IEEE


172 IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT, VOL. 55, NO. 1, FEBRUARY 2008

TABLE I
SURVEY OF FACTORS CRITICAL TO ERP FAILURES

problem statement for rescuing IS projects. Section IVprovides factors is, therefore, difficult. As a matter of fact, rescuing an
the background to the KuDrink project and discusses a SAP ERP project demands more than risk management; it involves
implementation model. Section V offers a number of anecdotes the psychology behind social behavior.
that illustrate the problems of the project. Section VI describes This section will review three diverse subject areas as the
the solution implemented by the KuDrink management. The res- background of the work. First, we look at difficulties that may
olution was an exemplar of the art and science of management, cause ERP project failures. Second, we briefly consider team
illustrating the crucial point that it is not as important that the composition from a social perspective. Third, we consider how
top management know about IT project problems as it is that we might go about learning from failing projects, perhaps the
they know about the owners of the problems. The final section most challenging aspect of handling a troubled IT project.
offers our conclusion.
A. ERP Project Failure
II. LITERATURE REVIEW The literature on the key factors required for success and
Although there are a number of postmortem studies on aban- failure of an ERP implementation is diverse. Jiang and Klein
doned IT projects [3]–[5], with the focus on key lessons learned report that IS project risk is a multifaceted issue with different
from the successful IT projects [6]–[8], classifications of trou- interacting parts. Table I provides a checklist of ERP issues as
bled IT projects [9], [10], and risk assessment, usually con- reported in the literature as they relate to critical success factors
ducted at the beginning of the projects, to avoid IT project (CSF). Note, however, that for purposes of comparison (see Ta-
failures [11]–[14], little attention has been given to the not so ble I), we list the opposite of the CSF for ERP implementation
uncommon problem of dealing with a troubled IT project, turn- in and SAP implementation in, with the understanding that no
ing the project around, and completing it. Often, troubled or literature has reported that the opposite of CSFs would neces-
failing IT projects do not start with good risk management. In sarily or sufficiently be the factors critical to project failures.
this case, data collected from the projects can be incomplete We also include a list from in which Zhang et al. have consoli-
and inaccurate. Snow and Keil report that the riskier the project dated a number of ERP factors in [15]–[21]. Table I also shows
is, the more inaccurate the status reports are. Reassessing risk factors from five real ERP failure cases in China reported by
LUI AND CHAN: RESCUING TROUBLED SOFTWARE PROJECTS BY TEAM TRANSFORMATION: A CASE STUDY WITH AN ERP PROJECT 173

Xue et al. [22]. For comparison, we also include problems hierarchy can contribute their experience and exercise more po-
discovered from empirical studies on software development sitional power to resolve task conflicts, whereas the end users
failures [3] and from gap analysis between organizational needs can contribute their hands-on experiences.
and ERP functionality [23].
Strong management support is sufficient to successfully
implement ERP; however, when dealing with troubled IT C. Issues in Learning From Experience
projects, the top management plays a dominant role in rescuing The central difficulty in learning from our experience of a
such projects. Management involvement is, thus, a necessary software projects comes with the fact that, due to the dynamic
condition. and multifactor nature of many software project problems, the
immediate or even timely identification of the root causes of
problems may be impossible. For example, changing require-
B. Team Composition
ments during implementation may lead to the ultimate aban-
Past research has suggested that heterogeneity in the ages donment of a software project and this may arise from many
of team members is associated with higher individual turnover causes, such as misunderstanding system limitations owing to
rates [24] while demographic diversity enhances creativity [25]. ineffective user training; a lack of user involvement at the user
Drawing an analogy between demographic diversity and po- requirement stage; or/and lengthy implementation requiring a
sitional (or functional) diversity in an organization, it can be review of potentially out-of-date business requirements. Given
quickly noticed that the ERP team composition influences team that the immediate identification of root causes is not always
performance. Any ERP project must start with the establishment possible, it would be wise for problem solvers to take the
of a project team in which the members are selected from dif- Hippocratic approach and to first ensure they do no harm by
ferent positions in major functional units of the company. Such offering premature solutions.
heterogeneity in a team will, on the one hand, contribute to a We interpret facts, reports, events, and perceptions purpose-
staff member’s functional expertise in reengineering a business fully. Forer [30] found that people with the belief that their
process and designing an integrated system, and, on the other temperaments are considered to be governed by the relative po-
hand, will cause conflicts of interest among different functional sitions of stars tend to accept general personality descriptions as
departments, which, to some extent, may lead to schedule slip- uniquely applicable to themselves and ignore the fact that such
ping, failure to keep to the budget, and low morale among team descriptions might be applied to everyone. Even though we may
members [26]. It has been confirmed during a CRM devel- try to avoid intentional belief, what we perceive greatly depends
opment cycle (e.g., inception, process design, implementation on what we learn and expect [31].
and go-live) that team morale is lowest at the implementation From an engineering management point of view, it would
stage [27]. be interesting for project managers to know whether the diffi-
Regarding team composition and team effectiveness, a survey culty of learning from both software projects and engineering
of the ERP implementation was conducted in mid-February projects (e.g., building a tunnel) is the same. Engineers must
2003 in Taiwan. A total of 102 questionnaires were returned, solve technical problems using models, which, in the field, may
of which 88 were analyzed and the following conclusions were be necessary to further simplify. Experienced engineers can ap-
drawn. ply their experience from one project elsewhere on another.
1) Empirical finding 2.2.1: Functional diversity is negatively In contrast, software engineering involves technical problems,
associated with team performance [25]. which are based on artificial rules and which can have many so-
Team members from different functional areas (e.g., sales lutions. It is by no means clear whether a particular problem is
and marketing, finance, distribution, information systems, etc.) typical or representative or that we will see that kind of problem
may bring different viewpoints to a team and this may not be again.
desirable. This finding is consistent with a previous study [28], Jørgensen and Sjoberg believe that much IT learning is incor-
[29] in which functional diversity was the main source of task rect. Postmortem reviews may include much incomplete and/or
conflicts. incorrect information, and invalid conclusions about causal re-
2) Suggestion 2.2.2: The study recommends that the ERP lationships may be drawn. However, they suggest how learning
package training should be taken into consideration as mistakes can be avoided by using statistical techniques to sup-
soon as an ERP team is established [25]. port our learned experience, by considering alternative perspec-
The project’s effectiveness can be increased and task conflicts tives of the same event, and by understanding the biases and
reduced if staff understand the functions of an ERP package limitations of our learning about a particular event.
early in the life of a project. Brehmer [32] reports that there is a very low correlation be-
3) Empirical finding 2.2.3: Positional diversity has a positive tween length of IT experience and the quality of professional
influence on team performance [25]. judgments. Five factors may contribute to this: 1) we do not
Heterogeneity in terms of job position and positional diversity always understand our own decision-making process [33]. For
may lead to lower levels of conflict within a team and greater example, French or German music was played on alternate days
effectiveness. from an in-store display featuring French and German wines.
4) Suggestion: Involve staff from a broader range of organi- The results were that German wines outsold French ones when
zational levels. Those who are in higher ranks in the company German music played on that day and vice versa. However, only
174 IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT, VOL. 55, NO. 1, FEBRUARY 2008

one out of 44 wine customers could mention the music [34];


2) learning correctly from experience is difficult; 3) we may not
be aware of a fact that we learn nothing from project issues that
have no clear root-cause relationships and what we actually do is
formulate hypotheses or establish beliefs that can explain those
issues, justifying to ourselves our false beliefs. These beliefs
become valuable to us once they can serve as explanations of
our observations of the project events [35]; 4) project events are
fragmented in nature but they are often recalled and reorganized
by us in order to make sense of a whole picture from the flow of Fig. 1. Total annual costs of KuDrink (U.S.$44 million in 2002).
events, thereby distorting our understanding; 5) what we have
learned from one project is so unique that the knowledge is hard
This incentive had no specific target beneficiary. KuDrink had
to generalize from and apply in other situations.
as many as 4000 clients.
Incentives constituted nearly 40% of the total annual costs,
III. RESCUING IS PROJECTS
implying there should be a proper control over how these re-
Table I shows many project issues that can give a project a sources were allocated, in particular, to high repeat customers.
red light. Although we collect exciting project data, interview The legacy ERP application on IBM A/S 400 in KuDrink could
project members, and conduct risk assessments before taking not provide an analysis of past performance, e.g., profitability
remedial actions, a study shows that many troubled projects are on operational transactions of outlets, which would provide a
already in trouble early in the lifecycle. If so, data collected rational way for making decisions and would serve as a baseline
from these projects are inaccurate. How project members tell for judgment. Thus, a project goal was clearly established by
about the project can be biased as project members will speak KuDrink Senior Management; the new ERP application must be
out of self-interest, rather than with the goal of identifying real able to handle rebates, customer sales contracts, logistic costs,
problems with the project. promoter bonus, etc., which altogether produce a profit-and-loss
In many software projects, a small problem often becomes a statement for each customer. This sheet would allow KuDrink
big issue because there is no problem owner or there may be to better allocate its resources and determine the right level of
more than one problem owner. When the problem is one of a service, pricing, and distribution.
conflict-of-interest between two parties, the situation can create SAP has been implemented in a number of KuDrink brew-
political fighting. Moreover, once the project gets into trouble, eries around the world as the adoption of SAP was a KuDrink
project members may be firefighting with (or even hiding) their Cooperate IT policy. Thus, KuDrink Hong Kong decided to im-
own problems and be unconcerned as to how these problems plement SAP R/3 and hired MirageTech Ltd. (not a real name),
might create other problems. And then, there is also the problem a company that in 1999 was certified as a TeamSAP partner col-
of social loafing [36] in which people contribute less to achieve laborating with SAP to deliver business solutions and services to
a goal when they work in a group than when they work alone. SAP clients. Veterans from SAP and MirageTech established an
This paper considers a real case in which a troubled IS project on-site hybrid consultancy team for the KuDrink Project, with
is rescued by reviewing and changing an existing project team MirageTech as team leader. In this paper, we refer to this hybrid
structure so that every project issue always has one problem team as “the ERP consultants” and the team leader as “the ERP
owner to follow up. manager” (see Fig. 2).
Sections IV-A–IV-C will provide background to the KuDrink
IV. KUDRINK AND SAP project and briefly review the project team structure and a SAP
methodology called AcceleratedSAP [37].
KuDrink had a versatile business in the Hong Kong market:
purchasing their products from KuDrink’s China manufactur-
A. Project Team Structure
ing unit, selling its alcoholic drinks, distributing other indirect
competitors’ alcoholic drinks, providing supporting services, As in many SAP projects, roles and responsibilities were
marketing promotion activities and sponsoring various activi- clearly addressed in the Project Charter by the ERP vendor,
ties. Fig. 1 provides a cost breakdown for the major activities including skills and experience required to do the job. Problem-
of the Hong Kong operation in 2002. Administration and staff atically, however, at the beginning of the project, neither the
salary were fixed costs. They did not vary depending on produc- ERP vendor nor the project manager assessed the criteria for
tion or sales levels. Product costs were related to sales volume. qualifications for each role according to the Project Charter. As
“Customer Fund/Outlet Promotion” and “Free Beer/Discount” a result, the team structure failed to reflect the actual competence
were two incentives offered by KuDrink. The first was a time- of KuDrink’s team.
based rebate agreement between KuDrink and each individual Fig. 2 shows the project organizational chart containing brief
customer (which are typically wholesalers rather than end con- descriptions of each role. The Steering Committee consisted
sumers), in which a sales (or volume) target must be achieved of senior executives of KuDrink, the Managing Director from
in a specific period. The latter was a trade discount, including Europe, the General Manager (HK), and the Managing Direc-
discount by percentage and free beer (e.g. buy ten get one free). tor of MirageTech. The project team was headed by a Project
LUI AND CHAN: RESCUING TROUBLED SOFTWARE PROJECTS BY TEAM TRANSFORMATION: A CASE STUDY WITH AN ERP PROJECT 175

Fig. 2. Project team structure by functional module.

Manager, who provided management direction, resolved issues tomer Statements). Both the Module Owners and MIS Pro-
in the project team, and ensured adequate project resources were grammers were from KuDrink’s IT Department.
allocated for all the tasks to be undertaken at each phase of the
project.
The ERP Manager from MirageTech assisted the KuDrink
Project Manager in the definition and execution of project de- B. Project Methodology: Accelerated SAP
liverables and the weekly management of the entire project. The ERP consultants provided detailed electronic documents
The ERP consultants provided SAP software expertise. There describing how to implement a SAP R/3 called Accelerated SAP
were five application consultants mainly responsible for SAP (ASAP). The method required the project team to define de-
application setup and problem solving according to Blue Print tails of project scope, to establish a project organization for
Documents (see Fig. 3). As mentioned, there were consultants roles and responsibilities, to produce a set of business process
from both SAP and MirageTech. blueprint documents, and to control project progress including
A Process Owner had ownership of the process area(s) and quality management, acceptance procedures, regular meetings,
the project deliverables, along with day-to-day management of and distribution of the minutes.
the business area(s). In SAP, Material Management includes The ASAP framework shown in Fig. 3 divides an imple-
both warehouse and purchasing functions, which were man- mentation project into five phases, each of which indicates
aged by two departments in KuDrink. Thus, there were two a specific purpose contributing to the achievement of project
Process Owners for Material Management. The position titles goals. “Project Preparation” provides initial planning and prepa-
of four Process Owners were Sales and Marketing Director, ration for a SAP implementation project. This includes defining
Group Finance Controller, Senior Manager of Warehouse, and a unique objective, scope, priorities, and so on. The purpose of
Group Purchasing Manager. the “Business Blueprint,” as the name suggests, was to create
A Process Support was responsible for the execution of a business process map, which is a detailed document describ-
the detailed process design, master data configuration of the ing how the company ran its business before it implemented
company’s business processes with SAP, and develop Business the SAP applications and how it intended to run it afterward.
Blueprint Documents (see Fig. 3). Process Supports were either “Realization” involves implementing the blueprint or realizing
line managers or KuDrink supervisors. the planned business processes. The phase ends by signing off
A Module Owner had the technical ownership of the process on an integration test that simulates a reengineered business
area(s). model and confirms that it works as expected. “Final Prepara-
An MIS Programmer was responsible for the data mapping, tion” finalizes everything in readiness for going live with the
design, development, and testing of conversion programs, in- new systems and processes. This covers end-user training, data
terface programs, SAP custom reports, and forms. Before the migration, and “cut over” activities, etc. On successful comple-
project’s inception, KuDrink recruited two programmers with tion of this phase, a client is ready to run their business with
one and a half years of work experience in writing SAP ABAP their live SAP system. “Go Live and Support” is the last phase
customized reports (e.g., Ageing Analysis Reports and Cus- of the AcceleratedSAP roadmap [38].
176 IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT, VOL. 55, NO. 1, FEBRUARY 2008

Fig. 4. Costs of a 1.1 million SAP project (in U.S. Dollars).

training course. Nor did SAP provide any feedback to the


KuDrink management as to the exact benefits that KuDrink staff
are presumed to have gained from the training courses, which
cost U.S.$99, 000. Our findings with regard to ERP training can
be summarized in two points: start training as early as possible
and have training postassessments.

V. TROUBLED IT PROJECT
The Consulting Services Agreement between MirageTech
and KuDrink was signed in September 2001. The Agreement
had an estimated productive operation date at the end of March
2002. Seven months after the kickoff meeting, the project re-
ported a number of problems to KuDrink’s senior management.
Many project issues seem to be intuitive and nontechnical,
Fig. 3. Overview of SAP implementation (where • denotes quality check and as if they might be easily understood by non-IT executives yet
signing off at the end of the process).
may require or assume an unexpected or unperceived technical
knowledge. For example, suppose two man-months are required
C. Project Budget to build a house. We may calculate the proportion of resources
required to build n houses as a linear relationship. However,
Fig. 4 provides a final breakdown of the spending in
because of the factors of software complexity and reusability,
KuDrink’s SAP project, but it is being presented here for its
such a linear relationship cannot, however, be used to estimate
cautionary value, as a static picture of spending of this type can
the efforts of writing 20 SAP ARAP reports. The KuDrink’s
be quite misleading as the tendency of costs is to expand and
senior executives, having educational backgrounds in Law and
contract as project conditions change. Here, for example, train-
Business Administration, did not necessarily understand every
ing, and software and hardware equipment were almost constant
point of the issues outside their fields, in this case, business
but two items, consultancy and temporary staff, were prone to
control and project management implementation, but especially
vary widely in range if the project schedule slipped. This is im-
those relating to the resolution of ERP issues.
portant to know because it suggests that it would be prudent to
Sections V-A–V-H provide anecdotes from the project, each
also control the consultants’ activities against the plan, rather
of which reflects one or more of the projects’ wide array of
than as sometimes happens, treating them as being above or
problems. Each of these sections is independent. These sections
outside the plan. The breakdown also shows that consultancy
mainly serve as recounts of what happened in the KuDrink
was the largest portion of the project budget. Since the consul-
project. We include original email texts and PowerPoint slides
tants were paid per hour, so when a project schedule slipped, it
for reference. Section V-H provides our analytical discussion
was likely that the consultancy costs would quickly exceed the
about those projects events in connection with factors critical to
original budget. This is problematic even without regard to the
ERP Failures shown in Section II-A.
conflict of interest which is implied in paying someone more to
fix a problem the longer that they take to fix it. A final cost to
A. Project Plan
note is the 5% markup, which was allowed on top of the total
budget in case of unforeseen contingencies. Fig. 5 shows the original plan, which estimates the project
The team members received the SAP R/3 training from SAP as lasting for 7 months, from September 2001 to March 2002.
before the project actually started. This is consistent with the The system SAP would go live on April 1, 2002. During “Real-
recent recommendation on taking team learning behaviors into ization,” the project schedule was revised and extended several
consideration in training packages as soon as the ERP imple- times: in January, early March, late March, and May 2002.
mentation team is formed. However, it was not known whether The reported reason for extending the length of the project
SAP could provide users with assessment programs after each was that Process Support could not complete their tasks
LUI AND CHAN: RESCUING TROUBLED SOFTWARE PROJECTS BY TEAM TRANSFORMATION: A CASE STUDY WITH AN ERP PROJECT 177

B. Meeting Minutes
The ERP Manager suggested that Process Supports should
take the minutes for their own areas. However, this created
an unnecessary workload (or confusion) because it meant that
two Process Supports would in all likelihood both record the
same event and this would, in turn, mean two follow ups to
the recorded event. For instance, many inventory operations
Fig. 5. Project schedule (∗ denotes revision of the project plan).
would involve both FI (Finance) and MM (Logistics) Process
Supports. Taking minutes in the suggested way would cause
some confusion. In addition, different parts of the duplicated
minutes also had to be consolidated on a single record for filing
on schedule. The tasks were to develop business blueprint and distribution. Following is the email regarding the meeting
documents, change their requirements, request non-SAP stan- minutes. Later, it was agreed that the project manager would
dard reports, perform system tests, and sign off on their work. take all the minutes.
Yet, Process Support complained about their heavy workload in
producing lengthy documents and drawing flow diagrams, and
also reported encountering unknown SAP configuration prob-
lems during testing. Process Supports often reported that previ-
ous testing cases tested okay but, that the same cases failed after
the ERP consultants changed some settings. They felt frustrated
and lost confidence in the system (or, rather, in the consultants
and the Module Owners).
In March 2002, an announcement was formally made by the
General Manager of KuDrink that a special project bonus would
be awarded to the whole project team if the system could go
live in July. However, no significant progress was subsequently
achieved. By the end of April, not only did the project schedule
seem tentative, but team members were confused as to their
responsibilities and the overall schedule, because April, May, C. Work Allocations
June, and July had formerly been proclaimed as an unchangeable Work allocation involves intergroup coordination. As an ex-
project deadline (see Fig. 5). ample, Fig. 6 shows a detailed flowchart for how an invoice note
At the end of April, the KuDrink Project Manager asked all was developed in the KuDrink project.
Process Supports and Module Owners to vote for the date of the Such a clear-cut work allocation caused an unexpected sit-
system going live. Her e-mail follows. uation. The MIS programmers did not know how to issue an
invoice from the SAP. They only knew how to develop the in-
voice note. As a result, they could not perform step 11 (a unit
test) by themselves and could only test their programs with
sample data prepared by the Module Owners. Moreover, the
Module Owners had never programmed a report. Thus, even
in the presence of both the Module Owner and MIS Program-
mer, it was not possible to identify many problems related to
reports.

D. Decision Support
The ERP manager and consultants were not able to pro-
vide many supporting points for decision making and direction
guidelines. They always said, “From my previous experience,
my clients also encountered the same situation and they used
this technique to solve it.” Unfortunately, the ERP manager
could not provide more information about how the previous
case was related to KuDrink and how the result could be mea-
sured after applying the technique. The following e-mail demon-
strated a way in which the ERP manager gave professional
advice.
Note that the parallel run was a system implementation ap-
proach in which a new ERP system was operated along with
178 IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT, VOL. 55, NO. 1, FEBRUARY 2008

Fig. 6. Work allocation and workflow.

Fig. 7. Parallel run versus the straight cutover.

formally announcing their refresh plan to all the project mem-


bers. On the same day, the Process Support staffer wrote an email
explaining that changes to the Chart of Accounts did not cause
any conflict with the previous version of the Chart of Accounts.
the old system for a short period before the old system was shut Overnight, the issue disappeared without the two managers pro-
down. In contrast, the straight cutover was to implement a new viding any explanation of how it was resolved. Changing the
ERP system in operation after shutting down an old system [39]. requirements was not expected in Realization. In conclusion, it
The two approaches are depicted in Fig. 7. is hard to know whether changing the requirements caused the
trouble, or ERP consultants made mistakes owing to the change
of the requirements, or whether the problems had some other
E. Consultant Management source.

In a Process Owner meeting at the “Realization” Stage


(February, 2002), the ERP Manager addressed the necessity for F. Overwhelming Documentation
a system refresh (i.e., reinstallation) due to significant changes In a regular meeting at the Business Blueprint stage, a
in the Chart of Accounts, which was defined in the “Business Process Support requested a hard-copy system manual for
Blueprint.” The refresh would take four working days and incur their reference. The ERP manager replied that manuals were
extra costs (from U.S.$3500 to 18, 000). available only on CD. She strongly advised that they should
The case was challenged by a Process Support staffer who was ask on-site ERP consultants for help as this would be
responsible for the Chart of Accounts while both managers were faster and better. The same Process Owner also requested
LUI AND CHAN: RESCUING TROUBLED SOFTWARE PROJECTS BY TEAM TRANSFORMATION: A CASE STUDY WITH AN ERP PROJECT 179

might be considered quite inflammatory as it created an im-


pression that nonaccountants were challenging accounting prac-
tices. The following response came from the KuDrink’s Process
Owner.

H. Analysis
This section discusses the project events reported in
Sections V-A–V-G according to Section II-A. When no evidence
from the events in the KuDrink’s project is related to a factor
that is critical to a project failure, we can only conclude that
factor is not an issue in the project. We also provide our opin-
ions as to what type of problem the project was suffering from.
KuDrink’s project objective was clear and realistic: the ERP
application can produce a P&L statement by the customer. User
Fig. 8. PowerPoint slide presented to the Steering Committee. involvements did not appear to be any problem throughout the
project. The KuDrink project appeared not to suffer from prob-
lems with (2.1.2) an unrealistic project objective, (2.1.3) wrong
some references such as another companies’ “Business ERP strategy, (2.1.4) unsuitable ERP software and hardware,
Blueprint” at the Business Blueprint Stage, “SAP Business (2.1.6) difficulty aligning to specific requirements, (2.1.17) neg-
Processes and Procedures (BPP),” and “End-user Training” ative user characteristics, (2.1.18) a lack of user involvement,
documents at the Realization Stage. The ERP manager ar- (2.1.20) data conversion behind time, (2.1.21) inadequate tech-
gued that these works were irrelevant because they involved nical knowhow, or (2.1.22) problematic technology base.
different businesses and KuDrink Process Supports might be It may not be straightforward to justify whether the project
misled or be accused of plagiarism. This could be unex- encountered the other issues shown in Table II. For example,
pectedly detrimental to successful SAP implementation, and in (2.1.15) unqualified personnel and (2.1.16) incorrectly es-
thus, the request was denied. Ultimately, Process Support timated staff learning curve, KuDrink recruited two MIS pro-
had to develop substantial amounts of documentation from grammers, who just had very basic knowledge of SAP ABRA
scratch. (SAP report writing language) and knew little about SAP ap-
plications. Thus, no one in KuDrink had SAP applications ex-
perience or SAP project management before. The team got the
G. Internal Conflicts
knowledge through the SAP training courses, but how much
On March 11, 2002, the ERP manager reported the status they actually learnt depends on the individual’s capabilities.
of the project, showing PowerPoint slides on human resources Furthermore, there was no assessment after training. As we
and potential issues. These galvanized Senior Management into could not know how SAP knowledgeable the team members
accepting the difficulties. For reference, a full set of the Power- were, we could not tell whether some problems could have been
Point slides is given in the Appendix. prevented if the KuDrink team had been comprised of staff that
The slide in Fig. 8 supports the idea that the project suffered had 3–5 years of SAP application experience. But one thing
from a human resource problem arising from users requesting an is certain: the success or failure of the KuDrink project would
overwhelming number of reports. The solution could be either to depend on the staff’s learning curve.
increase existing resources sevenfold or to prolong development The original project schedule was perfectly timed for
by 7 months. KuDrink. Alcoholic drinks were seasonal products in Hong
A full set of slides was distributed to the project mem- Kong, as shown by the monthly beer sales volume. This im-
bers after the Steering Committee meeting. One slide (say- plied that workload would be heavier during high season and
ing the concept of accounting accruals having been over used) at that time, internal resources would be harder to reallocate.
180 IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT, VOL. 55, NO. 1, FEBRUARY 2008

TABLE II at the Business Blueprint and the users should not revise COA
FACTORS CRITICAL TO THE KUDRINK’S ERP PROJECT
at the later phases. As expected, the ERP consultants would be
reluctant for any change because rework would be a must. What
came along with such changing would be conflicts between the
users and the ERP consultants.
In Section V-G, the project manager reported to the man-
agement in the steering committee meeting a project dilemma
between more resources without 7 month delay and existing
resources with 7 month delay. The project manager may have
intended to simplify the calculation for presentation or she may
have been unaware of some of the salient characteristics of soft-
ware project management. It is true that building one house for
one man-month, we build seven houses for either 7 man-months
or 7 month men. But, men and months are not interchange-
able [41] in software projects.
The unwillingness of project managers to report negative
news is a contributor to the phenomenon of troubled software
projects. Tan, Smith, and Keil [42], [43] report that people from
collectivistic cultures (i.e., China and Hong Kong [44]) may
be disinclined to report bad news about software projects when
In this case, Months 1–6 and 11–12 were opportune periods they believe that more time given could help the team to get
to implement ERP systems for KuDrink. Interestingly, ERP the projects back on track. In the KuDrink project, we could
vendors’ resources would normally be tight around the end of not conclude that the project manager failed to report the prob-
year because they needed to help almost all clients work on lems to the top Management early enough. It might be equally
year-end closing. Neglecting seasonal effects on a project might possible that the top management were not equipped to fully
contribute to an unrealistic schedule that would suffer from re- recognize all of the implications of IT project issues.
sources deadlock and load imbalance (i.e., 2.1.7. reallocating Although the IT department reported that they did not have
tight internal resources). the full support of top management (2.1.1), this may not be the
Voting for a “go-live” date appeared to be politic as it was case. It may just have been that when management had doubts
not anonymous and the purpose was not for effort estimation. about the practices adopted by the project manager, they delayed
What came along with the voting would be rumors (i.e., 2.1.19 making a decision as to whether to support the IT department.
dishonest Information Policy).
It is normal practice that meeting minutes be taken by only
one person [40], although it may not always be the same person VI. TEAM TRANSFORMATION
at each meeting. Considering the ERP manager had years of SAP Many project problems can be classified into different types
project management experience, we conclude that she would not of issues (Section V-H). The KuDrink problems might be con-
make such a mistake. However, her intention in asking Process sidered in terms of factors critical to ERP failures, although the
Owners to take minutes was unclear. root causes of the problems have not yet been addressed. It is
KuDrink’s problem with work allocations was an example of easier to group an event or a problem by type than to identify
bureaucratic project organization (2.1.14), which often occurs its cause. In this way, we can identify a problem by its type and
in a sizable project. The bureaucratic culture was also demon- place it in a 2-D matrix: process area and functional module.
strated when KuDrink Hong Kong failed to try to second one For example, a problem like user disagreement with new re-
or two IT staff from other KuDrink breweries (for example, port formats may correspond to three functional modules, but it
KuDrink Malaysia) where they already ran SAP. will only correspond to one process area for SAP, APRA report
There was no mechanism for measuring the quality of the ERP development.
consultants’ work products and services. As shown in Sections The process area and functional module can link up with two
V-D and V-E, the ERP vendor’s unilateral suggestion did not types of team structures. A team can be established according
sound reasonable to the Process Supports. From a user point of to the functional model shown in Fig. 2. This is a common
view, this is a problem with the ERP consultant (i.e., 2.1.10). It ERP implementation model. The drawback is the main source
appeared that mutual trust was not building and distrust could of task conflicts, which has been discussed in Section II-B. A
undermine both productivity and morale. team can also be structured with process areas. Many software
When ERP implementation takes more than one and a half development teams are formed in this way: this team is usually
years, changing requirements during the implementation period composed of three or more subteams, one of each being respon-
seems unavoidable. Business needs to respond to the competi- sible for business requirements, development, and testing.
tive market and changing can be for both excellence and surviv- A 2-D matrix of process area and functional module can be
ing. The KuDrink project was not a long project, just 7 months formed. Problems can be identified in the matrix. For exam-
in the original plan. Therefore, COA should be clearly defined ple, problem A shown in Fig. 9 involves both user acceptance
LUI AND CHAN: RESCUING TROUBLED SOFTWARE PROJECTS BY TEAM TRANSFORMATION: A CASE STUDY WITH AN ERP PROJECT 181

Fig. 9. Team composition by either functional module or process area.

testing and report development. Problem B involves three func-


tional modules. The matrix provides an analytical tool for use in
decision making. When the ERP project problems are oriented
to more functional modules than process areas, the team struc-
ture should be organized by process area so that each problem
is managed by one subteam. For example, in problem B (see
Fig. 9), a team structure by functional module (see Fig. 2) says
nothing about which subteam is responsible for dealing with the
Fig. 10. Project team structure by process area.
problem B. However, when a team structure is formed by process
area, the problem B belongs to the subteam of the user accep-
tance test. The team and its subteams can concentrate on their A weekly meeting chaired by the General Manager was held as
commitment to accountability, responsibility, and transparency usual to report and monitor the progress of the project.
in the way the team manages and operates the ERP project. The structure would be of assistance for a General Manager in
coordinating operation flows (i.e., user acceptance testing) and
information flows (i.e., report customization and data migration
A. KuDrink’s Management Decision from A/S400). In addition, whereas the previous team structure
At the end of May, KuDrink was still at the Realization had six reporting channels to the Project Manager, the new
stage. The senior management had a formal meeting with the structure had only four and these flowed to the General Manager.
Managing Director of MirageTech and discussed the approach The project plan was revised based on the decision made by the
to completing their late ERP project. The management sketched four team leaders.
a diagram (i.e., Fig. 9) on the whiteboard. It would seem that the It should be noted that the KuDrink management deempha-
KuDrink management did not pay due regard to that fact that sized the title of Project Manager in the new structure. The
problems can recursively cause other problems. This contrasted General Manager of KuDrink (Hong Kong) was not the project
with the usual practices of IT project managers. As a matter manager because he only chaired the ERP meeting and he had
of fact, the more IT experience a person has had, the more no time to track the project progress on a daily basis.
chance he/she likes to work out what problems might arise. The Accountability is clarified. Team leaders can solve project
KuDrink management found that many problems belonged to problems by function. For example, in the anecdote of work
System Configures and UAT. Each problem should have a sin- allocations reported in Section V-C, the reporting team leader
gle owner responsible for it. Thus, a team structure should be can simplify the report development workflow.
changed to link with the situation. When accountability and responsibility are crystal clear, we
On June 1, 2002, the KuDrink management formally an- can achieve higher project visibility. Ideally each problem or is-
nounced a restructure of the project team shown in Fig. 10. Tasks sue would be bound to one process area, and hence, be managed
in Realization and Final Preparation were combined into just one by one team leader. This would assign every problem to a single
stage. The Finance Director was appointed to take over all the owner. KuDrink management believed that problems that had
responsibilities of user acceptance testing. One Module Owner no owner or more than one owner could not be solved quickly
would be in charge of the report writing. The project manager no matter how simple they might be.
would only be responsible for data migration and one MIS pro- SAP went live on September 1, 2002. After a month, the
grammer was responsible for technical server problems. These legacy system A/S 400 ceased operation. Although the project
four team leaders and the ERP manager reported directly to the was 5 months behind the original go-live date, April 1, 2002, the
General Manager. With this structure, the team leaders managed project was completed within budget and the system achieved
their resources with the direct support of the top management. its goal, a customer P&L statement.
182 IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT, VOL. 55, NO. 1, FEBRUARY 2008

B. Discussion be. If there is no owner or more than one owner, restructure the
This section generalizes some principles from action the team so that for each problem there is always one owner.
KuDrink’s management took to rescue their troubled project. 4) (KuDrink Management) Action 2: In a failing IT project,
The principles are conclusively demonstrated in comparison executives should concentrate on what process areas and func-
with usual practices. tional modules the (technical/nontechnical) problems occur (see
In a survey of 122 projects, project managers in 8 of 76 Fig. 9), rather than what the problems could actually be. Ap-
successful projects and 11 of 48 failed projects were changed propriate actions can then be taken.
during the projects. However, the study did not provide infor- Some executives may promote a project member of staff to
mation as to whom these project managers reported as there can take over some major part of a project manager’s responsi-
be a difference between a decision made by a Chief Executive bilities. The reason is obvious: the project member who has
Officer (CEO) and a Chief Information Officer (CIO). CEOs participated in the project and has understood the organization’s
who know little about the complexity of IT projects may just operations will be highly motivated and committed as such expe-
replace the project manager. This would be like trying to turn riences provide a valuable opportunity for professional growth.
around a losing company by finding a new CEO [45]. 5) Usual Practice 3: Senior Management assigns a person at
1) Usual Practice 1: Senior management pulls the plug on a the same rank or just below that rank as the project manager
leader who fails to drive the project (i.e., the project manager) in the company to take over the functional title of the project
as far as the management is able to put another right person in manager.
charge. To deal with a troubled project, it is advised that the senior
However, CIOs who are expected to be ERP literate believed executive sit in all regular meetings. This has three major ben-
that failing team leaders/members should not be replaced until efits: it rebuilds the team spirit, resolves any politics among
the project is completed or cancelled. Removing team lead- team members, and produces a positive influence on team effec-
ers/members from a late project is counterproductive since we tiveness through positional diversity, as mentioned in empirical
lose the valuable experience they have gained in the project. finding 2.2.3.
Neither adding people nor replacing team members made the 6) (KuDrink Management) Action 3: Senior Management
KuDrink project faster. should assign a person with a higher rank than the project
When to ask a team leader to leave his project is as much manager in the company to take the chairperson role for each
art as it is science. The KuDrink management understood that project meeting.
asking team members to leave the project would be of no help.
The management decided to narrow down the responsibilities
of the project manager to one particular area. Changing the VII. CONCLUSION
team structure resolves (or relieves) existing conflicts between The paper contributes to our understanding of the process
Project Manager and Process Supports, as in the case described of rescuing a troubled ERP project. Past studies have em-
in Section V-G. phasized problem analysis [3]–[14], but this merely provides
2) (KuDrink Management) Action 1: Senior management a postmortem rather than a feasible solution to a troubled IT
should narrow down the responsibilities of the driving staff project. We have also considered why it is difficult to learn from
(e.g., project managers) so that they need to focus on only one failing IT projects. Without understanding what has caused a
or two process areas. project to fail, enforcing software process control and changing
Most senior executives were uninformed about technical im- current practice are little more than trial-and-error approaches
plications. Drawing incorrect conclusions from problems in IT and should be implemented only as last options. However, the
projects can be disastrous for a project. KuDrink case has shown that it is not necessary to spot the root
3) Usual Practice 2: Executives who have engineer- causes of problems when rescuing troubled IT projects. When
ing/science backgrounds tend to prefer determining causality a problem involves two or more subteams, there is a need for
in a failing project, and then, to immediately take remedial ac- greater transparency. It may, in fact, be a better approach to
tions to rescue the project. first confine the problem to one process area under one team
The importance of root-cause analysis has been mentioned in leader, and then, to restructure the project team along a dimen-
Agile Software Development Management [46]. But, as in fail- sion in which accountability, transparency, and responsibility
ing projects, the most secure way to do root-cause analysis is to are in connection with the problem.
base it on project data, rather than causal reasoning. Overcoming Although team transformation is not a silver bullet that cures
the temptation to think in a cause–effect way is particularly hard IT project troubles of all kinds, in our view, the principles behind
for CIOs who are able to combine information technology skills KuDrink’s team transformation are certainly sufficiently generic
with an understanding of the organization across all functions within the corporate-industrial environment that they should be
from problem solving to strategies [47]. The KuDrink manage- applicable to many, if not all, other failing software projects.
ment knew nothing about IT, and hence, would not bother to Keeping the project manager in place but narrowing down the
understand the causality. What the management looked at was scope to one process area, making people responsible for other
not remedial actions, but the connection between the team struc- process areas, and having the General Manager chair the ERP
ture and problem domains. The management tried to understand meetings are ultimately steps that reflect the corporate nature of
who owned the problem rather than what the problem could contemporary software development, which ultimately puts an
LUI AND CHAN: RESCUING TROUBLED SOFTWARE PROJECTS BY TEAM TRANSFORMATION: A CASE STUDY WITH AN ERP PROJECT 183

essentially nontechnocratic and business-oriented emphasis on [27] J. Anton and N. L. Petouhoff, Customer Relationship Management : The
accountability, transparency, and responsibility [46]. Bottom Line to Optimizing Your ROI, 2nd ed. Upper Saddle River, NJ:
Prentice-Hall, 2002.
[28] L. H. Pelled, K. M. Eisenhardt, and K. R. Xin, “Exploring the black box:
REFERENCES An analysis of work group diversity, conflict, and performance,” Administ.
Sci. Quart., vol. 44, pp. 1–28, 1999.
[1] A. P. Snow and M. Keil, “The challenge of accurate software project status [29] K. A. Jehn, G. B. Northcraft, and M. A. Neale, “Why differences make a
reporting: a two-stage model incorporating status errors,” IEEE Trans. difference: A field study of diversity, conflict, and performance in work-
Eng. Manage., vol. 49, no. 4, pp. 491–504, Nov. 2002. groups,” Administ. Sci. Quart., vol. 44, pp. 741–763, 1999.
[2] D. Cooke, L. Gelman, and W. J. Peterson, “ERP trends, the conference [30] B. R. Forer, “The fallacy of personal validation: A classroom demonstra-
board,” Rep. R-1292–01-RR, 2001. tion of gullibility,” J. Abnorm. Psychol., vol. 44, pp. 118–121, 1949.
[3] K. Ewusi-Mensah, Software Development Failures: Anatomy of Aban- [31] S. Plous, The Psychology of Judgment and Decision Making. New York:
doned Projects. Cambridge, MA: MIT Press, 2003. McGraw-Hill, 1993.
[4] J. Verner and W. Evanco, “In house software development: What software [32] B. Brehmer, “In one word: Not from experience,” Acta. Psychol., vol. 45,
management practices are needed for project success,” IEEE Softw., pp. 223–241, 1980.
vol. 22, no. 1, pp. 86–93. [33] R. E. Nisbett and T. E. Wilson, “Telling more than we can know:
[5] M. Jørgensen and D. Sjoberg, “The importance of not learning from Verbal reports on mental process,” Psychol. Rev., vol. 84, pp. 231–259,
experience,” in Proc. Eur. Softw. Process Improv., Copenhagen, Denmark, 1997.
2000, pp. 2.2–2.8. [34] A. C. North, D. J. Hargreaves, and J. McKendrick, “The influence of
[6] C. P. Holland and B. Light, “A critical success factors model for ERP in-store music on wine selections,” J. Appl. Psychol., vol. 84, no. 2,
implementation,” IEEE Softw., vol. 16, no. 3, pp. 30–36, May/Jun. 1999. pp. 271–276, 1999.
[7] N. Welti, Successful SAP R/3 Implementation: Practical Management of [35] J. Preston and N. Epley, “Explanations versus applications: The explana-
ERP Projects. Reading, MA: Addison-Wesley, 1999. tory power of valuable beliefs,” Psychol. Sci., vol. 16, no. 10, pp. 826–832,
[8] Z. Zhang, M. K. O. Lee, P. Huang, L. Zhang, and X. Huang, “A framework 2005.
of ERP systems implementation success in China: An empirical study,” [36] J. M. Jackson and J. M. Harkins, “Equity in effort: An explanation of
Int. J. Prod. Econ., vol. 98, no. 1, pp. 56–80, 2005. the social loafing effect,” J. Pers. Soc. Psychol., vol. 49, pp. 1199–1206,
[9] J. Smith, Troubled IT Projects: Prevention and Turnaround. Stevenage, 1985.
U.K.: Institution of Electrical Engineers, 2001. [37] S. M. Stewart, AcceleratedSAP: Implementation at the Speed of Business.
[10] E. Yourdon, Death March, 2nd ed. Upper Saddle River, N.J.: Prentice- New York: McGraw-Hill, 1998.
Hall, 2004. [38] K. M. Lui and K. C. C. Chan, “Capability maturity model and SAP:
[11] T. DeMarco and T. R. Lister, Waltzing with Bears: Managing Risk on Toward a universal ERP implementation model,” Int. J. Enterprise Inf.
Software Projects. New York: Dorset, 2003. Syst., vol. 1, no. 3, pp. 69–95, 2005.
[12] D. Remenyi, Stop IT Project Failures Through Risk Management. [39] B. D. Hiquet and A. F. Kelly, SAP R/3 Implementation Guide: A Manager’s
London, U.K.: Butterworth Heinemann, 1999. Guide to Understanding SAP. Indianapolis, IN: Macmillan Technical,
[13] R. Schmidt, K. Lyytinen, M. Keil, and P. Cule, “Identifying software 1998.
project risks: An international delphi study,” J. Manage. Inf. Syst., vol. 17, [40] M. E. Haynes, Effective Meeting Skills: A Practical Guide for More Pro-
pp. 5–36, 2001. ductive Meetings. Menlo Park, CA: Crisp, 1997.
[14] J. J. Jiang, G. Klein, and R. Discenza, “Information system success as im- [41] F. P. Brooks, The Mythical Man-Month: Essays on Software Engineering,
pacted by risks and development strategies,” IEEE Trans. Eng. Manage., Anniversary ed. Reading, MA: Addison-Wesley, 1995.
vol. 48, no. 1, pp. 46–55, Feb. 2001. [42] H. J. Smith and M. Keil, “The reluctance to report bad news on troubled
[15] M. Al-Mashari, A. Al-Mudimigh, and M. Zairi, “Enterprise resource plan- software projects: A theoretical model,” Inf. Syst. J., vol. 13, pp. 69–95,
ning: A taxonomy of critical factors,” Eur. J. Oper. Res., vol. 146, pp. 352– 2003.
364, 2003. [43] B. C. Y. Tan, H. Jeff, H. J. Smith, M. Keil, and R. Montealegre, “Reporting
[16] J. S. K. Ang, C. C. Sum, and L. N. Yeo, “A multiple-case design methodol- bad news about software projects: Impact of organizational climate and
ogyfor studying MRP success and CSFs,” Inf. Manage., vol. 39, pp. 271– information asymmetry in an individualistic and a collectivistic culture,”
281, 2002. IEEE Trans. Eng. Manage., vol. 50, no. 1, pp. 64–77, Feb. 2003.
[17] P. Bingi, M. K. Sharma, and J. K. Godla, “Critical issues affect- [44] D. W. Karolak, Global Software Development: Managing Virtual Teams
ing an ERP implementation,” Inf. Syst. Manage., vol. 16, pp. 7–14, and Environments. Los Alamitos, CA: IEEE Computer Society, 1998.
1999. [45] J. Collins, Good to Great: Why Some Companies Make the Leap—and
[18] O. M. Burns and D. Turnipseed, “Critical success factors in manufacturing Others Don’t. New York: Harper Collins, 2001.
resource planning implementation,” Int. J. Oper. Prod. Manage., vol. 11, [46] K. Beck, eXtreme Programming Explained, 2nd ed. Boston, MA:
pp. 5–19, 1991. Addison-Wesley, 2005.
[19] J. F. Cox and S. J. Clark, “Problems in implementing and operating a [47] Y. Li, C.-H. Tan, and H.-H. Teo, “Innovative usage of information tech-
manufacturing resource planning information system,” J. Manage. Inf. nology in Singapore organizations: Do CIO characteristics make a dif-
Syst., vol. 1, pp. 81–101, 1984. ference?,” IEEE Trans. Eng. Manage., vol. 53, no. 2, pp. 177–190, May
[20] K. K. Hong and Y. G. Kim, “The critical success factors for ERP im- 2006.
plementation: An organizational fit perspective,” Inf. Manage., vol. 40,
pp. 25–40, 2002.
[21] V. A. Malbert, A. Soni, and M. A. Venkataramanan, “Enterprise resource
planning: Managing the implementation process,” Eur. J. Oper. Res., Kim Man Lui received the Bachelor’s degree from
vol. 146, pp. 302–314, 2003. Tamkang University, Tamsui, Taiwan, R.O.C., the
[22] Y. Xue, H. Liang, W. Boulton, and C. A. Snyder, “ERP implementation Master’s degree from the University of Witwater-
failures in China: Case studies with implications for ERP vendors,” Int. srand, Johannesburg, South Africa, and the Ph.D.
J. Prod. Econ., vol. 97, pp. 279–295, 2005. degree from the Hong Kong Polytechnic University,
[23] C. Rolland and N. Prakash, “Bridging the gap between organizational Kowloon, Hong Kong.
needs and ERP functionality,” Requirements Eng., vol. 5, pp. 180–193, He has held a number of IT positions in the com-
2000. mercial sector in Hong Kong and China. Currently, he
[24] C. A. O’Reilly, D. F. Caldwell, and W. P. Bamett, “Work group demog- is responsible for the setting up of a software devel-
raphy, social integration, and turnover,” Administ. Sci. Quart., vol. 34, opment center in China. The center adopts agile soft-
pp. 21–37, 1989. ware practices and is funded by a city government.
[25] J. Pfeffer, “Organizational demography,” in Research in Organizational His research interests include agile software development and e-marketing. He
Behavior, vol. 5, L. L. Cummings and B. M. Staw, Eds. Greenwich, is the lead author of two Chinese books on CMM (published by the Tsinghua
CT: JAI, 1983, pp. 299–357. University Press, Beijing, China, 2002) and eXtreme Programming (published
[26] Y. J. Yeh and H. W. Chou, “Team composition and learning behaviors in by the Publishing House of Electronics Industry, Beijing, China, 2005), re-
cross functional teams,” Soc. Behav. Pers., vol. 33, no. 3, pp. 391–402, spectively. His latest book, which is titled Software Development Rhythms:
2005. Harmonizing Agile Practices for Synergy will be published by Wiley in 2008.
184 IEEE TRANSACTIONS ON ENGINEERING MANAGEMENT, VOL. 55, NO. 1, FEBRUARY 2008

Keith C. C. Chan received the B.Math. degree in


computer science and statistics, and the M.A.Sc. and
Ph.D. degrees in systems design engineering from the
University of Waterloo, Waterloo, ON, Canada.
From 1989 to 2003, he was with IBM, Canada
Laboratory, Toronto, ON, where he was engaged in
research on software engineering tools. In 1993, he
joined the Department of Electrical and Computer
Engineering, Ryerson University, Toronto, as an As-
sociate Professor, and in 1994, he joined the Depart-
ment of Computing, Hong Kong Polytechnic Univer-
sity, Hong Kong, where he is currently a Professor and Head. He is also a Guest
Professor of the Graduate University of the Chinese Academy of Sciences,
Beijing, China. He is active in consultancy and has served as a consultant to
government agencies and various companies in Hong Kong, China, Singapore,
Malaysia, and Canada. His current research interests include agile software en-
gineering, data mining, and bioinformatics.

View publication stats

You might also like