Decision Tree

IT Sourcing Decision
Zachary H. Crawford
CIS540 ­ Professor Rachdan




-1-
Introduction
The in or outsourcing decision should be based on a wide

variety of assessments of current and future needs. The point,

of course, is that there are many decisions that contribute to

the final one of whether to make or buy. The decision process

should begin with an honest assessment of current processes. It

gains the organization nothing if it only automates processes

that have evolved into unnecessary hindrances.


Article Review: "A multiple-criteria framework for
evaluation of decision support systems"
Phillips-Wren, Hahn and Forgionne (2004) published "A

multiple-criteria framework for evaluation of decision support

systems" in the August, 2004 issue of Omega. Their purpose was

to evaluate the usefulness of using decision support systems

(DSSs) in making the determination of whether to make new

systems in-house or to buy packaged products from outside

vendors. Phillips-Wren, Hahn and Forgionne (2004) cite other

authors (P. Humphreys, R. McIvor and G. Huang 2002. An expert

system for evaluating the make or buy decision. Computers &

Industrial Engineering 42(2-4):567-85) as stating that "firms do

not have formal methods for evaluation of the make versus buy

decision and often make the decision on the basis of a single

criterion such as overhead costs" (p. 323).




-2-
The stories of any industry professional certainly can bear

out this view. One such company, IMT Custom Machine Company

(IMT-CMC), found itself with a system that was fragmented and

outdated, to the point that individuals shopped for and added

their own PCs rather than limit their computer access to IMT-

CMC's old mainframe system (Brooks and DeHayes, 2000). The

company held a wealth of problems, including trying to patch

together incompatible systems inherited through acquisition of

other companies.

Phillips-Wren, Hahn and Forgionne (2004) note that DSSs

typically have "been evaluated on only a single criterion such

as the outcome from decision making" (p. 323), a result of using

them within the same context in which most make-or-buy decisions

are made (i.e., the single criterion). The authors found that

though DSSs as they currently are used have the potential to

affect the outcome of the decision-making process, the use of a

DSS "did not offer a statistically significant improvement in

terms of outcome-related characteristics" (Phillips-Wren, Hahn

and Forgionne, 2004; p. 323). Organizations still need to

undertake thoughtful decision-making, whether they are

approaching the make-or-buy decision with or without the

assistance of a DSS.

At IMT-CMC, the company already had invested more than

7,000 hours into trying to make a "free" program work, without


-3-
success. The cost of the "free" program likely was higher than

the cost of an efficient and integrated system produced

specifically for IMT-CMC. The company already had wasted

literally thousands of hours trying to make a free program and

outdated equipment meet the current needs of the business,

apparently without a thought of the needs of the future. It had

choices ranging from bringing in an outside consulting company

to design, build and implement an entirely new system to

approaching the problem in the way it chose at the outset. Had

the company's management chosen the option for a totally new

system, however, it still would have faced many of the same

questions. It also may have ended up with a brand new system

that met current needs, but only automated processes that should

have been eliminated.

One of the first assessments that the organization needs to

make is whether it has sufficient in-house talent to make its

own system. This assessment should include competence, of

course, but it goes much further than only that. Even if in-

house personnel have the technical ability to design, build and

implement a new system, do they have the time available to them?

If so, then that determination raises yet another question: if

technical personnel have the time to undertake such a project,

does the organization approach staffing wisely?




-4-
Beginning the Assessment
A wise approach is to begin the decision-making process

with a hierarchical decomposition of the current system,

relative to identified needs for both the present and the

future. Martin, Brown, DeHayes and Hoffer (2002) give a concise

definition of hierarchical decomposition that made sense to this

reader who has no technical background of any kind. The authors

describe "the system" as "a set of interrelated components"

(Martin, et al., 2002; p. 318), and that any one of the

components making up the larger system can itself be viewed as a

system. In turn, this subsystem can further be broken down into

a more narrowly-defined set of components, which also may be

able to be viewed as subsystems themselves. Martin, et al.

(2002) explain that there are five primary goals of hierarchical

decomposition:

· Break the larger system down into more manageable and
understandable segments;

· Analyze or change only part of ­ rather than all ­ a
system;

· Allow chronological design or construction of a specific
subsystem;

· Allow targeted use of a specific subsystem; and

· Allow more independent operation of system components.




-5-
This necessarily involves both high- and low-level analysis

of system operation and goals. Lee (n.d.) provides a

nontechnical explanation of high- and low-level analysis. High

level: [Go(Supermarket),Buy(Milk),Buy(Bananas),Go(Home)] ... It is

a long way from instruction fed to the agent's effectors (Lee,

n.d.).

In contrast to this high level analysis is the low level

approach, which Lee (n.d.) also illustrates in nontechnical

terms: [Forward(1 cm),Turn(1 deg),Forward(1 cm),...] (Lee, n.d.).

In this example, the instructions and the effects gained

from those instructions are much closer both in proximity and in

time. Change in the system's effects, results, usefulness,

relevance or any other characteristic can be done on this level;

high level analysis provides an overview of how the system

arrives at its visible results. Martin, et al. (2002) provide

the reasons for viewing and analyzing systems in this way in

their list of the five primary goals of hierarchical

decomposition. It allows different people to work on different

subsystems simultaneously, whether they are designing, building

or altering existing systems. It also allows systems to be

changed to provide information in the form it is most useful.


Two Beginning Points
Beyond the questions of what the system should do also are

those of how it should accomplish those ends. As example, if


-6-
the company has many users, it can be more cost effective to use

the thin client, three-tier approach rather than buying many PCs

capable of supplying all processing needs, then adding an

applications server to balance the deficiencies. It is also

more efficient to update applications on a single server than on

many client machines if users are dispersed over a wide area.

Smaller organizations or departments may find it more cost

effective to operate only a database server in a two-tier system

while providing users with client machines with greater

capabilities.

One of the leading points of consideration in the make-or-

buy decision must be whether sufficient in-house expertise

exists, as stated above. Following are two possible scenarios,

one where there is a great deal of in-house expertise, the other

where there is not.

Sufficient In-House Abilities
IT is famous for its failed projects, unintentionally

expanded projects and slipped development and implementation

schedules. Keil and Montealegre (2000) report that a survey

recent to 2000 "revealed that 30% to 40% of all IT projects

exhibit some degree of escalation" (p. 55). Most are not

brought under control and end up being implemented well beyond

schedule and over budget, if they are implemented at all (Keil




-7-
and Montealegre, 2000). One way to avoid this problem is to

purchase a packaged product.

Advantages to purchasing a system include shorter cycle

time to implementation; lower total cost; in-house resources are

free for other uses. Application quality may be higher, and a

packaged product can "infuse new expertise" (Martin, et al.,

2002; p. 379) without increasing departmental or organizational

headcount. Disadvantages include risk of lack of product

knowledge; greater risk in the form of required business process

changes; and long-term dependency on an external vendor.

Key decision rules should include comparison of overall

costs of each option; whether there are tight time limits on

implementation; and whether use of the packaged product would

require business process changes. Total cost should be

determined through cost-benefit analysis or capital budgeting.

In the case of IMT-CMC, the company had invested more than 7,000

hours into trying to make a "free" external program work

(Martin, et al., 2002), without success but at high cost. An

analysis of total costs could have pointed the company to a

packaged product that would have been operational in a

relatively short time, at a cost much less than 7,000 hours of

salaries and benefits.




-8-
Little or No In-House Expertise
The three leading concerns would be total cost, business

process effects and making the best overall choice. There would

be other considerations, but most of those other points

ultimately would lead back to one of these three.


Total Cost
The first concern should be with total cost ­ not just the

purchase price and cost to implement, but also the costs of not

acting at present. Management should want to know what

operating and maintenance costs would be over the expected life

of the package, and it should want to have as accurate a dollar

figure possible of how much a specific package could save

(Horngren, et al., 2002). Will it allow the company to add new

employees at a slower rate, or better yet, will it allow the

company to avoid hiring a certain number of people as the

business grows? Is the package a complex one (such as one used

in computer-aided manufacturing) that will need a babysitter

that the company will have to hire and keep? Will it be a

solution or a problem?


Business Process Effects
Will the software make current processes more effective, or

will it require that the company change important processes to

accommodate its needs? One reason to use any software product

is to make processes more efficient (Turban, et al., 2002), but



-9-
if a specific package requires that people serve it instead of

it serving people, then it likely something that management

should want to be certain to avoid. On the other hand,

management also should not want to be putting any time into

automating processes that should be changed.

This is an apt place to invoke high-level system analysis.

What is the immediate need that the company seeks to meet

through the proposed change? If it is merely automation of

current processes, then what the company needs from the new

system will be far different than if the company also is seeking

to streamline processes for the purpose of enhancing internal

efficiency. With the greatly heightened need for increased

efficiency across all industries in all economic sectors,

management can be considered to be remiss if it does not

consider efficiency-related options as well.


Choosing the Best Option
How will management know if it decides on the most useful

and trouble-free system that also fits the needs of the

business? The company likely could benefit from having a

consultant in on the process of identifying available choices

and choosing the best fit for the business, but that can be

costly in itself.

An appropriate approach for management would be first to

ensure that it had identified all of the things the company


- 10 -
would need software to do for it in both the present and the

future. When all involved are satisfied that they have

identified every current and potential need they would like to

see a new system fill, then it would be time to call in that

consultant. The consultant's role would be to review the list -

s/he's certain to add to it ­ and to make suggestions of

available products.

Management should then give software salespeople the job of

demonstrating how their software is best for the company's

current and future needs, letting their employers take on labor

costs for that stage. At the end of this process, the

consultant should return to review what the salespeople have

provided and to recommend a final choice.

The company's management may still have misgivings, but it

could have far more confidence that the purchase is a wise one.

The final decision would have been arrived at by people more

technically knowledgeable than general managers, and there would

be no unpleasant surprises in terms of total costs when the plan

selected is one designed by the vendor selected to implement it.


Conclusion
When approached systematically and in consideration of all

costs involved, the make-or-buy decision is one that has the

ability to bring more benefit than only the fact of having a new

system to use. The right decision can positively contribute to


- 11 -
the organization's competitive advantage, but the wrong one can

be insidiously detrimental.




- 12 -
References

Brooks, Christopher L. and Daniel W. DeHayes. (2000). Case
Study 1-1: IMT Custom Machine Company, Inc.: Selection of a
hardware platform. In Martin, E. Wainright, Carol V.
Brown, Daniel W. DeHayes, Jeffrey A. Hoffer and William C.
Perkins, eds. (2002). Managing Information Technology, 4th
ed. (Boston: Pearson Custom Publishing), p. 133.

Horngren, Charles T., Gary L. Sundem and William O. Stratton
(2002). Introduction to Management Accounting, 12th ed.
(Boston, Mass.: Pearson Custom Publishing).

Keil, Mark and Ramiro Montealegre (2000, Spring). Cutting Your
Losses: Extricating Your Organization When a Big Project
Goes Awry. Sloan Management Review, 41(3), p. 55.

Lee, Kwang-hyung (n.d.). Chapter 12: Practical Planning.
Retrieved August 28, 2004 from
http://ai.kaist.ac.kr/~jkim/cs570-99/Lectures/cs570-12.ppt.

Make or Buy Decision for Software Vendors -Electronic Software
License Generation & Distribution. (2002). Macrovision.
Retrieved August 27, 2004 from
http://www.globetrotter.com/articles/makebuy.pdf.

Martin, E. Wainright, Carol V. Brown, Daniel W. DeHayes, Jeffrey
A. Hoffer and William C. Perkins. (2002). Managing
Information Technology, 4th ed. (Boston: Pearson Custom
Publishing).

Phillips-Wren, Gloria E., Eugene D. Hahn and Guisseppi A.
Forgionne (2004, August). A multiple-criteria framework
for evaluation of decision support systems. Omega, 32(4),
p. 323.

Turban, Efraim, David King, Jae Lee, Merrill Warkentin and H.
Michael Chung (2002). Electronic Commerce 2002. (Boston:
Pearson Custom Publishing).




- 13 -