Ross l Strategic IT Architecture Competency

CREATING A STRATEGIC IT
ARCHITECTURE COMPETENCY:
LEARNING IN STAGES1

Jeanne W. Ross Executive Summary

MIT Center for Information IT architecture is often assumed to follow business strategy, to align IT with the
Systems Research business’s strategic objectives. Increasingly, though, many business strategies
depend on specific underlying IT capabilities. To develop a synergy between
business strategy and IT architecture, firms must develop organizational com-
petencies in IT architecture. My research has identified four IT architectural
stages, each with its own requisite competencies. The “application silo archi-
tecture stage” consists of IT architectures of individual applications. The
“standardized technology architecture stage” has an enterprise-wide IT archi-
tecture that provides efficiencies through technology standardization. The “ra-
tionalized data architecture stage” extends the enterprise-wide IT standards to
data and processes. And the “modular architecture stage” builds onto enter-
prise-wide global standards with loosely coupled IT components to preserve the
global standards while enabling local differences. Each stage demands different
organizational competencies to implement the architecture and prepare the firm
to move to the next stage.2

At United Parcel Services (UPS), strategic planning My research suggests that few firms experience such
involves discussions about the opportunities the firm’s “happy surprises.” UPS creates business opportunities
IT capabilities present. For example, as the Internet by leveraging its centralized package data, low-cost
started gaining steam in the mid-1990s, IT executives processing environment, and integrated core applica-
noted that UPS’ package tracking capability was eas- tions. But most firms’ IT capabilities limit rather than
ily transferred to the Web. Later, management ob- create new business opportunities. Their limitations
served that package tracking data made new prod- come from their history of applying IT as a response
ucts—like guaranteed delivery—affordable. Package to specific business needs. The business needs are
tracking data also led to the creation of new customer isolated and their solutions rarely combine to create a
services, such as allowing a customer’s customers strategic capability.3 Consequently, when buffeted by
(package recipients) to view an online summary of changing market conditions, most firms see their IT
expected delivery times from selected suppliers. These architectures as competitive liabilities.
IT-enabled business opportunities have been unantici-
pated sources of value for UPS. CEO Mike Eskew But that is not always the case. My colleagues and I
calls them “happy surprises.” 12 have written 40 case studies4 of firms that are evolv-

3
Broadbent, M. and Weill, P., "Management by Maxim: How Busi-
1
Cynthia Beath was the senior editor accepting this article. ness & IT Managers Can Create IT Infrastructure," Sloan Manage-
2
My thanks to Peter Weill, David Robertson, George Westerman and ment Review, Spring 1997, pp.77-91, describes how IT infrastruc-
Nils Fonstad for their significant contributions to my thinking on IT tures result from either maxims, which focus individual IT invest-
architecture. I am particularly indebted to Jack Rockart, John ment decisions on the firm’s strategic intent, or deals, which build IT
Mooney, Leslie Willcocks, and Bob Zmud for timely, insightful infrastructure through a series of isolated decisions.
4
comments that led to radical revisions. I am grateful to the managers The conceptual framework for this paper resulted from a project with
who participated in this research and shared their experiences and Peter Weill in 2001 that examined IT architecture through case stud-
insights. This research was made possible by the support of CISR ies at 8 global firms. I refined the framework based on several other
sponsors and especially CISR patron Microsoft Corporation. research studies, including a project in 2002 with David Robertson,

© 2003 University of Minnesota MIS Quarterly Executive Vol. 2 No. 1 / March 2003 31
Ross l Strategic IT Architecture Competency

ing their IT architectures from sets of isolated solu- of the relationship between IT and business processes.
tions to planned capabilities that support their strategic The resulting architecture does, indeed, identify the
business processes. This evolution has followed a mass of complex linkages among technology compo-
learning process. The firms have not derived value nents, but it does not highlight the few IT capabilities
simply by linking IT to their business processes. critical to enabling the firm’s strategic objectives.8
Rather, they have learned how to benefit from IT by Accordingly, the city plan metaphor has failed to cap-
developing a competency in creating and evolving an ture the strategic potential of enterprise IT architec-
enterprise IT architecture. ture.

In some firms the enterprise IT architecture acts as a

What is an IT Architecture? tool for aligning IT and business strategy.9 This
alignment focuses on the IT components that enable
The term IT architecture lacks a universally accepted critical business processes. Thus, at the enterprise
definition. In fact, the terms architecture and infra- level, an IT architecture is the organizing logic for
structure are sometimes used interchangeably, with applications, data, and infrastructure technologies, as
architecture seen as the plan for the next infrastruc- captured in a set of policies and technical choices,
ture. 5 More often, IT architecture refers to a firm’s list intended to enable the firm’s business strategy. Ac-
of technology standards. But viewing IT architecture cordingly, the enterprise architecture implies certain
only as technology standards does not connect it to IT capabilities. These capabilities are the objectives of
business requirements. the IT architecture, specifying what the architecture
enables the business to do. IT capabilities would in-
The enterprise IT architecture concept, though, does clude, for example, being able to access specific data
place technology standards in the context of business for new applications, quickly add channels to existing
requirements. Consultants and researchers often refer processes, integrate data from related processes, en-
to an enterprise IT architecture as a kind of city plan sure secure processing for electronic transactions,
that details policies and standards for the design of provide an extended online customer service, or repli-
infrastructure technologies, databases, and applica- cate systems in new locations. Rather than develop an
tions.6 The city plan concept has given birth to a breed exhaustive list of possible IT capabilities, a well-
of IT architects who develop detailed drawings of the designed enterprise IT architecture highlights the IT
interconnections between processes, infrastructure, capabilities that are most critical to a firm’s strategic
data, and applications.7 However, these detailed draw- objectives.
ings often provide only the technologist’s perspective
What is an IT Architecture
Competency?
George Westerman, and Nils Fonstad, which involved 16 case stud-
ies on the relationship between IT architecture and business strategy. A competency in enterprise IT architecture is the abil-
A 2002 project with Cynthia Beath and John Mooney explored 3
new cases on Web services implementations. Between 1998-2003 I ity of a firm to create a mutually reinforcing pattern of
wrote 13 additional case studies that assessed IT architecture in the evolving, tightly aligned business strategy and IT ca-
context of e-business implementations, ERP implementations, merg- pabilities. The logical sequence for developing an en-
ers, service level agreements, and IT governance. I am indebted to
all the research colleagues who participated in this research and
terprise IT architecture is assumed to be as follows:
shared their insights.
5
Weill, P. and Vitale, M., “What IT Infrastructure Capabilities are 1) Define the firm’s strategic objectives.
Needed to Implement E-Business Models,” MISQ Executive (1:1),
March 2002, pp. 17-34, define IT infrastructure as the base founda- 2) Define key IT capabilities for enabling those ob-
tion of budgeted-for IT capabilities (both technical and human)
shared throughout the firm as reliable services, and centrally lo- jectives.
cated.
6
Keen, P.G.W., and Cummins, J.J., Networks in Action, Wadsworth,
Belmont, California, 1994, p. 279.
7
Most proposed frameworks link technology to business process (see,
8
for example, Zachman, J.A., "A Framework for Information Sys- See Goodhue, D.L., Kirsch, L.J., Quillard, J.A. and Wybo, M.D.,
tems Architecture,” IBM Systems Journal (26, 3), 1987, reprint “Strategic Data Planning: Lessons from the Field,” MIS Quarterly
G321-5298, and Sowa, J.F. and Zachman, J.A., "Extending and (16:1), March 1992, pp. 11-34, for a discussion of how detailed
Formalizing the Framework for Information Systems Architecture." planning of data architecture can overwhelm planners with volume
IBM Systems Journal (31, 3), 1992, reprint G321-5488). However, and cause them to lose focus. The result is little value from architec-
these frameworks do not help distinguish the relative importance of ture activities.
9
processes. The output of the architecture process is often volumes of For additional discussion of how IT architecture aligns IT with busi-
detailed drawings that are overwhelming in their volume and scope. ness strategy, see Sauer, C. and Willcocks, L.P., “The Evolution of
This output may be the cause or a result of the architecture function the Organizational Architect,” Sloan Management Review, Spring
often being buried in lower levels of the IT organization. 2002, pp 41-49.

32 MIS Quarterly Executive Vol. 2 No. 1 / March 2003 © 2003 University of Minnesota
 Ross l Strategic IT Architecture Competency

3) Define the policies and technical choices for de- ple of the tradeoffs firms face, as one IT architect de-
veloping the IT capabilities. scribes:

Completing this sequence is challenging, at each step. The security function is working very, very
A major difficulty in the first step is obtaining the hard at locking down the corporation so that
firm’s strategic objectives. One firm’s chief architect we can't be penetrated and put out-of-
described a common scenario among firms trying to business by hackers, viruses, and the like.
align business strategy and enterprise IT architecture: But if we really do that job to the extreme,
we can cripple our ability to compete on the
So we started working on understanding the Web.
business strategy, and what we discovered
in that process was that they really didn't Thus, the process of developing an enterprise IT archi-
have a business strategy. What they had tecture is not as orderly as assumed. More impor-
were a lot of promises. “We are going to tantly, defining and developing IT capabilities to sup-
grow. We are going to use branding. We port business strategy is merely a starting point. The
are going to run our plants more effectively. objective is to get to the point where IT capabilities
We are going to increase our volume.” But shape business strategy while business strategy shapes
they hadn't figured out exactly how they IT capabilities in response to changing market condi-
were going to do it… And what I said was, tions and organizational realities. To do this the firm
“It is very difficult for me to write an IT must develop an IT architecture competency to dy-
strategy to support your business strategy namically adjust strategies and technologies.
when you don't have that defined.”
Our case studies illustrate that firms hone their ability
The second step, defining a set of critical IT capabili- to define and align IT and business strategy by accu-
ties with lasting value, is equally challenging. Once mulating architecture-related experiences. When used
strategic objectives have been defined, they generally to enrich organizational learning, these experiences
demand multiple IT capabilities, which are likely to be can create enterprise IT architecture competencies.
interdependent, possibly contradictory, and perhaps Johnson & Johnson offers one example of this cumu-
unachievable given the firm’s legacy. lative learning process.

For example, in 1991, John Reid articulated the vision An Example of Developing Enterprise IT
for Citibank to include continuous innovation, agility Architecture Competencies: Johnson &
in meeting customer needs, and attention to cost.10 Johnson
Citibank Asia Pacific pursued this strategy by devel-
oping four IT capabilities: a low-cost, high-volume In 1995 Johnson & Johnson, a respected pharmaceuti-
processing environment, global look-and-feel to Citi- cal, health care, and medical devices firm, had over
bank access points, global accessibility to customer 150 operating companies generating total revenues of
systems, and electronic access to all customer sys- approximately $15 billion. Analysts both inside and
tems. But some capabilities were not immediately outside the firm attributed J&J’s success in large part
achievable. For example, Citi relied on a mainframe to to its autonomous management structure, which held
build its low-cost processing environment. But, at that managers accountable for the financial results of their
time, no mainframe software offered global accessibil- individual operating companies.11 Despite its virtues,
ity to customer systems. Although the bank ultimately this decentralized approach frequently left customers
achieved its targeted capabilities, it could not achieve frustrated. Many of them had to deal with multiple
them all in the short term. sales calls, multiple invoices, and multiple contracts
with J&J operating companies. They wanted a single
The third step highlights yet another difficulty in point of contact.
building IT capabilities. The policies and technical
choices for developing IT capabilities must reflect Management vowed to respond by presenting a single
organizational realities and thus inevitably require face to its key customers. However, its IT capabilities
tradeoffs. Security policies provide a common exam-
11
See Tanouye, E., "Johnson & Johnson Stays Fit by Shuffling Its Mix
of Businesses," Wall Street Journal, December 22, 1992, p. A1; also
Weber, J., "A Big Company That Works," Business Week, May 4,
10
Details are available in Brand, A., Weill, P., Soh, C. and Periasamy, 1992, pp. 124-132; and also Ross, J., “Johnson & Johnson: Building
P., “Citibank-Asia Pacific: Positioning IT as a Strategic Resource,” an Infrastructure for Global Operations,” CISR Working Paper 283,
Melbourne Business School, 1999. 1995.

© 2003 University of Minnesota MIS Quarterly Executive Vol. 2 No. 1 / March 2003 33
Ross l Strategic IT Architecture Competency

had developed around the decentralized business tems to standardize critical data across their operating
model. IT supported the individual needs of the oper- companies. Through these efforts, J&J has evolved its
ating companies, not the demands of global custom- business strategy, its IT infrastructure, and its technol-
ers. To meet these new customer demands, senior ogy management practices toward a more strategic
management not only needed to implement technol- enterprise IT architecture competency.
ogy that would provide a single view of the customer
but they also needed to reorient everyone in the firm J&J’s journey is not unique. Despite differences in
to think about IT—and organizational—capabilities at industry, culture, IT requirements, and organizational
the corporate level as well as the operating company structure, firms attempting to design, implement, and
level. The technology and organizational change ef- leverage enterprise IT architectures share common
forts proceeded in parallel. experiences. I have identified four distinct stages that
capture evolving architectural designs and their chang-
One of the first IT initiatives to align the strategic re- ing strategic implications and managerial demands.
orientation and a more integrative IT environment was These stages help us understand why firms cannot
a series of training sessions for small groups of IT simply “declare” that they will use IT strategically.
managers from the various J&J operating companies. They must continuously build their competency to do
These training sessions, which took place over 18 so.
months, explained the role of IT in enabling the corpo-
rate vision and proposed corporate-wide IT standards.
Early participants thought the idea of corporate stan- Four IT Architecture Stages
dards was radical and dysfunctional. Participants in
the later sessions, though, arrived for the training al- Firms’ common experiences in evolving their IT ar-
ready aware that managers were trying to operate dif- chitectures suggest four distinct stages of increasing
ferently. They had learned the value of corporate enterprise IT architecture competency. These stages
standards and were looking for ways to facilitate inte- differ in the logical design of their applications, data,
grated communications across the operating compa- and infrastructure; the IT capabilities they provide; the
nies. strategic opportunities they present; and the IT man-
agement and governance processes they demand. The
An early corporate initiative involved installing a sin- four stages are:
gle global network and desktop configuration. J&J had
traditionally funded IT initiatives within the operating 1) An application silo architecture – The architecture
companies. Senior management recognized that this consists of architectures of individual applications
funding model would delay implementation of the rather than an architecture for the entire enter-
corporate infrastructure, so J&J provided some corpo- prise;
rate funding (with eventual chargeback) to stimulate
the standardization. Through this early infrastructure 2) A standardized technology architecture – The IT
initiative, management started to learn how to assess architecture becomes enterprise-wide and pro-
and fund corporate-wide IT investments. vides efficiencies through technology standardiza-
tion and, in most cases, centralization;
Over the years, J&J has continued to evolve its enter-
prise IT architecture, reflecting organizational learning 3) A rationalized data architecture – The enterprise-
about viable strategies and emerging IT capabilities. wide IT architecture expands to include
J&J did not dismantle its operating company structure. standardization of data and processes;
Its strategic objectives continue to foster strong oper-
ating companies, while leveraging cross-company 4) A modular architecture – The architecture builds
synergies where appropriate. Thus, IT capabilities— onto enterprise-wide global standards with loosely
and accompanying policies and technical choices— coupled applications, data, and technology com-
must reconcile sometimes competing IT needs. J&J ponents to preserve the global standards while
has also implemented a shared services organization enabling local differences.
to introduce the efficiencies of a standardized IT envi-
The evidence from my research suggests that firms
ronment. It has created committees to establish and
can generate significant business value at each stage
monitor technical standards. New organizational units,
when they capitalize on the architecture’s benefits.
such as sectors, create formal structures that link op-
Organizational learning at each stage helps them un-
erating companies that have shared customers and
derstand how to realize those benefits and how to po-
markets. Some sectors are introducing common sys-
sition themselves for the next architectural stage.

34 MIS Quarterly Executive Vol. 2 No. 1 / March 2003 © 2003 University of Minnesota
 Ross l Strategic IT Architecture Competency

Firms that attempt to skip stages consistently find that Despite its benefits, the application silo architecture is
either the benefits are severely delayed or they must largely outdated. Over time, the applications form the
backpedal to acquire the missing organizational com- firm’s legacy, which consists of independent applica-
petencies. Following are the characteristics of each tions on multiple technology platforms with embedded
stage and their benefits and risks.12 data. The developers’ freedom to innovate becomes
offset by the difficulty of linking new applications to
The Application Silo Architecture Stage related systems. The applications become as much a
burden as a blessing. A plant manager at Dow Corn-
Firms in the application silo architecture stage focus ing, a $2 billion manufacturer of silicon products, de-
their IT resources on delivering individual applica- scribed the limitations of his firm’s legacy applica-
tions. Typically, IT develops or buys an application to tions prior to implementing an ERP system:
address a specified business need and hosts each ap-
plication on the best available technology platform. When I went to the Midland Plant in
This approach lets them develop superior applications 1992,…[the legacy] was becoming an enor-
for IT-enabled processes. But, in most cases, these mously ugly patchwork quilt. The systems
processes are limited to a single function or geogra- wouldn’t talk to each other. Nice functional-
phy. ity, but they wouldn’t talk to each other.

In this stage, firms thus allocate their IT resources By allowing variety in technology platforms, applica-
primarily to application development. They might tion silos are expensive and difficult to maintain.
have a centralized data center for transaction process- More importantly, multiple data definitions make de-
ing, but they have few shared infrastructure services. velopment of electronic linkages between related ap-
And they rarely manage data apart from transactions. plications cumbersome. Many IT professionals are
Each new system defines its own data. Firms in the quite adept at making disparate systems look inte-
application silo stage rarely think of an enterprise ar- grated, but the code required to link applications be-
chitecture. Business users focus on the value of their comes increasingly complex and ultimately extends a
applications—occasionally complaining about high new system’s time to market. A systems manager at
operations costs. For the most part, the IT unit is re- an investment bank commented on the complexity of
sponsible for systems implementations, although the code required to link applications: “Everything we
business users are usually expected to generate the do revolves around straight-through processing with
benefits. no manual intervention, but it’s a miracle our systems
work.”
Benefits and Risks of the Application Silo Architec-
ture Stage. The strategic goal of the application silo The Standardized Technology
stage is local optimization. Organizationally, applica- Architecture Stage
tions align naturally with the firm’s functional or geo-
graphic structures. The architecture encourages inno- The standardized technology architecture stage is the
vation, because it imposes no constraints on develop- most common among the forty firms in my sample.
ment. Developers can satisfy end users by pursuing These firms have shifted resources from application
full functionality without regard to other applications development (in their application silo stage) into a
or organizational units. Consequently, functional, shared infrastructure. (See Figure 1 for a graphic of
plant, and geographic managers often respond posi- how IT resources are allocated during each stage.)
tively to applications developed in their silos. Applica- They have established technology standards to limit
tion silos can compete for capital funding using simple technology choice and reduce the number of platforms
cost-benefit analyses. System benefits are predictable they manage. This standardization also significantly
(albeit frequently overstated), and outcomes are meas- reduces the number of vendor packages that perform
urable. similar functions. For example, one firm reduced the
12
number of order management systems from 28 to 4.
Based on the characteristics described here, I assigned each of our 40 Technology standardization, however, does not over-
case sites to an architecture stage. My classification put 8 firms in
the application silo stage, 22 in the standardized technology stage, come the application silo problem of application-
and 10 in data rationalization. These results might be slightly specific data. Firms in this stage do introduce data
skewed to the right relative to the general population. Discussions warehouses to share access to data, but the transaction
with consultants in early 2003 suggest that 80-90% of firms are in
either the application silo or standardized technology stages. I did data is still embedded in the individual applications.
not attempt to secure a representative sample in my selection of
firms because my goal was to learn the practices of leading-edge
firms.

© 2003 University of Minnesota MIS Quarterly Executive Vol. 2 No. 1 / March 2003 35
Ross l Strategic IT Architecture Competency

Figure 1: Changing Resource Allocations Across Architecture Stages

Strategic Implications of IT
Local/Functional IT Process Strategic
Optimization Efficiency Optimization Choices
Local Applications
Non-core Customization
Business Needs

Local Knowledge
Specific Worker Support
Business Infrastructure
Wired
Needs
Business
Core Process Core
Integration

Technology
Standardization
Data
Shared Product Accessible Core
Data Center &/or Customer Data Objects
Data Warehouses Data Bases
Standardized Rationalized
Application Silo Modular
Technology Data
Architecture Maturity

Most IT organizations in this standardized technology One IT architect noted that the standardized technol-
stage do conceptualize an enterprise architecture for ogy architecture has given him the right to say “no,”
the shared infrastructure. Even those that do not con- when a vendor product is off-standard. In the past,
ceptualize will often hire an IT architect to lead the every application request resulted in lengthy debates
standardization effort. Business managers rarely par- and fact-finding efforts between IT architects, ven-
ticipate in developing the enterprise architecture. They dors, and users. The authority to eliminate off-
defer to the IT organization to set the technology poli- standard technologies greatly simplifies the lives of
cies and standards. However, senior business manag- architects and developers.
ers are anxious for IT cost savings, so they support the
CIO’s efforts to standardize and centralize infrastruc- A key risk of the standardized technology stage,
ture technologies by mandating compliance with the though, is managerial resistance to both the concept of
technology standards. standards and, in some cases, the dictatorial approach
used to implement standards. In the silo stage, busi-
Benefits and Risks of the Standardized Technology ness people would never allow IT’s concerns to con-
Architecture Stage. The goal of the standardized strain their business solutions. The migration to a
technology architecture stage is IT efficiency. Firms standardized technology architecture fundamentally
often move to this stage because senior management changes firms’ approaches to solutions delivery. In-
believes the IT costs have gotten out of line. Standard- stead of defining the solution and looking for the best
izing and consolidating technology platforms can lead technology, firms in this stage negotiate the best pos-
to significant cost savings. Several IT managers re- sible solution among the acceptable technology plat-
ported up to 20 percent reductions in their IT opera- forms. At first, business unit managers and developers
tions costs. Furthermore, by reducing complexity, cling to the belief that business needs should drive
technology standardization also increases IT main- technology choices. For many, their initial encounter
tainability, reliability, and security. with technology standards is the first time manage-

36 MIS Quarterly Executive Vol. 2 No. 1 / March 2003 © 2003 University of Minnesota
 Ross l Strategic IT Architecture Competency

ment allows IT, in any way, to shape business deci- algorithms. Business units have different needs for
sions. shared infrastructure at different points in time. Most
firms want to link funding to value received, but rela-
As the benefits become apparent, though, most man- tive value provided by a shared infrastructure is diffi-
agers became resigned to the standards. One CIO cult to assess.
noted that in leading the charge to standardization, he
won over business unit leaders by demonstrating cost Managers in firms with standardized technology archi-
savings: tectures have found that, after a few years,14 early bat-
tles and mistakes are forgotten. People no longer ques-
We’ve had successes where we’ve been able tion the commitment to standards or the sharing of
to reduce people’s costs by bringing in stan- infrastructure. This evolution positions firms for the
dardization. That has given us credibility. data rationalization stage, where standardization prac-
Their jaws hit the table when they see the tices expand to incorporate data and sometimes even
impact of standardization on their bottom business processes.
line.
The Rationalized Data Architecture
Although some firms justify technology standards Stage
based on economic arguments, most firms rely on a
senior management mandate that empowers the CIO Firms that have learned how to manage a standardized
to establish and enforce technology standards. This set of infrastructure services are positioned to apply
managerial approach sometimes comes as a culture similar discipline to their core data and processes.
shock, as one CIO described: Data rationalization refers to distinguishing the subset
of the firm’s data that must be unfailingly timely and
I come in and I pull every IT resource out of accurate for the firm to consistently meet customer
every closet and every back room. Then I demands. Standardizing and integrating that critical
centralize it. I tell them ahead of time this is data subset stabilizes the firm’s core activities and
going to hurt, but it will only be temporary. I increases predictability of outcomes.
have got to bring it all into a central loca-
tion, look at it, and figure out what is there, In the data rationalization stage, resources are shifted
what they do, and why they are doing it. away from application development and into data
That takes about a year and a half. management and infrastructure development. Data
management resources are used to develop centralized
The standardized technology stage also introduces data stores for the data that powers core activities. As
risks in managing the new standardized environment. a rule, data embedded in application silos must be ex-
For example, firms need a process for recognizing tracted and made easily accessible to all activities that
when a business need justifies an exception to a stan- depend on it. Numerous tools support data rationaliza-
dard. In addition, IT managers need to monitor and tion, including middleware, ERP, CRM, and internally
periodically upgrade standards to avoid obsolescence. developed customer information files or product files,
They have to clarify the costs and benefits of these such as UPS’ package level data file. These tools also
upgrades because business unit managers do not want make data available to the applications that need it.
to pay to replace something that already works.
Infrastructure activities in the rationalized data stage
Firms in the standardized technology stage also face involve integrating the activities of the firm’s core
new challenges in investment decisions. Funding for a processes. To retain data integrity while integrating
shared infrastructure demands that management con- core process activities, most firms introduce some
sider investments with longer payback periods than process standardization15 because when carried out
other classes of applications.13 Infrastructure devel-
opment costs lead to frequent debates over funding 14
Managers’ estimates varied from 2 to 6 years for the elapsed time
between starting technology standardization and fully absorbing the
13
Weill, P. and Broadbent, M., Leveraging the New Infrastructure, change culturally and technically in their firm.
15
Harvard Business School Press, Boston, Massachusetts, 1998, lists Firms can standardize processes by implementing a single instance of
four classes of IT investment with differing payback periods. Infra- an application or by implementing multiple instances of the same
structure investments had a slower payback than the three applica- application. Similarly, while some manufacturing firms may install
tion classes. See also Ross, J. and Beath, C.M., “Beyond the Busi- an ERP package to standardize the entire order-to-cash process,
ness Case: New Approaches to IT Investment,” Sloan Management other firms may rely on one module of a customer relationship man-
Review (43:2), Winter 2002, pp. 51-59 for a discussion of the differ- agement system to limit process standardization to a small set of
ent approaches to applications and infrastructure investment deci- functions in their call center. Differing needs determine the extent of
sions. standardization. See Davenport, T.H., Process Innovation: Reengi-

© 2003 University of Minnesota MIS Quarterly Executive Vol. 2 No. 1 / March 2003 37
Ross l Strategic IT Architecture Competency

effectively, process standardization ensures the quality then the process is better suited for customization than
of the central data stores. As long as the data is reli- standardization. Second, someone must be account-
able, core process activities become predictable—a able for the data. The predictability of core processes
given input always generates the same output. The depends entirely on the quality of the data, so ac-
business rules for core processes are thus in effect countability for data must rest with persons who are in
“wired” into the firm’s infrastructure. They ensure a position to ensure disciplined processes and data
consistency in customer response and become the ba- monitoring—i.e, business, not IT, people. One IT
sis for future innovation. leader described how data ownership shifted with the
implementation of ERP:
Appropriately defining the firm’s core processes is
thus critical to creating an effective rationalized data Historically within the company, we in IT
architecture. Correctly “wiring the core” requires a owned the data. Not only were we custodi-
dialog between IT and senior business managers to ans of it, we were the ones who got beat up
ensure that the core processes are indeed central to the whenever it got corrupted. In today's busi-
organization and that the business rules are stated de- ness process environment, the business own-
finitively. At a high level, managers can come to con- ers and the process owners are learning that
sensus on core processes fairly easily. they really own the data. We in IT can as-
sist them, we can make it available, and we
For example, at both Air Products and Nestle USA, can set some standards, but they own the
management decided to wire the supply chain using data. There has been little argument on this.
ERP. Both firms defined their business rules and built Once we organized around business proc-
those rules into the ERP package. The vendor set up esses, they started to realize and recognize
the package so that data flowed automatically to re- the value of data and the importance that
lated processes. Citibank Asia Pacific established a data plays in this company.
core set of standard personal and commercial banking
processes that could be replicated as new banks were Firms that moved into this third stage put a stake in
acquired or opened. the ground with regard to defining the business. Once
wired, changing the business core has become
In defining a core process, management must deter- harder—nearly impossible—while building on that
mine which activities are included and which are not. essence to create new products and services has be-
Management must also be explicit about the data these come easier and faster. Schneider National, a large
activities rely on, and how they share that data. For trucking firm, wired its operations from dispatch to
example, at Delta Air Lines, management determined delivery. In doing so, the firm can readily add new
that the firm’s two core processes were customer ex- services, such as satellite tracking and advance cus-
perience (reservations, gate check-in, seat assignment, tomer notification. But this wiring did not position
baggage claim, boarding, rewards miles, etc.) and air- Schneider to enter the logistics business. Fully aware
line operations (allocating resources, loading, flight of what its architecture did and did not support,
departure, flight arrival, unloading aircraft, and clean- Schneider designed a parallel IT architecture when it
ing aircraft). These two processes relied on 9 data- added a logistics business.
bases: customer, maintenance, location, schedule,
equipment, employee, aircraft, supplies, and ticket. Benefits and Risks of the Rationalized Data Archi-
Delta created a layer of middleware that provided a tecture Stage. The rationalized data stage provides
publish-and-subscribe environment, so that when a significant business process efficiencies, but the pri-
piece of data changed, a message about that change mary objective of a rationalized data architecture is
was sent to all applications that “subscribed” to know- process optimization. In particular, implementing a
ing the change. The result was single sources of data rationalized data architecture involves embedding core
for many systems, which meant that all systems were processes in the firm’s IT infrastructure. They become
simultaneously updated. part of the definition of the firm. In exchange for tak-
ing the risk of “permanently” wiring core process,
Two factors limit which processes and data belong in firms gain a platform positioned for innovation.
the core. First, the business rules for a core process
must be rigid. If the firm envisions regular exceptions, Extracting data from a firm’s legacy applications is a
nontrivial technical challenge, and thus a risk. More
overwhelming still are the management challenges of
neering Work through Information Technology, Harvard Business adopting data and process standardization. First, man-
School Press, Boston, MA, 1993.

38 MIS Quarterly Executive Vol. 2 No. 1 / March 2003 © 2003 University of Minnesota
 Ross l Strategic IT Architecture Competency

agement must clearly explain the concept of “core” as each other to do 3 or 4 things very, very well
well as the data driving that core. Firms that “get it and not try to do 400 things.
right” can then move faster than their competitors.
Dell’s business model offers an example of a firm that This focus on a small set of priorities is critical be-
wired its core process to how it planned to do busi- cause IT architectural changes demand significant
ness. Many fallen dot-coms did not. managerial resources. Senior management commit-
ment to the change effort is important to avoid confu-
A second risk is an implementation risk. A rational- sion and resistance. Regardless of the level of com-
ized data architecture requires disciplined processes mitment, though, change management is always a dif-
and a strong central organization. In many cases, the ficult task.17
data architecture dictates a fundamentally different
way of working. At many firms, standardizing core Senior and IT business managers in firms in the ra-
business processes has involved ripping away control tionalized data stage have learned how to articulate
of business processes from local business unit leaders. strategy to define IT capabilities and how to identify
Business process standardization is thus much harder strategic opportunities the IT capabilities create. The
to sell to local managers than technology standardiza- architecture allows the firm to protect its core proc-
tion. One CIO described the discomfort of installing esses while identifying opportunities to leverage those
ERP and accompanying process standards: processes. Leveraging a rationalized data architecture
eventually involves moving to a more modular archi-
It became very clear that what it was going tecture.
to take to do this [implement ERP] was ba-
sically senior executives telling the busi- The Modular Architecture Stage
nesses they were going to do this. That was
a huge culture change. We're taking the keys The rationalized data architecture enables process op-
to the car away from you so we can go build timization through judiciously applied data and proc-
this new car. ess standardization. The modular architecture, on the
other hand, enables strategic agility through custom-
This CIO said process standardization was “the most ized or reusable modules. These modules extend the
top-down thing we’ve ever done in this organization.” core processes, which have been wired into the infra-
structure during the rationalized data stage. Citibank
Another implementation risk is the risk of taking on Asia Pacific is nearing the modular stage. Its rational-
more change than the organization can absorb. For ized data architecture allows it to quickly implement
example, one manager whose firm attempted to jump its core products (e.g. checking accounts, personal
from an application silo to a rationalized data architec- loans, and credit cards) in banks that it opens in new
ture through an 18-month ERP implementation said countries. Over time, Citi will be able to provide al-
his firm found the organizational change requirements ternative channels and interfaces to its back-end proc-
overwhelming. “We have exceeded our capacity for esses through reusable or locally developed front-end
change,” he explained. Firms that move too aggres- modules.18
sively to change technology and organizational habits
greatly increase their risk of failure.16 Although no firms in this study have migrated to the
modular stage, those with rationalized data architec-
To implement change of this magnitude, senior man- tures provide insights into the promise they see in
agement must establish priorities for transforming the modularity. Top management has seized control of
IT architecture. Successful efforts usually involve lev- data and core processes from local managers and has
eraging existing capabilities while building the most
critical new capabilities. Successful firms focus on a
17
small set of priorities. As one senior executive de- Brown, C.V. and Vessey, I., “Managing the Next Wave of Enterprise
Systems: Leveraging Lessons from ERP Projects,” MISQ Executive
scribed her firm’s transformation: (2:1), 2003, describes the change effort involved in ERP implemen-
tation. ERP implementation is a good example of the change re-
We [senior management] as a team were quired in any implementation of a rationalized data architecture. See
committed to the vision and committed to also Robey, D., Ross, J.W. and Boudreau, M-C., “Learning to Im-
plement Enterprise Systems: An Exploratory Study of the Dialectics
of Change,” Journal of Management Information Systems (19:1),
16
Cohen, W. and Levinthal, D., “Absorptive Capacity: A New Perspec- Summer 2002, pp. 17-45, for a description of the learning challenges
tive on Learning and Innovation,” Administrative Science Quarterly associated with ERP.
18
(35:1), 1990, pp.128-152, describe the concept of organizational ca- See the full case write-up: Brand, A. Weill, P., Soh, C. and Perisamy,
pacity for change. The authors note that an “… organization needs P., “Citibank-Asia Pacific: Positioning IT as a Strategic Resource,”
prior related knowledge to assimilate new knowledge” (p. 129). Melbourne Business School, 1999.

© 2003 University of Minnesota MIS Quarterly Executive Vol. 2 No. 1 / March 2003 39
Ross l Strategic IT Architecture Competency

defined and standardized the core of the business. reuse modules that extend that core. Reusable modules
With a wired core, IT management can provide agility will build a thicker, denser core, providing greater
in two ways. efficiencies while allowing local customization. Cus-
tom modules will allow experiments to respond to
The first is to create reusable modules and allow busi- changing market conditions. In the modular stage,
ness units to select customer-oriented processes from firms will almost certainly reuse their expertise in
a menu of options. Web services, for example, offer process, data, and technology standardization.
modules as reusable business services.19 Firms will be
able to select Web services modules from internal and The greatest risk in the modular stage is that firms will
external sources. rush to introduce modules before they have com-
pletely rationalized their core data. Modules can re-
The second way to provide agility is to give business store the autonomy and innovativeness of the applica-
units greater discretion in their local processes, as long tion silo stage. But without a solid process base, mod-
as they can connect to the wired core processes. One ules run the risk of also restoring the anarchy of hun-
financial services institution is experimenting with this dreds of unmanaged applications.
model. To ensure reliable, low-cost processes, man-
agement specified systems to support some back-
office processing. However, management is encourag- How Organizational Competencies
ing local development of new process support sys-
tems, anticipating that some new modules may even-
Change Across the Architecture
tually be offered firm-wide. Teaming a standardized Stages
core with localized development can provide rapid
innovation and perhaps lead to reuse of locally- Figure 2 provides a summary of the characteristics and
developed modules. types of learning in each architecture stage. Each stage
defines a dramatically different relationship between
Modules allow for local customization, but they do IT and business executives and between IT architec-
not reduce the need for standardization. The enterprise ture and business strategy.
architecture for a modular architecture would require
an ongoing dialog between senior management and IT The application silo architecture stage allows an
executives to clarify which processes have one stan- arms-length relationship between IT architecture and
dard and are required, which processes have multiple business strategy. Business people can define strategy
standard versions to provide local choice, and which without IT input, and IT can deliver solutions without
processes may be developed in the business units for understanding the business strategy. A firm can gener-
maximum local flexibility. To continue to provide all ate value in the application silo stage and position it-
the benefits of standardization in the rationalized data self for subsequent stages by starting a dialog on IT
stage—efficiency, single face to the customer, and value between IT and business managers. Specifically,
process integration—modularity architectures extend, IT and business managers can jointly estimate, meas-
rather than replace, rationalized data architectures. ure, and communicate the value of IT-enabled busi-
Because few firms have adopted rationalized data ar- ness processes. To institutionalize this learning, man-
chitectures and because these data architectures are agement formalizes the use of business cases and post-
difficult to master, only a few modular architectures implementation reviews.
may be built in the next few years.
In the standardized technology architecture stage,
Benefits and Risks of the Modular Architecture business and IT managers build on their ability to
Stage. Modular architectures create the opportunity communicate IT value. They make decisions on IT
for strategic agility. By ensuring the predictability of standards based on their negotiated understanding of
core processes, they leverage the firm’s distinctive- the impact of IT on business strategy. They also nego-
ness. By enabling local customization, they encourage tiate funding models for the shared infrastructure, in-
innovation and customer responsiveness. But to bene- cluding replacing and upgrading technologies before
fit from modular architectures, firms need to learn they become obsolete. At this stage, IT and business
how to quickly identify the strategic opportunities that managers develop governance structures, such as ex-
best leverage their core and then quickly develop or ecutive committees, to formalize funding for the
shared infrastructure, both for new infrastructure de-
velopment and replacement (refresh). The executive
19
For a description of the potential of Web services, see Hagel, J. and committee also debates the appropriate organizational
Brown, J.S., “Your Next IT Strategy,” Harvard Business Review,
October 2001, pp. 105-113. level for IT standards. Architecture committees, typi-

40 MIS Quarterly Executive Vol. 2 No. 1 / March 2003 © 2003 University of Minnesota
 Ross l Strategic IT Architecture Competency

Figure 2: Characteristics of the Architecture Stages

Application Standardized Rationalized

Modular
Silo Technology Data
IT applications Firm-wide IT focused on Modules enable
serve isolated technology wiring core business model
IT Capability
business needs standards process extensions

Technology- Standardization Recognizing Practices

Key
enabled change and exception essence of the facilitating
Management
management management, business reusability
Innovation
refresh
ROI of Reduced IT costs; Improved Speed to market;
Business Case applications interoperability business strategic agility
for IT performance;
integration
Local control Senior Senior Senior mgmt, IT,
Locus of management management, IT, process, and local
Control support of CIO and process leadership
leadership
Estimate, Establish (local/ Determine core Define boundaries
Key measure, regional/ global) processes and for business
Governance communicate standard setting, funding priorities experiments
Issues value exception &
funding processes

cally populated with IT people, establish processes for introduce new governance mechanisms to encourage
developing, monitoring, and granting exceptions to component reuse, and they retain governance mecha-
standards. These governance structures enhance learn- nisms that support funding, standardization, and IT
ing by formalizing the negotiation process. value assessment.

In the rationalized data architecture stage, negotia- Figure 3 shows how governance mechanisms accumu-
tions become more sophisticated. IT capabilities late as firms advance through the four stages. These
shape, as well as respond to, business strategy. IT and governance mechanisms create and track organiza-
business managers clarify strategic intent, critical IT tional readiness for the various stages. Technologies,
capabilities, and the target enterprise architecture. such as middleware, Web services, and enterprise sys-
These discussions eventually produce consensus on tems allow firms to replace huge sections of their ex-
the firm’s core processes and the data that drives isting uncoordinated architecture with a more capable
them. The executive committee continues to address architecture. Although firms have transformed their
strategic IT prioritization and investment issues. The architectures, they have not been able to benefit from
firm applies its learning about standards to standardiz- new technical capabilities until their managerial capa-
ing data and processes. Management institutionalizes bilities catch up. In fact, one CIO noted that his firm
learning in governance mechanisms, such as project had attempted to leapfrog from an application silo
prioritization processes. architecture to a rationalized data architecture by in-
stalling an ERP system. They failed—twice. Two
Finally, the modular architecture stage introduces the other firms skipped from application silos to rational-
challenges of componentization, customization, stra- ized data architectures but took many years to receive
tegic experiments, and reuse. In this stage, IT and payback on their IT investment. The managers were
business management apply learning from the earlier not able to internalize the goals and objectives of the
stages to discussions about strategic direction. They firm’s architecture-related activities.

© 2003 University of Minnesota MIS Quarterly Executive Vol. 2 No. 1 / March 2003 41
Ross l Strategic IT Architecture Competency

Figure 3: Key IT Governance and Management Mechanisms

Strategic Implications of IT
Functional IT Process Strategic
Optimization Efficiency Optimization Choices
Business cases
Post implementa-
tion reviews Executive committee
Architecture exception
process
Centralized
infrastructure funding Mechanisms
Infrastructure refresh for Designing
process & Protecting
Process owners Architecture
Project prioritization
process
Data standardization
Component funding
process
Component tracking
process
Standardized Rationalized
Application Silo Modular
Technology Data
Architecture Maturity

all of a firm’s business processes will almost certainly

Lessons for Creating a Strategic stall. Focus on just the core ones.
IT Architecture Competency Second, don’t skip or rush through stages. Skipping
stages leads to either failures or delayed benefits.
Creating a strategic IT architecture competency is a
Firms benefit more from making improvements in
long, difficult process. It involves ongoing negotia-
their existing stage than from transformational efforts
tions about a firm’s business strategy and about how
that abruptly move them into foreign waters.
IT both shapes and responds to that strategy. It also
involves defining a target technology architecture (i.e. Third, recognize that complex organizations have
applications, data, and infrastructure technology) and multiple architectures, which may be at different
doggedly pursuing that architecture even when imme- stages. Like J&J, a firm may have a corporate archi-
diate business needs argue for leniency. Another step tecture, as well as divisional architectures and busi-
in building out a strategic IT architecture involves ness unit architectures. Because these different archi-
identifying the stage that best defines the firm’s archi- tectures have different objectives, they will likely be
tectural competency. A firm will want to excel both at different stages.
technically and managerially in that stage. Then, in
many, though not all cases, the firm will want to start Fourth, institutionalize learning about architecture in
working its way to the next architecture level. appropriate governance mechanisms. Governance
manages the architecture for value. Without govern-
The architecture stages model offers a number of les- ance, executives may find themselves with an expen-
sons: sive, but limiting, technology base.
First, focus architecture efforts on key business proc- Fifth, continue the dialog. Enterprise IT architecture is
esses. Architecture exercises that attempt to establish never complete. Ongoing dialog enables management
links among applications, data, and infrastructure for to continuously zero in on what matters. New people

42 MIS Quarterly Executive Vol. 2 No. 1 / March 2003 © 2003 University of Minnesota
 Ross l Strategic IT Architecture Competency

need to understand strategy and IT alignment. Markets

change. Learning and architecture atrophy if ignored.
Any firm can lose its strategic edge if management
does not regularly check its assumptions about its IT
capabilities.

Sixth, keep an architecture capability in-house. The

architecture task is difficult enough that management
might want help. But the negotiations that lead to un-
derstanding the links between business strategy and IT
architecture require a close working relationship be-
tween business and IT.20

The payback for enterprise IT architecture efforts is

strategic alignment between IT and the business.
Alignment will generate a higher return on the firm’s
IT investments and focus the firm’s project portfolio
on initiatives likely to have strategic impact. Ulti-
mately, enterprise architecture leads to “happy sur-
prises.”

About the Author

Dr. Jeanne W. Ross is Principal Research Scientist at
MIT’s Center for Information Systems Research
(CISR) where she lectures, conducts research, and
teaches executive education courses on IT manage-
ment practices. Her work has appeared in Sloan Man-
agement Review, Harvard Business Review, Journal
of Management Information Systems, and MIS Quar-
terly. She is a former associate editor of MIS Quar-
terly and serves as a founding senior editor of MISQ
Executive ([email protected], web.mit.edu/cisr/www/).

20
On the other hand, a number of firms had outsourced significant
numbers of IT services with no apparent impact on the quality of the
architecture.

© 2003 University of Minnesota MIS Quarterly Executive Vol. 2 No. 1 / March 2003 43