1. Information & Management 41 (2004) 431–443
Understanding the local-level costs and benefits of
ERP through organizational information processing theory
Thomas F. Gattikera,*, Dale L. Goodhueb
a
Richard T. Farmer School of Business Administration, Miami University, Oxford, OH 45056, USA
b
Terry College of Business, University of Georgia, Athens, GA, USA
Received 6 February 2002; received in revised form 13 March 2003; accepted 27 June 2003
Abstract
Using organizational information processing theory (OIPT), we suggest several factors that influence some of the enterprise
resource planning (ERP) costs and benefits that organizations are experiencing. Though we do not attempt to address all
important factors that contribute to an ERPs impact, we suggest two organizational characteristics that may have received
insufficient attention in other ERP literature: interdependence and differentiation. High interdependence among organizational
sub-units, contributes to the positive ERP-related effects because of ERPs ability to coordinate activities and facilitate
information flows. However, when differentiation among sub-units is high, organizations may incur ERP-related compromise or
design costs. We provide a case study that explores the viability of this framework. The case describes some local-level impacts
of ERP and provides some evidence of the validity of the model. Unexpected findings are also presented.
# 2003 Elsevier B.V. All rights reserved.
Keywords: Enterprise resource planning; ERP; Enterprise system; Standardization; Integration; Interdependence; Differentiation;
Manufacturing planning and control systems; Organizational information processing theory
1. Introduction additional complementary explanation for why some
ERP implementations are more successful than others.
Companies have experienced mixed results when Our IP-based framework suggests that, depending
using enterprise resource planning (ERP) systems: upon the amount of interdependence and differentia-
along with some inspiring success stories come a con- tion among sub-units of an organization, ERP systems
siderable number of conspicuous failures. In between may ‘‘fit’’ some organizations better than others. To
these extremes are many ERP implementations that are explore these ideas further and to see whether these
partly successful and partly problematic. While much constructs make sense in the context of an actual
of the literature focuses on implementation practices company, we conducted a case study of a manufactur-
as the critical determinant of success, this paper uses ing organization involved in a multi-plant ERP imple-
several constructs from organizational information mentation. The investigation suggests that OIPT is a
processing theory (OIPT) as a basis for exploring an worthwhile lens for understanding ERP systems, and
the case study provides evidence that the concepts
*
Corresponding author. Tel.: þ1-513-529-8013;
of interdependence and, especially, differentiation
fax: þ1-513-529-2342. may be important in predicting the impact of ERP,
E-mail address: gattiktf@muohio.edu (T.F. Gattiker). particularly at the local level.
0378-7206/$ – see front matter # 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0378-7206(03)00082-X
2. 432 T.F. Gattiker, D.L. Goodhue / Information & Management 41 (2004) 431–443
The goals of our research were to develop some rich uncertainty as a lack of information about statuses of
descriptions of ERP impacts at the local level and, tasks, the environment, and so on [10,12]. The amount
in particular, to refine our conceptualization of dif- and types of uncertainty vary across organizations and
ferentiation and the ways in which it appears to among individual sub-units within organizations.
affect ERP systems within organizations. Differences Numerous modes of coordination exist and these
among sub-units can be assessed only by examining differ in the degree to which they are suited for coping
them. Therefore, this paper focused on lower levels with various types and amounts of uncertainty. For
of the organization, specifically, on the manufacturing example, the theoretical literature states that mechan-
facility. isms, such as hierarchical referral and standard oper-
We acknowledge that concentrating on a limited ating procedures, are appropriate when uncertainty is
view of the organization misses important ERP costs low, while computerized IS and lateral relations are
and benefits that occur at the organization-wide level. better choices in high uncertainty situations. There-
However, we also note that a great deal of the existing fore, in order to prosper, the organization must match
discussion of ERP focuses at that level and perhaps the appropriate mode(s) of coordination with its par-
under-represents local-level costs and benefits. ticular uncertainties [11].
It follows that ERP, being a type of computerized
information system, is a coordination mechanism
2. Why ERP? that is appropriate under many circumstances but
less so under others. A recent study [9] examined
ERP systems are highly standardized systems. They the impact of various uncertainty reduction mechan-
typically employ a single logical database for the isms within manufacturing plants. Interestingly, IS
entire enterprise. This feature requires data standards were the only mechanism that did not significantly
across the enterprise [1,5,7]. ERP implementations moderate the negative relationship between uncer-
also entail a great deal of process standardization tainty and performance. However, the researchers
[13,22]. This is partially a result of data standardi- examined IT investment in the aggregate, rather than
zation. Furthermore, managers in many firms view considering potentially differing effects of different
ERP-driven reengineering of business processes to types of computerized systems (e.g. computer inte-
company-wide standards as beneficial from a business grated manufacturing, ERP, advanced planning and
standpoint [20]. Finally, ERP systems themselves scheduling software, etc.). Since these technologies
are limited in the processes that they can model. differ so widely, their impacts on performance prob-
For example, one estimate is that 20% of the typical ably differ. Therefore, it may be more revealing to
company’s legacy processes cannot be modeled in narrow the focus to a particular technology, such as
SAP [23]. As a result, an organization is sometimes ERP.
limited to the collection of options imbedded in the Information processing theorists have suggested
package or to modifying the business practice and not various sources or types of uncertainty, including:
the ERP code [3,29]. the characteristics of the self-contained tasks that
Integration—the linking together of the information sub-units must execute, instability of the external
and processes of distinct subsets of the organization— environment, interdependence with other sub-units
is one benefit of standardization. Indeed, ERP systems [27], and differentiation among sub-units [6]. Because
link all (or many) business functions and operating integration and standardization are two major char-
locations so all have access to relevant information as acteristics of ERP, we focus on the two sources of
transactions occur. uncertainty that are most related to them: interdepen-
dence and differentiation. Theory suggests that greater
interdependence among organizational sub-units is
3. Organizational information processing theory associated with greater benefits from ERP. On the
other hand, differentiation among organizational sub-
OIPT posits that resolving uncertainty is the central units can lead to some significant ERP-related costs.
task in organizational design. The theory conceptualizes This framework is depicted in Fig. 1.
3. T.F. Gattiker, D.L. Goodhue / Information & Management 41 (2004) 431–443 433
Organizational Information Processing Theory Other Theories
Impacts of ERP:
Interdependence
with other sub- Benefits from
units improvements in
coordination
with other sub-
units
Net impact of Other costs Other factors
ERP and benefits
Design and
compromise
Differentiation costs
from other sub-
units
Fig. 1. Impact of ERP on a sub-unit in an organization.
3.1. Interdependence influences ERP benefits 3.2. Differentiation influences ERP costs
OIPT posits that the benefits of a highly integrated, On the other hand, OIPT predicts that the costs of a
standardized system, such as ERP, are influenced standardized system, such as ERP, increase in propor-
positively by the level of interdependence among tion to the degree of sub-unit differentiation, which is
organizational sub-units [30]. Interdependence is the the uniqueness of tasks, technologies, environment,
degree to which sub-units must exchange information goals, etc. across sub-units [19]. According to a model
or material in order to complete their tasks [21]. posited by Goodhue et al. [14], when an integrated IS
Interdependence increases the need for ‘‘mutual is implemented across a number of differentiated sub-
adjustment’’ and decreases the degree to which activ- units, two types of costs result: design and compromise
ities can be pre-planned. When interdependence is costs as shown in Fig. 2. Designing a single common
low, simple coordination modes, like standard operat- language (or a single common system) to accommodate
ing procedures, suffice. By contrast, high interdepen- the diverse needs of sub-units that differ greatly from
dence increases the need for a common formalized one another is a difficult and expensive task, compared
language in order to enable the exchange of informa- to designing a system for a collection of relatively
tion among sub-units [26]. homogeneous sub-units. These design costs include
Managing interdependence and improving the flow the time and money spent developing common under-
of information across sub-units is a major reason many standings across very different points of view, creating
firms have implemented ERP [4]. If sub-units of a complex IS designs, and possible database structure
firm need to exchange information and materials, then redesign. On the other hand, when standardized systems
ERP should facilitate this flow. After all, data standards are implemented across differentiated sub-units without
eliminate the burden of reconciling or translating infor- such complex designs, some sub-units will experience
mation that is inconsistently defined across two or more compromise costs. These costs include decreased opera-
sub-units [18]. Data standards also do away with the tional performance or decreased data relevance due to
potential for translation or reconciliation errors as well one or more sub-units’ having to use an IS that is not
as ambiguity about a field’s true meaning [24]. Finally, well-tailored to the peculiarities of the tasks that the sub-
ERP improves the timeliness of information. unit must execute. Hong and Kim [17] found that the fit
4. 434 T.F. Gattiker, D.L. Goodhue / Information & Management 41 (2004) 431–443
Compromise Costs:
Lack of flexibility to
respond to unique needs of
individual sub-units
Integrated data and
applications implemented
across one or more
differentiated sub-units
Design Costs:
Costs of designing and
implementing sophisticated
global information systems
that meet unique individual
sub-unit local needs
Fig. 2. Differentiation, compromise costs and design costs.
of an organization’s ERP system with its task, data, and cialized, level of automation) that plants employ
related needs explains a significant amount of the internally should be consistent with these output
company’s ERP success. Our model extends this by characteristics [15,16]. Furthermore, to be successful,
suggesting that differentiation impacts fit. manufacturers must match these characteristics with
A recent study of small businesses found that the the configuration of their computerized planning and
greater the variety of functions within a business, the control systems, including ERP [2]. For example, job
less the effect of IT adoption on the amount and quality and batch approaches require features like detailed
of information made available to organization mem- order tracking, while these features impede produc-
bers [8]. One explanation is that greater functional tivity for repetitive and process manufacturers. Pro-
variety creates greater variety in information needs cessors often require the capability to maintain
across sub-units and thus greater compromise costs if a multiple bills of material for a single end product
single, standard system is broadly deployed. While and to track individual lots of raw material [25].
this is only one explanation, it does add some empiri- This line of thinking has strong implications for
cal basis for investigating our framework. ERP. A particular ERP implementation imposes a
particular set of computerized planning and control
system characteristics (configuration decisions that
4. Differentiation in a manufacturing context are made when implementing ERP modules like
material requirements planning, master scheduling
Manufacturing plants often exhibit high differentia- and purchasing) on a manufacturing plant. This con-
tion, making them a good domain for studying the figuration is usually determined at the division or
impact of differentiation on ERP implementation. company-wide level. Thus a particular ERP config-
Plants within an organization often differ significantly uration may be a good fit for the majority of plants in a
and observably in products produced, technologies company but may be a bad fit for one plant (or a few
employed, and markets served. plants) that differ substantially from the others.
The manufacturing strategy literature provides sev-
eral very similar frameworks for characterizing plants
and thus for comparing them in order to assess differ- 5. Research methodology
entiation among them. These characteristics include
volume and standardization of outputs. The manufac- In order to explore these ideas, we conducted a case
turing technologies (e.g. general purpose versus spe- study at a manufacturing plant that was part of a large
5. T.F. Gattiker, D.L. Goodhue / Information & Management 41 (2004) 431–443 435
ERP implementation. We interviewed eight indivi- resources. In fact, the team spent the entire imple-
duals at the site: Plant Manager, Controller, Account- mentation budget on just four plants, and the project
ing Clerk, Purchasing Manager, Customer Service was then frozen. Shortly thereafter SAP began releas-
Representative, Master Scheduler and Production ing revisions to version 2.1, some of which allowed
Supervisor, Shipping Supervisor. Most interviews SAP to handle the three-dimensional nature of FPC’s
lasted approximately 1 h and followed a semi-struc- items. Unfortunately, having already customized the
tured format. Most were taped and transcribed. We code to meet the needs of different plants during
also toured the facility and reviewed documents, such implementation, FPC felt that installing the upgrades
as organization charts and ERP planning documents. was not feasible.
Case write-ups were reviewed by the plant manager At this point a new IS vice president was hired. One
for accuracy and amended as necessary. The company of his major charges was to supervise ERP implemen-
identity is disguised. tation. He pointed out the problems in implementing a
unique configuration of SAP at each plant and stressed
the importance of developing one business ‘‘vision’’
6. Case description: forest products corporation for the entire company prior to moving forward with
ERP. In 1996, the Blueprint Team (including the
Forest products corporation (FPC) manufactures manager of the Augusta plant) was formed to develop
construction materials—primarily engineered lumber. this vision. A group of mid- and top-level managers
FPC has approximately $1.2 billion in sales annually, was assigned to this team for 6 months.
generated from approximately 20 manufacturing The plan assumed that after the company-wide
plants, all in North America. Finished goods are held blueprint had been created, SAP 3.1 was to be imple-
at the plants and shipped directly to customers and mented in all 20 plants in the following 2 years (prior
distributors. Our focus in this case was on FPC’s to the year 2000), with the help of a different outside
Augusta plant. consultant. In contrast to the approach taken during the
In the early 1990s management felt FPC’s systems first round, FPC did not allow its plants to modify SAP
were inadequate for several reasons. Current systems code. In fact, to facilitate implementing the software
were not Y2K compliant. Furthermore, the plants were on the tight 2-year timetable, a template-based
running various systems that did not communicate. approach to implementation was used. This approach
Thus customers had to place and track orders with stressed standardization and allowed for limited con-
multiple plants. In fact, it was difficult to even find out figuration to meet individual plants’ needs.
how much inventory FPC retained of a particular item The first implementation occurred at the Augusta
without making telephone calls to many plants. plant and was completed in August 1998. After the
In 1993 a decision was made to implement a variety system was implemented, 20 office staff plus 20
of SAP modules (version 2.1) throughout the com- operating managers and associates used the SAP
pany. The modules chosen were manufacturing and system at Augusta. A shop-floor data collection sys-
materials, finance and accounting, and sales and dis- tem called FACE was also implemented company-
tribution. Access to and control of the SAP system wide at this time. FACE provided a more user friendly
were to be maintained at the corporate IT level. Most interface to upload data to SAP.
plants, including Augusta, did not have IT staff.
With the help of outside consultants, a project team 6.1. The Augusta plant
from corporate IT began implementing SAP, starting
with the Augusta plant. Because every plant was The Augusta facility produced engineered beams
considered to have different needs, the company and had four production departments. In the first, logs
implemented a different configuration of SAP at each were received, prepared, and ‘‘unrolled’’ into sheets of
plant. This included significant customization of the veneer. In the next, the veneer was dried and then
SAP code. stored in bundles. In the press department, bundles of
The approach created several unforeseen problems. veneer were cut into strips, mixed with glue and fed
The high levels of customization required considerable into a press. As material left the press it was cut into
6. 436 T.F. Gattiker, D.L. Goodhue / Information & Management 41 (2004) 431–443
15–50 ft billets. In finishing, these were cut into final were conducted via telephone, which was time con-
dimensions, and these finished beams were bundled suming and error prone. The vision was to centralize
and sent into the finished goods storage yard. customer service and order management.
There were a number of stock items (standard Three of Augusta’s four production departments
depth, width and length) but many products were were satisfied with SAP. Unlike the finishing depart-
cut to order. Lower volume sawed-to-order material ment, these three production areas were very straight-
was termed N-STOK. After being cut and bundled in forward in terms of manufacturing planning and
the finishing department, stock and non-standard (N- control and they were typical of most operations in
STOK) products were placed in finished inventory. FPC. The dryer department manager’s evaluation of
Off-cuts (leftover sections of billets) and rejects con- SAP was representative of them. He stated that the
taining some usable material were put into recovery– simplicity of their processes had much to do with their
reclaim inventory for use later. Partially complete satisfaction:
orders (partial bundles of identical items) were also
When you take a log and you peel it, and you make
stored in recovery–reclaim, since, from this point in
a stack of 4 Â 8 veneer, all SAP is doing is tracking:
the process onward, the organization only tracked
that’s worth 90 cubes, 90 cubic feet. Then when the
complete bundles.
dryer uses it, through the FACE system, we with-
The plant’s customer service department took
draw that from inventory and it minuses out and it
orders. Orders that could not be fulfilled from finished
gets added to the next stuff along the line.
goods inventory were promised against future capa-
city. Non-standard orders were sawed to customer Furthermore, SAP eliminated many reporting and
specifications (from recovery/reclaim inventory mate- data entry chores. For example, prior to the first SAP
rial or new production). implementation, the dryer department employees had
The wide variety of finished goods configurations it to calculate the volume of each bundle of veneer
produced set the Augusta plant apart from other removed from green inventory and the volume of veneer
facilities in the company. In its product line, it was entered into dry inventory. This involved measuring
the one plant that handled non-stock lengths. For bundle height and multiplying by a factor. At 100–200
example, Augusta manufactured all European orders, bundles per shift, this practice resulted in many errors
which was a significant business segment but treated and, ultimately, in poor inventory accuracy. The first
as N-STOK because of the metric sizing. SAP implementation eliminated the errors by perform-
ing the calculations automatically, and the second
6.2. Benefits from ERP implementation went even further by almost eliminat-
ing data entry through bar-coding each bundle.
According to the interviewees, the system resulted The dryer department manager stated that the latest
in several important benefits in the Augusta plant. SAP implementation of SAP was a ‘‘piece of cake,’’ and
kept perpetual inventories of finished goods and of that, ‘‘Bottom line, it’s saving us an enormous amount
some intermediate materials. These increased the of time now not having to do all those hand calcula-
plant’s ability to make customer commitments and tions.’’ However, there were some valuable reports he
to keep them. Additionally, several interviewees stated received under the previous version of SAP that the
that SAP boosted accountability and self-discipline current implementation eliminated.
within the plant. SAP provided visibility of open
production orders and it provided accurate inventories 6.3. Problems from ERP
of intermediate materials, such as veneer.
At the company level, SAP improved coordination Many problems stemmed from the poor visibility of
among plants for fulfilling customer orders. A custo- information regarding N-STOK orders and recovery–
mer service representative in one plant could view all reclaim inventory. The accuracy and understandability
other plants’ inventories using the SAP system; some of performance measurement and accounting data
plants could enter finished goods orders against were also an issues. The problems often resulted in
another’s inventory. Before SAP, both these activities the use of informal systems or ‘‘going without.’’.
7. T.F. Gattiker, D.L. Goodhue / Information & Management 41 (2004) 431–443 437
6.3.1. N-STOK part numbers was required to locate the distinct package of N-STOK
One objective of the SAP 3.1 implementation was associated with the order being picked. There could be
to enable available to promise calculations. Doing so (and often were) numerous identical packages in
without adding unacceptable levels of complexity inventory, but the picker could not know this without
(and IS excessive IS staff) required drastically redu- accessing the text description on the order; this was a
cing FPC’s number of stock model numbers. The time-consuming task. For many products this insis-
largest part of the problem came from the many stock tence on obtaining the exact bundle would be no great
model numbers used for non-standard, lower volume inconvenience. However, it greatly complicated
products. operations at the Augusta plant, because the size of
Since most plants had little actual volume of non- the items (often 15–50 ft in length) made moving them
standard products, a compromise solution was found. difficult. Furthermore because N-STOK was identified
Starting with the SAP 3.1 implementation, FPC insti- by sales order, a bundle of N-STOK material could not
tuted configure-to-order or N-STOK part numbers for easily be moved from one sales order to another. A
all low volume, non-standard products. A particular customer service representative explained:
number was used for all non-standard items with a given
It’s locked in, all the way down to that package
width and depth, regardless of length. For example,
number. You might have 9 packages out there that
all items that were 3.5 in. wide and 12 in. deep were
are 39 foot, the same size and everything, but each
assigned part number 897 686, regardless of their
one is assigned to a specific order and it has to ship
length. Since N-STOK part numbers do not indicate
with that order. So, if the customer calls and wants
length, they could not be used as a way to identify
to change his mind or he wants to take that same
exactly what had been ordered on a particular sales
package and move it to another one of his trucks,
order. The only way to determine that was to pull up the
that is chaos.
individual line item of the sales order and look at the text
description. Nor was it clear what ‘‘quantity on hand’’
or ‘‘quantity produced’’ meant for N-STOK items. 6.3.2. Recovery–reclaim inventory
Thus, a bundle of N-STOK product in the plant could Off-cuts were stored in recovery–reclaim inventory
not be fully described by its part number; the only way until they could be used on orders. Similarly, partially
to identify the configuration of product in such a bundle completed bundles of product were kept in recovery–
was to indicate its order number and line number. reclaim until they could be completed. Because of the
For most plants in FPC, this way of handling non- high variety of configurations it produced, Augusta
standard products worked well enough. However, had some of the ‘‘toughest cuts’’ in the company. This
because of its heavy volume of non-standard business, resulted in many leftover billet sections and thus a
relative to other plants, Augusta experienced problems. large recovery–reclaim inventory to manage.
For example, to create the schedule, the master sche- Since FPC’s ERP system was part-number driven,
duler needed to know what orders had been placed and the only way to track the contents of recovery–reclaim
what recovery–reclaim was available for cutting. For inventory would be to assign a part number to every
any customer order consisting of N-STOK, the SAP piece that was placed into recovery–reclaim. FPC
standard reports he received indicated the quantity, management did not consider the costs of doing so
depth, and width of the material needed, but not the to be worthwhile because of the sheer variety of
length. Discerning the length required opening up the lengths of off-cuts and rejects. Therefore, all material
appropriate line item of the specific sales orders. Thus placed into recovery–reclaim was inventoried under a
the master scheduler spent approximately 30 min per single part number and SAP only tracked the total
day looking up N-STOK requirements in sales orders. cubic feet in this inventory.
N-STOK created the same type of visibility-related This poor visibility into recovery–reclaim resulted in
problems in the customer service area also. problems for the plant. To avoid wasting usable mate-
Using sales order number as part of the identifier rial, the master scheduler needed to schedule cuts from
meant that identical items were not interchangeable, material in recovery–reclaim whenever possible; he
e.g. when physically assembling orders, the picker wanted to keep a perpetual recovery–reclaim inventory
8. 438 T.F. Gattiker, D.L. Goodhue / Information & Management 41 (2004) 431–443
in a spreadsheet; however, doing so would require though I know my way around the system, and I
a full-time clerk, which the plant could not afford. don’t trust the numbers that it gives me.
Therefore, making the best use of recovery–reclaim
Similarly, the senior plant accountant and the plant
was difficult and the master scheduler estimated that
manager had reservations about the accuracy of SAP-
working to do so added several hours to his work week.
generated reports. Some problems were a function of
The finishing department, too, needed to perform
the way FPC configured SAP. For example, because an
manual calculations in order to manage the daily
N-STOK part number did not signify length, it was
material flow into and out of recovery–reclaim and
difficult to track the cost per cubic foot produced. This
the associated reporting. This activity required about
created difficulties valuing certain business segments,
one man-day per day and the department hired a clerk
such as European sales. Other deficiencies were due to
solely to work on it. The department manager spent
errors in the logic implemented. The plant manager
60–90 min per day on such tasks when the initial,
and accountant noted many discrepancies between
customized version of SAP was installed, and approxi-
SAP-based reports and historical averages and also
mately 30 min per day using their old, totally custo-
the plant’s manually generated reports. Corporate
mized system before any ERP was installed.
investigated some of these and found the plant’s
internal figures or estimates to be correct.
6.3.3. Performance reporting
As a result, the plant manager and his subordinates
In addition to information visibility issues, there
made many decisions based on the manually gener-
was also a problem with the meaningfulness and
ated plant production report. This report drew data
accuracy of SAP reports. This manifested itself in a
from the FACE system, and from various departmental
number of different areas and was often addressed by
reports, some of which were generated from tally
maintaining alternate parallel reporting systems.
sheets on the plant floor.
For example, in the finishing department, produc-
tion and cost numbers generated by SAP were not
6.4. Complexity and understanding
useful to the department manager. The major perfor-
mance metric for the finishing department manager
Even though the senior plant accountant had years
was recovery (net finished goods production divided
of experience in diverse manufacturing environments,
by the amount of material received to be sawn into
he had difficulty in understanding how costs were
finished goods). However, because of the way the
calculated and passed on from one department to
system handled recovery–reclaim, SAP did not pro-
another in the SAP system. Furthermore, because
vide a meaningful recovery metric that the department
SAP limited his access to information and because
manager could use to gauge daily performance. The
it restricted the format into which this information
problem with the system (as implemented at FPC) was
could be organized, SAP made it more difficult to
that it only considered, as production, those completed
investigate and resolve concerns.
bundles placed into finished goods inventory; it did not
The senior plant accountant believed that eventually
count material in recovery/reclaim, including partial
the problems would be fixed (and that reports tailored
bundles even though (from the standpoint of measur-
to the needs of the plant would then be available) but
ing the department’s efficiency) they were finished
that resources to do so were scarce. However, he noted
production. As a result, yield was understated some
that, in previous systems, he would have done much of
days and overstated on others. Because the SAP daily
this customization himself. Because of ERPs integra-
production report was not managerially meaningful,
tion of plants and the associated complexity, he had to
the finishing department manager and others depended
rely instead on the corporate SAP team. He stated:
on a manually generated department production report
maintained on Excel. One department manager said: If you tinker with something, you’re tinkering with
something that 19 other plants are using, too, and
Even though SAP is what the plant’s performance you just can’t do that. Not all the plants are exactly
is based on, I don’t really use a lot of the numbers the same, so you don’t know what you’ll screw up
that we’re entering. They’re not easy to get to even somewhere else.
9. T.F. Gattiker, D.L. Goodhue / Information & Management 41 (2004) 431–443 439
7. Discussion of the case with Augusta by saying, ‘‘This plant is easier to manage
with SAP than some of the other plants . . .. At Augusta
Because many other studies have focused on global there are more exceptions and so they have more poten-
impact, we chose to look more closely at local impacts tial to stub their toe.’’ Experiences with SAP at Stockton
here. Guided by OIPT, we looked for evidence of rela- were overwhelmingly favorable. Employees did com-
tionships between interdependence and ERP benefits ment on the system’s rigidity and on the cumbersome-
and also between differentiation and ERP costs. How- ness of some of the user interfaces but SAP was clearly
ever, because of the focus on the local level, we were not creating problems of the same magnitude as those
better able to surface evidence about differentiation. at Augusta. For example, N-STOK created significant
Additionally, we sought to refine our understanding problems at Augusta, whereas Stockton employees
of differentiation in a concrete business situation. We described SAP’s handling of N-STOK as ‘‘somewhat
found evidence supporting the contention of Goodhue clunky’’ but a minor problem. The Stockton Master
et al. that integrating in the presence of differentiation Scheduler spent less than 30 min per week dealing with
results in either design costs or compromise costs. N-STOK parts and often no time at all. Furthermore,
Furthermore we developed some understanding about recovery–reclaim was not a problem, because the plant
the sources of differentiation in manufacturing organi- was able to keep this inventory to almost zero.
zations. For another perspective, we also interviewed four
individuals at company headquarters (over the tele-
phone). These quoted numerous metrics that suggest
8. Differentiation and ERP-related costs that, overall, the project was a success. For example,
capacity utilization went from 60 to 85%, the order fill
8.1. Sources of differentiation among rate improved, and the end month close cycle was
manufacturing plants reduced from 20 to 5 days. Interviewees stated that
SAP has worked well for most plants, and these
As a construct in OIPT, differentiation is often used metrics seem to confirm it. Furthermore, the inter-
in general and abstract terms; in order to apply the views indicated that the company followed a sound
construct to a specific domain, we suggested frame- methodology during the ERP configuration process.
works from the manufacturing strategy literature. A key Therefore, it would be difficult to attribute Augusta’s
in both these is the level of standardization (or variety) difficulties to configuration blunders. For example, the
of products produced. We found that standardization company utilized a high level steering committee; the
did set the Augusta plant apart from others in the project general manager was a respected former plant
company. The plant differed due to the wide variety manager and users were heavily involved. In fact,
of non-standard finished products produced. This there were many indications that teams went to great
included custom dimensions (such as those specified lengths to solicit input from all plants when making
by architects) and products for export (sold in low important decisions. However, there were also
volumes and not conforming to American standards). instances when a small number of plants would protest
As a check on the accuracy of Augusta interviewees’ a configuration decision but would be overruled.
reports about their plant, we visited another plant and Interestingly, the impact of custom products was not
conducted several interviews. We refer to this plant as felt in three of the four departments of the Augusta
Stockton. Stockton and Augusta used the same produc- plant. Operations in these three departments (green,
tion technologies and produced the same family of drying, and press) are virtually identical from plant to
products. Furthermore, both plants underwent both of plant regardless of the degree of customization of the
the company’s SAP implementations (versions 2.1 and plants’ final products. Only in the finishing department
3.1). However, while Augusta manufactured a signifi- did custom products make Augusta truly different
cant portion of non-standard items, almost 100% of from the other plants in FPC, and there they forced
Stockton’s product mix was standard material for resi- a difference in operations. It is problems that originate
dential construction. Stockton management confirmed in the finishing department that have caused the
this; the manufacturing manager contrasted his plant Augusta plant its greatest difficulty.
10. 440 T.F. Gattiker, D.L. Goodhue / Information & Management 41 (2004) 431–443
8.2. Compromise costs solutions to difficult local-level problems are extre-
mely costly. Thus even when the technical potential
OIPT predicts that ERP will impose costs when for such accommodations exists, local tailoring or
differentiation exists among entities in an enterprise, configuring requires substantial design costs—the
because it increases the likelihood that the ERP will be costs associated with meeting the idiosyncratic needs
a poor fit with the operational and informational needs of a differentiated sub-unit.
of some local units. At FPC, the finishing department In part because of the need to install ERP before
of the Augusta plant differed substantially from its Y2K and partially because of management’s imple-
peers. When faced with high levels of differentiation, mentation philosophy, FPC followed a deliberate
companies may implement sophisticated ERP config- strategy of avoiding design costs. The plant manager
urations in order to attempt to accommodate them. To at FPC’s Augusta plant acknowledged the possibility
the extent that these attempts fail or are not made, that the SAP system might be capable of meeting the
compromise costs will be incurred. Forest products did plant’s needs if it were configured differently. How-
not attempt such accommodations with respect to the ever, FPC’s IT staff concentrated on issues common
Augusta plant. Thus the plant experienced compro- to the majority of manufacturing facilities, not to the
mise costs. idiosyncratic problems plaguing a single unique
The higher end-product variety at the Augusta plant plant. Because of a desire to economize on time
resulted in the need to make more extensive use of and personnel costs, FPC decided not to attempt to
N-STOK part numbers and recovery/reclaim than its accommodate the differences of the Augusta plant.
peers. When FPC configured SAP to handle N-STOK For example, one interviewee stated:
and recovery/reclaim, their decision-makers assumed
Nobody’s going to back you up now because you’re
that non-standard cuts and non-standard products were
the only plant with a problem. The problems that
rare. This was a valid assumption for most plants but
are getting fixed are problems that are pretty similar
not for Augusta. The organization made a deliberate
throughout all plants. The problems that aren’t
decision to accept the ‘‘ungraceful’’ manner in which
getting fixed are the ones that are individual pro-
the ERP system handled N-STOK and recovery–
blems, specific to plants.
reclaim. This trade-off was worthwhile at the other
plants, because they seldom had to deal with non-
standard business. But at Augusta this solution led to 8.4. Unexpected findings: closing the gap versus
operational problems for the master scheduler and the bridging the gap
finishing manager, as well as reporting problems that
extended up to the plant manager level. The Augusta A significant gap resulted between the business
plant responded by utilizing several manual systems, processes that the plant needed to follow and the
but these were resource intensive and still did not solve business processes supported by the ERP systems,
the information accuracy problems. as implemented. This gap needed to be closed some-
how. Most of the discussion today focuses on two
8.3. Design costs and the role of resources types of responses to such misfits: (1) change the
business process or (2) modify the ERP system.
In supporting the needs of locally differentiated However, the plant we studied did not have the
sub-units, SAP and other ERP systems may well have authority or resources to modify the ERP system.
more flexibility to model a wider variety of business Moreover, changing business processes to conform
processes and conditions than was used at FPC. This to ERP standards was not viable, because these pro-
functionality often exists within the configuration cesses were dictated by relatively immutable factors:
options of the ‘‘vanilla’’ ERP. Additionally, systems the heavy use of recovery/reclaim and N-STOK was
may be customized, and/or ‘‘bolt-ons’’ may be added determined by the nature of the products that the plant
to provide desired functionality. However, sufficiently produced and the markets it served.
skilled personnel (consultants or employees) and Because neither modifying the ERP system nor
the time needed to analyze problems and potential changing business processes were options, the plant
11. T.F. Gattiker, D.L. Goodhue / Information & Management 41 (2004) 431–443 441
could not close the gap. Rather plant employees were information that ERP provides being used for high-
forced to bridge the gap instead. Gap-bridging beha- level decision making?.
viors were based on reliance on alternative IS (Excel, We made an effort to confirm our observations
paper logs) and using more labor (longer hours, hiring about the Augusta plant (especially regarding differ-
additional personnel). entiation and the impacts of ERP) by examining a
contrasting plant and by interviewing employees at
corporate. However, it remains possible that we over-
9. Interdependence and ERP benefits estimated the level of differentiation at Augusta or that
the problems are widespread throughout the company.
Following OIPT, we suggested that interdepen- One way to eliminate this concern would be to study
dence would contribute to ERP benefits. Many ERP each plant in the company. The fact that we have not
benefits which exploit interdependence, such as done so is a limitation of our work.
improved centralized decision making, manifest Case studies are useful for their exploratory value,
themselves most clearly at the global level. Because including investigating the applicability of a particular
our study focused on the plant level, we were not able theoretical framework to a specific domain. However,
to discern all corporate-wide benefits. confirming theoretically specified relationships
However, the case manifests one very significant requires confirmatory work.
locally visible benefit influenced by interdependence:
At FPC, the ERP provided better coordination among
plants by allowing customer service representatives to 11. Conclusion
see inventory across multiple plants and then take
cooperative action. The ERP thus supported a small As a collection of highly integrated subsystems,
degree of reciprocal interdependence between the ERP systems can be described as ‘‘tightly coupled’’, in
plants because each plant’s actions could be influ- contrast to the ‘‘loosely coupled’’ subsystems that
enced by their knowledge of conditions at others. make up most of the installed base of pre-ERP systems
However, there was minimal interdependence invol- in organizations. To be sure, tight coupling facilitates
ving the plants. coordination among subsystems, and this is typically
Our work was well positioned to discern both critically important in today’s competitive business
benefits and costs at the local level, and we saw some environment. On the other hand, Weick [28] argues
laudable local benefits. For example, the ERP that loosely coupled subsystems have potential advan-
improved coordination and discipline between depart- tages as well:
ments within the Augusta plant and it automated some
They adapt more easily to unique local require-
manual tasks. We do not claim that interdependence is
ments.
the only predictor of all ERP benefits, and FPC’s gains
They may provide a more sensitive boundary sen-
are certainly cases in point.
sing mechanism.
They provide more room for innovation and adap-
tation to environmental change by accommodating
10. Limitations of this study
more variety and novelty.
A ‘‘breakdown’’ in one subsystem is not automa-
This paper presents a model that uses interdepen-
tically transmitted to others.
dence and differentiation as predictors of some impor-
Finally, since they interact little with one another,
tant costs and benefits of ERP. Understanding
changing them requires fewer resources.
differentiation among sub-units requires the study
of sub-units. Thus we focused at this level. Although For many organizations, implementing ERP means
we did gather some data at the corporate level, we did moving from a confederation of loosely coupled sys-
not concentrate our efforts there. Thus, we are missing tems to a tightly coupled one. Drawing on OIPT, we
many of the benefits (and perhaps some costs) that have suggested that tightly coupled systems produce
accrue at the global level, e.g. how is the aggregated significant benefits, including better coordination
12. 442 T.F. Gattiker, D.L. Goodhue / Information Management 41 (2004) 431–443
among sub-units and administrative efficiencies. The Canadian firms, Canadian Journal of Administrative Sciences
15, 1998, pp. 180–199.
more interdependence among sub-units in an organi-
[9] B.B. Flynn, E.J. Flynn, Information processing alternatives
zation, the more we expect this type of benefit to for coping with manufacturing environmental complexity,
accrue. However, we have also argued that in some Decision Sciences 30, 1999, pp. 1021–1052.
cases ERP systems will create either compromise or [10] J.R. Galbraith, Designing Complex Organizations, Addison-
design costs because of the standardization that they Wesley, Reading, MA, 1973.
require. The severity of these costs increases with [11] J.R. Galbraith, Organization design: an information proces-
sing view, Interfaces 4, 1974, pp. 28–36.
differentiation among sub-units. Here, we have pro- [12] J.R. Galbraith, Organization Design, Addison-Wesley, Read-
vided a description of differentiation in a manufactur- ing, MA, 1977.
ing context and a better understanding of some of the [13] T. Gattiker, D. Goodhue, Software driven changes to business
costs. processes: an empirical study of impacts of enterprise
We do not suggest that the case presented here is resource planning systems at the local level, International
Journal of Production Research 40, 2002, pp. 4799–4814.
representative of all ERP experience. Rather, we argue [14] D.L. Goodhue, M.D. Wybo, L.J. Kirsch, The impact of data
that organizations differ in the amount of interdepen- integration on the costs and benefits of information systems,
dence and differentiation among their sub-units, and MIS Quarterly 16, 1992, pp. 293–311.
that these two organizational characteristics merit [15] R. Hayes, S. Wheelwright, Link manufacturing processes and
consideration. product life cycles, Harvard Business Review 57, 1979, pp.
133–140.
Differentiation does not make it impossible to [16] T. Hill, Manufacturing Strategy: Text and Cases, 3rd ed.,
implement ERP well, but it may make doing so more Irwin, Boston, 2000.
costly. In most situations, firms will find that the [17] K.-K. Hong, Y.-G. Kim, The critical success factors for ERP
benefits of integration are sufficient to outweigh com- implementation: an organizational fit perspective, Information
promise and/or design costs. Occasionally, this will and Management 40, 2002, pp. 25–40.
[18] G. Huber, Organizational information systems: determinants
not be the case and wise firms may opt for looser of their performance and behavior, Management Science 28,
coupling by choosing to omit a plant or some other 1982, pp. 138–155.
sub-unit from an ERP implementation. However, firms [19] P.R. Lawrence, J.W. Lorsch, Organization and Environment,
are in no position to make such determinations without Harvard, Boston, 1986.
[20] V.A. Mabert, A. Soni, M.A. Venkataramanan, Enterprise
an understanding of differentiation and its effects in
resource planning survey of US manufacturing firms,
the ERP context. Production and Inventory Management Journal 41, 2000,
pp. 52–58.
[21] J.E. McCann, D.L. Ferry, An approach for assessing and
References managing inter-unit interdependence, Academy of Manage-
ment Review 4, 1979, pp. 113–119.
[1] N.H. Bancroft, H. Seip, A. Sprengel, Implementing SAP r/3, [22] J.W. Ross, M. Vitale, The ERP revolution: surviving
2nd ed., Manning, Greenwich, CT, 1998. versus thriving, Information Systems Frontiers 2, 2000,
[2] W.L. Berry, T. Hill, Linking systems to strategy, International pp. 233–241.
Journal of Operations and Production Management 12, 1992, [23] J.E. Scott, L. Kaindl, Enhancing functionality in an enterprise
pp. 3–15. software package, Information and Management 37, 2000, pp.
[3] J. Connoly, ERP: corporate cleanup, Computerworld, vol. 33, 111–122.
1999, pp. 74–78. [24] A.P. Sheth, J.A. Larson, Federated database systems for
[4] D.P. Cooke, W.J. Peterson, SAP implementation strategies managing distributed, heterogeneous, and autonomous data-
and results, Conference Board, Research Report, 1998. bases, ACM Computing Surveys 22, 1990, pp. 184–236.
[5] T. Curran, G. Keller, SAP r/3 Business Blueprint: Under- [25] S.G. Taylor, Are the process industries different? in:
standing the Business Process Reference Model, Prentice- Presented at the 23rd Conference of the American Production
Hall, Upper Saddle River, NJ, 1998. and Inventory Control Society, Los Angeles, 1980.
[6] R.L. Daft, R.H. Lengel, Organizational information require- [26] J.D. Thompson, Organizations in Action, McGraw-Hill, New
ments, media richness and structural design, Management York, 1967.
Science 32, 1986, pp. 554–571. [27] M.L. Tushman, D.A. Nadler, Information processing as an
[7] T. Davenport, Putting the enterprise in the enterprise system, integrating concept in organizational design, Academy of
Harvard Business Review 76, 1998, pp. 121–131. Management Review 3, 1978, pp. 613–624.
[8] M. El-Louadi, The relationship among organization structure, [28] K.E. Weick, Educational organizations as loosely coupled
information technology and information processing in small systems, Administrative Science Quarterly 21, 1976, pp. 1–19.
13. T.F. Gattiker, D.L. Goodhue / Information Management 41 (2004) 431–443 443
[29] C. Wilder, B. Davis, False starts, strong finishes, Informa- Before obtaining his PhD from the University of Georgia, he
tionweek, 1998, pp. 41–53. worked in operations and inventory management—most recently at
[30] M.D. Wybo, D.L. Goodhue, Using interdependence as a Rockwell Automation/Reliance Electric.
predictor of data standards: theoretical and measurement
issues, Information and Management 29, 1995, pp. 317–328. Dale L. Goodhue is a professor of MIS
and the C. Herman and Mary Virginia
Thomas F. Gattiker is an assistant Terry Chair of Business Administration
professor of Operations Management at at the University of Georgia’s Terry
Miami University in Oxford, Ohio, and College of Business. He has published
is an affiliate of the interdisciplinary in Management Science, MIS Quarterly,
Engineering Management Program. He Decision Sciences, Sloan Management
has published in Production and Inven- Review and other journals, and is an
tory Management Journal, International associate editor for Management Science
Journal of Production Research, Quality and a former associate editor for the MIS
Management Journal and The Decision Quarterly. His research interests include
Sciences Journal of Innovative Educa- measuring the impact of information systems, the impact of task-
tion. His current research is the applica- technology fit on individual performance, the management of data
tion of information technology, to the operations and supply chain and other IS infrastructures/resources, and the impact of ERP
areas. He was the 1999 APICS George and Marion Plossl Fellow. systems on organizations.