Sunteți pe pagina 1din 6

ERP IMPLEMENTATION

Ensuring
E-Business
Success by
Learning from
ERP Failures
Herb Krasner

I
n the 1990s, a plethora of companies at- ing, and then moved into ERP. SAP started by
tempted to use enterprise resource planning specializing in manufacturing automation before
systems to support and improve their internal expanding into other areas. Thus, each product
business processes or to address Y2K com- derives its strengths and weaknesses from its his-
pliance issues. During my recent involvement in tory and its companys current business strategies.
troubleshooting several failed ERP projects, I dis- Some vendors design their products to be flex-
covered a set of common problems that besets ible in capturing and using customer business
these projects.As the shift toward e-business accel- processes; others dictate the processes to be used.
erates in this decade, more companies will face For example, Oracles ERP product is among the
similar problems. Un- most flexible; SAPs is among the least flexible.
Although enterprise derstanding the impli-
cations of these prob-
Over the past few years, ERP products special-
ized for particular industry segmentssuch as the
resource planning lems and learning how apparel industrys JBA System 21 Stylehave
to mitigate the risks also emerged.
may be the toughest they pose will be vital to Highly complex, ERP systems contain many
project youve ever the success of ERP pro- hardware, software, and peopleware components
jects on and off the that can be interconnected in a variety of patterns.
attempted, doing it Web. Pop the hood on one of these systems and youll
right will be the key find dozens of parts and millions of lines of code.
ERP DEFINED These systems are further complicated by the het-
to implementing ERP productssuch erogeneity of the connected components, the
as those by SAP, Baan, newness of the underlying technology, and the
a successful JDE, SSA, JBA, Or- need for integration into the clients total IT envi-
e-business. acle, and PeopleSoft ronment. Some ERP components are self-con-
conceptually contain a tained software packageseither custom-built or
set of functional com- standard products. For example, SAP/R3, Baan,
ponents, integrated around an enterprise data and others now offer various client-server prod-
warehouse. These components provide auto- ucts that reside within an overall ERP system
mated support in traditional business process architecture.
areas such as inventory control, material require-
ments planning, and order processing.
With each product suite emerging from a dif-
ferent historical perspective, todays ERP prod- Inside
ucts offer a wide variety of capabilities. People-
Soft, for example, began by specializing in back-
office systems, then expanded into the front office.
Strategic Options for
Oracle specialized in relational database man- Enterprise Reengineering
agement systems, branched into data warehous-

22 IT Pro January February 2000 1520-9202/00/$10.00 2000 IEEE


DOCUMENTED ERP PROBLEMS applying lessons learned from earlier implementations
Since the early 1990s, many companies worldwide have to later implementations.
attempted to implement ERP systemsand failed. Maga-
zines like Fortune, Datamation, and Information Week have Another report (Stanley Wee, Juggling Toward ERP
published articles on these failures. This publicity has cre- Success, ERP News, Apr. 1999, pp. 7-8) also focused on
ated the perception that large ERP implementation proj- management process issues as one lesson learned from
ects are dangerously failure-prone.To support my consulting interviewing ERP customers. The key to success in over-
work, I have gathered a large compendium of such failures. coming management problems, according to this report, is
Consider the following excerpt from the Harvard Business to establish and maintain momentum by paying attention
Review (Thomas H. Davenport,Putting the Enterprise into to the following factors: focus, teamwork, defined scope,
the Enterprise System, July/Aug. 1998, p. 2). business case analysis, planned training and support, smart
reengineering, overall architecture, rigorous project man-
ERP systems are expensive and difficult to implement.The agement, effective communication of expectations, and
growing number of horror stories about failed or out of top-management involvement.The report did not address
control projects should certainly give managers pause. ... technology issues.
Some of the blame for such debacles lies with the enor-
mous technical challenges of rolling out enterprise sys-
temsthese systems are profoundly complex ..., and In one survey, 62 percent of
installing them requires large investments of money, time respondents cited people issues
and expertise .... The biggest problems are business prob-
lems. Companies fail to reconcile the technological imper- as a prime problem in taking
atives of the enterprise system with the business needs of
the enterprise itself.
ERP projects live.

ERP systems resemble many other large, complex,


sophisticated IT systems. Extremely risky and failure- User problems
prone to begin with, an ERP system by its very nature Financial services firm Deloitte & Touche conducted in-
courts greater risk than other system types, as the follow- depth interviews with 164 individuals at 62 Fortune-500
ing comment shows (Susan Conner,And the Keyboards companies (The Review, Maximizing the Value of ERP-
Connected to the Backbone, Chemistry and Industry, 18 Enabled Processes, Deloitte & Touche, 18 Jan. 1999). All
May 1998, pp. 1-2):The most obvious problem with ERP companies included in the report manufacture consumer
implementations is that these projects are so large and so products, and all use ERP systems by vendors such as SAP,
complex that you cant tell the implementation of the soft- Baan, Oracle, and PeopleSoft.The study focused on going
ware apart from re-engineering the business. live: that point when the ERP system is made available for
These complexities can cause problems with manage- general use within the company.
ment, users, and technical issues. The study concluded that when issues and obstacles
occur before going live, they do so for the following reasons
Management problems and in the proportions reported (the reports listed values
Most ERP disasters reported in the trade press and open do not total 100 percent):
literature stem from management failuresalthough few
companies will admit to it. Many more such disasters have people obstacles, 62 percent;
gone unreported, an assessment supported by studies of business process issues, 16 percent; and
IT project failures in general (CHAOS, The Standish information technology (technical) issues, 12 percent.
Group, http://standishgroup.com/visitor/chaos.htm).
Previous articles have focused on the management dif- After going live, people issues still dominate, but the
ficulties involved in running a complex ERP project. For emphasis of concern shifts to such areas as ongoing sup-
example, Marie Benesh (Managing Your ERP Project, port, business performance, reporting, system transition,
Software Testing and Quality Engineering, July/Aug. 1999, and training. Within IT issues, only about 5 percent of
pp. 38-43) describes five areas of common management respondents considered software functionality an obsta-
pitfalls that involve shortcomings in or a lack of cle both before and after going live.
Unfortunately, the point at which a project should go
integrated project team planning, live is not universally defined and may not always be a firm
managed communications across many people, milestone on the basis of rigorously preestablished crite-
a formal decision-making process, ria. Despite the relative proportions cited for the problems
an integrated test plan and managed test process, and identified, the impact of those problems does not occur in

January February 2000 IT Pro 23


ENTERPRISE
RESOURCE PLANNING
Strategic Options
for Enterprise
the same proportions.Thus, how you deal with these obstacles
Reengineering may be quite different.
Significantly, systems that suffered problems when going live
A first step in enterprise resource planning is to were the best of the lotthe ones that survived most of the
reengineer, downsize, rightsize, or otherwise stream- process, management, and technology risks that kill so many
line business processes in pursuit of a competitive other projects before they can go live. So its really no surprise
edge or greater efficiency. This process, sometimes that the remaining issues that confronted these projects
called enterprise reengineering, promises big re- involved people problems.
wards but also carries big risks.
ER initiatives start with a planned strategy for Technical problems
streamlining and integrating the companys opera- When you autopsy a failed ERP project, identifying the prob-
tions. The following choices describe the spectrum lems that occurred along the way can provide indicators of the
of strategic options available for reorganizing a failure to come. Many ERP projects are plagued by complex
companys business processes. technical problems, which fall into the following general cate-
gories:
Functional centralization. This option man-
dates that a single organizational unit performs nonrobust and incomplete ERP packages,
all common business functions, such as human complex and undefined ERP-to-legacy-system interfaces,
resources or accounting. middleware technology bugs,
Functional standardization. Taking a cloning poor custom code, and
approach, this option has one business process poor system performance.
fit all business units and mirrors that
uniformity in the IT systems im- The ERP package vendor promises that its product
plementation. Functional stan- usually the next versionis robust and has all the func-
dardization implies that a com- tions and features to support your business processes.
pany must standardize on hard- If you naively believe it, go to the end of the line.
ware platforms, choose a standard Independent certification of such a package can be
suite of ERP packages, then con- worth its weight in gold, and firms are setting up shop
vert each and every business unit to do this (for an example, see Hugh Klipp, Quality
to those standards. Assurance Services for ERP, Proc. Ninth Intl Conf.
Distributed functional specialization. This Software Quality, American Society for Quality, Milwaukee,
option designates a selected business unit as a Wis., 1999, pp. 131-144).
center of excellence for a specific business func- To function, an ERP package must be placed in an IT envi-
tion; the selected unit then performs that func- ronment that already contains other systems. The interfaces
tion for the whole corporation. Partial functional between these systems are complex and significantly affected
specialization allows each business unit to con- by the overall IT integration approach.You will need rigorous
trol its own core MIS functions, such as sales legacy-systems analysis to precisely understand the ERP-inter-
processing and inventory control, but also cen- facing requirements. You should also document each major
tralizes a limited set of business functions, such interface definition.
as accounting, across the company. Middleware technology has emerged as the glue to interface
Functional business unit autonomy. This de- these various systems and applications over a distributed ERP
centralizing option exemplifies the noninte- system. You can choose from several standard approaches
grated enterprise approach, allowing each unit based on CORBA (Common Object Request Broker Archi-
to go its own way. tecture), DCOM (Distributed Component Object Model), and
Radical redesign of business processes. An other distributed architecturesor roll your own. Be aware
organization could choose this option in that buggy client-server messaging or object-sharing tech-
response to a paradigm shift in which a new nologies will bring the entire system to a halt.
approach replaces the normal business flow. Most ERP packages dont cover all the specific business
requirements, and therefore you may need a certain amount of
Hybrid strategies combine features of these basic customization to complete the fit. In some troubled projects, cus-
options. For example, you can loosely coordinate tom-code development started without sufficient thought and
standardization, allowing each business unit to pur- planning.The subsequent undisciplined development produced
sue an independent IT strategy, but insisting that it poor-quality code with unexpected, unlocalized side effects that
still cooperate with business units on bulk hardware arose from poor structuring.Custom code also significantly raises
and software purchases. the total cost of ERP system ownership over the long haul
because IT shops must upgrade the
customizations each time a vendor Figure 1. E-business system architecture.
upgrades the base ERP package.
Once an ERP system is fully loaded
with the necessary corporate data and
hundreds or thousands of people start ERP e-b2b
E-commerce and supply
using it, performance can suffer if you other chain
applications
havent designed for scalability. Users MIS/IT management
who must wait more than a few sec- Customers systems applications Suppliers
onds to see the on-screen results of
Products/ Parts/
their input will not be happy. Nor will services services
the regular appearance of the dreaded
blue screen of death please them. Integrated, enterprisewide
information system
ERPs silver lining
Although trade publications tend to
focus on ERPs failures, many ERP
projects have succeededthey just
dont seem to get much press. One well-documented excep- but that rarely happens. Even progressive companies have
tion is Cisco Systems, which reportedly installed a com- done so well on the front end that they are now unable to
pany-wide ERP system for about $30 million in only nine meet customer demands with sufficient product offerings
months (Cisco Systems, Inc.: Implementing ERP, Harvard or services.
Business School Publishing, Report 9-699-022, 20 Oct. 1998, Lack of strategic thinking seems to get many companies
http://www.hbsp.harvard.edu). When you look closely at in trouble. For example, no matter how much a given com-
what Cisco did, you realize it has taken IT to a new level, pany may want its e-business to run off a turnkey system,
integrating with its suppliers and customers to an amazing the components that make up these systems will not come
degree, and realizing tremendous efficiencies and capabil- from a single source any time soon. Even if such a system
ities as a result. It showed that a strong focus on business existed, a companys business processes are unlikely to
goals in defining a system could lead to business success. match the default embedded processes. Further, vendors
can implement the systems component pieces with dif-
E-BUSINESS SYSTEM CHALLENGES ferent underlying technologies and designs, forcing the
E-business systems are the latest fad in advanced IT and organization to overcome much greater integration chal-
business process improvement. Figure 1 shows the emerg- lenges than if the underlying infrastructure was consistent.
ing architecture for such a system in a large company. Consequently, I believe that new middleware technology
While watching some of my clients join the e-commerce must emerge that lets us glue these systems together. I dont
game, Ive often observed that the following ease into it know who will invent this technology or even when we will
scenario works well. These clients start easy, establishing see it. Its not Java or ActiveX. Its not CORBA or DCOM,
a Web site presence. Then they incrementally redesign because I also wonder where organizations will find expert
their Web page to add better organization, additional integrators who know how to architect the whole solu-
information, enhanced searching, linking, and interactive tionat present they are in extremely short supply.
capabilities. Next, they attempt the challenging task of tak-
ing customer orders or handling customer transactions via Assessing project risks
their Web page. If they clear this hurdle, they attempt to Most of todays corporate CIOs and CTOs dont under-
integrate their Web-based front-end system with existing stand the complexities of establishing an e-business.Aside
legacy IT systemsa task so difficult that many organi- from the enormous management and people challenges,
zations fail to implement it. If they successfully integrate many technical challenges confront such projects.
their legacy systems, they then try adding other e-com- Examining e-business through the lens of the FURPSI
merce capabilities, like supply chain management. Finally, functionality, usability, reliability, performance, supporta-
with all these capabilities in place, they begin to see new bility, integratabilitymodel of system quality (Herb
ways of running their business and start improving their Krasner, Using FURPSI for Software Product Quality
systems and processes accordingly. Evaluations, to appear in Software Quality Professional,
June 2000), I see increased technical risks in several areas.
Strategic vision is key
If the organization has analyzed, planned, and archi- Functionality. Systems will need to support a more inte-
tected well, the system functions as a fit and useful whole grated style of business processes, including womb-to-

January February 2000 IT Pro 25


ENTERPRISE
RESOURCE PLANNING

AN E-BUSINESS
Figure 2. Typical project life-cycle model. SYSTEM PROJECT MODEL
A corporate e-business project typi-
E-business goal cally consists of five major phases, as
shown in Figure 2. These phases, per-
Strategy definition formed incrementally, usually overlap
heavily.

Existing-systems analysis and Strategy definition. Phase one ad-


new system capability planning dresses the creation of a corporate e-
business strategy that contains a vision,
objectives, and an approach to
System definition and development
automating, improving, and integrat-
ing the companys business processes
across the entire enterprise. The strat-
Deployment
egy also defines the general scope of
an e-business system in terms of how
it will support the organizations busi-
Use and evolution ness processes.
Existing-systems analysis and new sys-
tem capability planning. Phase two
Time begins with an analysis of existing
legacy systems, followed by the plan-
ning and architectural definition of the
tomb management of customer, company, contractor, new systems capabilities.
and supplier relationships. System definition and development. Phase three involves
Usability. Systems must support the emerging point-and- the precise definition and development of the integrated
click generation. e-business computer system that will provide support
Reliability. Systems must achieve the uptime goal of 24 for each of the companys designated business units.
hours a day, seven days a week, with 99.9999 percent Deployment. Phase four focuses on deploying the
availability as the backup goal. Systems also need to be e-business system within one business unit or across sev-
safe and resistant to illegal penetrations. eral to support the organizations business processes.
Agility. Demand for shorter Web response times will Use and evolution. Phase five occurs when the organi-
grow as people tire of the World Wide Wait. zation effectively uses the deployed systemfor
Supportability. Systems must improve their capabilities example, to improve the organizations efficiency, effec-
in a smooth evolution rather than through a constant tiveness, or competitiveness. Lessons learned from this
barrage of herky-jerky upgrades and bug fixes. use feed back into earlier phases as the system evolves.
Integratability. Complexity will drive the movement
toward component-based integration and glueware To reap the e-business systems promised benefits, all
the Lego block approach. More organizations will move phases must succeed. Mistakes made in the first phase will
toward a distributed system built around a tiered archi- poison efforts in the second, and so on.
tecture. In some cases, an organization will outsource portions
of the project. For example, instead of the companys own
When you add up the management, people-resistance, and MIS or IT department, you could contract with an IT sys-
technical-complexity risks, its a wonder any substantive e- tems integration company to develop the e-business sys-
business system project successfully achieves its lofty goals. tem. This solution presents a double-edged sword,
In this decade, existing ERP systems will expand to meet however: While outsourcing can supplement your com-
e-commerce and supply chain management systems. Major panys systems knowledge, it also exposes your company
new versions of ERP software will be rearchitected to use to some risks in terms of performance and a knowledge
new technologies like Java and XML.The common archi- gap within your organization.
tectural stratification standards may also give way to new In the past decade, most ERP projects occurred over a
distributed-systems approaches. For example, its not clear multiyear cycle. In this decade, however, increased com-
how existing client-server systems will meld with Web- petitive pressures will demand that e-business systems be
based technologies. delivered even more rapidly. The challenge of accelerat-

26 IT Pro January l February 2000


ing these phases will lead to rapid, incremental, and iter- growing knowledge base of common risks encountered,
ative approaches to the essential activities of planning, thus enabling less problematic acquisitions of such pack-
systems analysis, system definition, hardware acquisition, aged systems. For more information about how this tech-
software acquisition, custom software development, sys- nique has been used in past situations for organizational
tems integration, testing, documentation, pilot use, instal- risk management, see the article From Project Risks to
lation, and evolution. Organizational LearningA Path to Practical Knowledge
In ERP projects, Ive seen many cases in which signifi- Management (Joyce Statz and Don Oxley, Application
cant iteration occurred both within the early phases and Development Strategies, June 1998, http://www.teraquest.
across those phases. For example, it is not uncommon to com/risk/).
see the first and second phases iterate several times before

I
the planning is sound. Often, people think they know what f you consider your e-business system project to be cru-
processes they want to automate right to the point when cial to the future of your organization, you should want
the requested software packages actually come in-house. to treat the knowable risks to success formally and
Then gap analysis tells them it would be a lot cheaper to proactively so that you can focus resources and mitigate
adapt their processes than to rebuild the software, so its potential project failures. The techniques Ive presented
back to the planning phase again. here can provide tools for starting this process.

MANAGING ERP RISKS Herb Krasner is an independent software systems/man-


Based on my experiences with ERP projects in the agement excellence coach and troubleshooter at Krasner
1990s, Ive found that for your project to succeed, you must Consulting. Contact him at hkrasner@cs.utexas.edu.
prevent problems in the following high-priority areas:

e-business strategy,
project management approaches,

RENEW
complex technology and systems, and
end-user resistance.

A small investment in understanding, analyzing, track-


ing, and preventing these major problems should pay off
well given that most such problems are common, expected, your Computer Society
avoidable, and detectable early. One viable approach to membership for
reducing the likelihood of failure is to identify the crucial
risks at each point of the project life cycle, then build plans The best rates for
to address them before they become real problems. I have IT Pro
found this technique useful in evaluating ERP projects
and fully expect it to be useful for evaluating e-business 12 issues of Computer
system projects as well.
My colleagues at TeraQuest and I have built a risk fac- Member discounts
tors model-and-analysis toolset for these types of e-busi- on periodicals,
ness-system projects (see http://www.teraquest.com).The conferences, books,
model captures the experience gained from studying ERP proceedings
failures as well as from identifying areas of improvement
in the management capabilities of organizations that per- Free membership in
form large software projects using component packages. Technical Committees
For example, the toolset identifies 76 risk factors in eight
key categories for consideration when selecting an ERP or Digital library access
e-commerce package. It then rates these factors as low-, at the lowest prices
medium-, or high-risk indicators.Working with the factors
and toolset, a project team uses the high-rated factors to Free e-mail alias
plan for the mitigation of a risky selection decision. @computer.org
IT personnel currently use the toolset to analyze lessons
learned on completed projects, review the status of proj-
ects under way, and assess risk planning as projects get http://www.ieee.org/renewal
started. Data collected from all these sources provides a

January February 2000 IT Pro 27

S-ar putea să vă placă și