Sloan School of Management.

Reference papers on market oriented management systems online

. (page 1 of 11)
Online LibrarySloan School of ManagementReference papers on market oriented management systems → online text (page 1 of 11)
Font size
QR-code for this ebook




NOV 2 1966
Arnold E. Amstutz

Henry J. Claycamp i

John D. C. Little
Daniel S. Diamond-^
Tlieodore E. Hlavac ^ *

Prepareil for the Tliird M.I.T. Industrial Liaison Symposium, November 3, '966


NOV 7 1966
/I, 1. T. LIBRAKilS




The Evolution of Market Oriented Management Information Systems —
Arnold E, Amstutz

This presentation examines characteristics of successful management
information systems in use by operating and policy managers concerned
with the marketing function. It focuses on the attributes and impact
of representative systems and the characteristics of managers responsible
for their successful implementation. Bases for the evaluation of
information systems will be proposed. Representative systems performing
retrieval, monitor, advisory, and decision functions are examined.
Management implications of on-line, real time, and simulation based
system structures are also discussed.

See paper entitled, "The Marketing Executive and Management Information
Systems" .

Total Environment Management Control Systems — Arnold E. Amstutz

This presentation focuses on the development, testing, and implementation
of large scale computerized micro-analytic simulation of market behavior.
The characteristics of management information systems founded on the use
of this technique are discussed using examples from systems in use.
Problems associated with model conception and specification, function
verification, and system and sub-system testing and validation are
discussed with reference to operating systems. Sample simulation runs
based on test markets are evaluated and representative management uses
of the system are discussed.

See paper entitled, "Simulation Techniques in the Analysis of Marketing

Interactive Operating Systems — A Demonstration

A Geographic Model of an Urban Automobile Market — John D, C. Little
and Theodore E. Hlavac

A person chooses a store to shop at partly on the basis of the difficulty
of getting there. An understanding of the relationships involved is
important in marketing strategy, particularly for site selection purposes.
We study an urban new-car markets As a measure of shopping difficulty,
the distance from a person's residence to the car dealer is used.
Assumptions about buyer behavior lead to a model of competitive interaction
among dealers and car makes„ The model is fitted to three months of
sales data for metropolitan Chicago, An interactive computer system has
been programmed to make the mode) available for on-line investigation of
a variety of site selection and inter-brand competition questions. For
example;, a user can add, eliminate, or move a dealer and then ask for the
model -predicted changes in the sales and penetration of any dealer or

See paper entitled, "A Geographic Model of an Urban Automobile Market",



Computer Aided Preparation of Magazine Advertisement Formats —
Daniel S. Diamond

While most attempts toward quantification in advertising have involved
budgets, media selection, and m.edia scheduling, the present presentation
deals directly with magazine advertisement design. It is first
proposed that before an advertisement can have any effect on the consumer,
it must attract his attention. Readership is suggested as a measure of
the attention-getting power of an advertisement. Next, a set of six
Starch Magazine Advertisement Readership models is constructed by means
of multiple regression analysis on 1,070 advertisements appearing in L JFE
Magazine. The independent variables are predominantly advertisement
format descriptors, such as size, number of colors, and position in
magazine - twelve in all. Finally, a conversational computer program
is developed which requests from its user a readership objective function
(a function of the six Starch readership scores for the advertisement
and its cost), format restrictions, and a budget constraint. The
program then prepares that advertisement format, which while conforming
to all restrictions, maximizes the objective function.

See paper entitled, "Computer Aided Preparation of Magazine Advertise-
ment Formats".

Adaptive Management Control Systems - John D. C. Little

Companies try to conduct their marketing operations so as to respond
to changing market conditions. A model of such a process is studied
for the case of setting promotion rate. Company sales are functions
of promotional spending, but the relationship changes with time. An
adaptive system is devised that works as follows: Information about
sales response is collected by performing an experiment. The
experimental results are used to update a sales response model. Pro-
motion rate is chosen to maximize expected profit in the next time
period. The cycle is repeated. In designing the experiment, sample
size is chosen to minimize the cost of imperfect information plus the
cost of experimentation.

The model employs a quadratic sales response function with a parameter
that changes according to a randorr; walk. The optimal adaptive system
turns out to involve exponential smoothing of the experimental results.
A numerical example is studied analytically and by simulation. The
adaptive system is found to work better than various other policies.
In a sensitivity analysis, an adaptive system derived for one underlying
model of the market is found to perform well even when certain other
models actually apply.

See paper entitled, "A Model of Adaptive Control of Promotional Spending".



Arnold E. Amstutz


Arnold E. Amstutz


This paper is concerned with the present state of the art in market
oriented management information and control systems and the impact of these
systems on policy and operating management. As such, this discussion will be
narrowly focused. Normative questions of what management should or should not
expect or obtain from an information system will not be considered.
Ob jectives

The two major objectives of this paper may be simply stated as:

1. To identify relevant characteristics of successfully implemented manage-
ment information and control systems.

2. To evaluate the impact of representative systems on the managements
they were designed to serve.

The Process of System Evolution

It is seldom possible to Impose a pre-packaged information system on a
management group. There are no generalized management information systems.
Each company's management has unique information requirements; a unique
perspective on the environment within and outside of their firm; unique
priorities; and a style of management which is the unique product of the parti-
cular personalities making up the management group. Successful system develop-
ment is a matter of evolution. Effective management systems evolve over time
as management and system specialists learn to communicate, structure problems,
and achieve a joint focus on increasingly broad information needs.

About the Author. Arnold E. Amstutz is Assistant Professor of Management at
the Sloan School of Management, Massachusetts Institute of Technology. He is
engaged in research and consulting work relating to applications of computer
technology in marketing management and is the author of several articles in
this area.

*A talk presented to the American Marketing Association Meeting in Bloomington,
Indiana, September 1, 1966.

Page 2 .

The process of designing and developing a management information system
to meet the requirements of a particular management's decision style makes
significant demands on management time and thought. If the resulting product
is to be compatible with management's perspective, priorities, and systems of
measures, management's models of the decision environment must be made explicit
and used as the basis of system design. In addition, management must evaluate
the implications of alternative system structures in terms of criteria which
only they can supply. Intelligent choice between alternatives must be based on
management understanding of the implications of available structures. In order
to achieve understanding the manager must take the time necessary to become
familiar with the management implication of basic system design concepts.

Just as there is no single generic management information system, there
is no single payoff associated with all systems. It may be argued, in fact,
that the benefits gained from a particular system are largely determined by
the demands which management makes of the system and the capabilities established
in the course of system design.

Dimensions of Evaluation

No two managements have the same information needs. As indicated earlier,
the characteristics of systems developed for different managements are as
different as the managers' personalities, the problems they face, and their
approach to the complex art of management. Despite these differences it is
possible to identify sets of dimensions for use in evaluating specific systems
and isolating similarities and differences between systems. Figures I and 2
illustrate four such dimensions.
Information Recency

The first dimension, information recency, refers to the time lapse between
occurrence of an event in the environment and inclusion of data describing that

Page 3.

event in the system. This may range from several weeks in the case of certain
market developments to a few hours or minutes for automated inventory control.
Information Aggregation

The second dimension, information aggregation, describes the detail with
which information is maintained in system data files. Inventory control systems
in which information regarding product components or sub-assemblies is maintained
at the item level are representative of relatively disaggregated (micro) data
maintenance while industry market share statistics of the type developed through
trade associations are representative of highly aggregate (macro) measures.

As illustrated In Figure 1, there is normally a relationship between level
of aggregation and the time delay involved in incorporating associated data in
the system.
Analytical Sophistication

The tliifrd dimension, analytical sophistication, refers to the sophistica-
tion of models or structure encompassed by the system. As illustrated in
Figure 2, the lowest level of analytical sophistication is that required to
identify a particular file and record. At this level it is only necessary for
the computer to retrieve the specified record and display the information which
it contains. The second level of analytic sophistication involves aggregation -
gathering together numbers from within one or more records to produce a total
or sub-total. At the third level the computer may be programmed to perform
arithmetic averaging or to compute differences. The fourth level, logical
analysis, introduces the use of classification schemes through which various
types of data are aggregated within sub-sets or conditionally segmented.

At the fifth level of analytical sophistication statistical analyses may
be employed to develop extrapolations from historic data, statistical best
estimates, analyses of variance, or trend estimates.

The term learning Is used in Figure 2 to indicate adaptive system processes
through which the computer is programmed to modify parameter values or model

Page 4.

structures on the basis of experience (data Inputs received) over time.

At the most advanced level of analytic sophistication, simulation, models
on which the system is based constitute an artificial environment paralleling
real world markets referenced by the information systems. The managers'
perception of the environment has been explicitly modeled to a sufficient extent
and detail to justify the assumption that the models making up the simulated
environment duplicate in all relevant aspects the response pattern of the real
world environment monitored through the Information system. Inputs to the
information system are directly related to process variables In the simulation
models. At this stage in development the system provides management with the
capability of testing proposed policy and strategies in the simulated environ-
ment; choosing between alternatives on the basis of resulting output; implementing
the policies in the real world environment; and evaluating the effectiveness
of implemented plans through the information system. The manager references
the simulated environment to ask "What if"? and the information system monitoring
the real world environment to determine "What is"?.
Computer Authority

The final dimensions of system evaluation, authority delegated to the
computer, is closely associated with the system's analytic sophistication.
Management is more willing to delegate authority to sophisticated systems and,
conversely, as management places greater demands on an information system, a
greater level of analytic sophistication must be embodied in the system structure.

At the lowest level management may delegate to the computer authority to
retrieve Information from specified records and files - entrust to the computer
system processes associated with identification and retrieval . Once a retrieval
capability has been established it Is usually a short step to the next level
of computer authority. Recognizing that the computer has access to all records
In the file, management concludes that while the computer is "looking at" the
contents of each record it might as well check the reasonableness of record

Page 5 .

content to Insure against gross clerica! errors. At this stage the computer
is delegated a supervisory function checking on human personnel responsible for
input .

As management comes to accept computer review for purposes of error
detection, they normally begin to think In terms of other functions which the
computer could perform "while looking at all those records". It follows quite
naturally to have the computer perform additional analyses on records which it
is reviewing and refer for further review and action situations meeting criteria
established by management.

Management frequently finds that certain classes of monitor output are
consistently subjected to additional analyses to determine whether or not
action is warranted. In such situations it Is natural to suggest that the
computer be programmed to perform the additional calculations in order to add
a recommendation for action to the monitor report.

As management gains experience with computer based recommendations they
may find that In most situations they are able to implement computer recommenda-
tions without further investigation. Criteria may be modified to isolate non-
typical cases requiring additional review. The computer is then given authority
to take action on the remaining cases in which its recommendations are a valid
basis for action.

The hierarchy of Figure 2 suggests that delegation of authority to
predict involves a higher level of management dependence on the computer than
authority to act. While the models on which the computer bases its action
normally involve prediction, the potential impact of computer based prediction
is often greater than computer action. Computer originated actions may adversely
affect the firm's position at a point in time. However, actions relate to the
operating sphere while predictions are the basis for planning. Thus, inaccurate
prediction may have a damaging effect on the firm's activities for months or

Page 6.

years while erroneous actions can be corrected in days or weeks.

The chance for successful computer based prediction Is ironically reduced
by the very nature of management-computer interaction. Since predictions are
often based on relatively sophisticated models, management is frequently
hesitant to accept the computer's prognostication until they have gained
experience with the system and had an opportunity to "see how it does". With
the passage of time management's satisfaction with predictions which are
verified by subsequent experience increases. However, as time passes the
modeled environment may change - the original models may become less and less
applicable. Finally at that point when management is ready to take action
based on the computer's predictions the models may be completely outdated and
no longer accurately represent the decision environment. When this happens
the stage is set for disillusionment or worse. It is considerations such as these
which argue strongly for management involvement in the system design process,
familiarity with system structure, and understanding of models on which system
decisions and predictions are based.

Characteristics of Successful Systems

While specific functions performed by successful systems are as varied
as the managements to which they contribute, four common characteristics of
successful systems or perhaps more correctly the environment in which successful
systems operate, can be noted.

1. The system is founded on management's conception of the decision
envi ronment.

2. The user-manager understands the system structure.

3. The system Is based on disaggregated data files.

k. System development has proceeded to increasing levels of sophistication
through a process of gradual evolution.

Page 7.

Management's Conception of the Environment

If a system Is to provide meaningful Information to a particular manage-
ment It must reflect that management's priorities and provide Information of
a type and In a form which Is assimilable In the context of existing manage-
ment decision processes. In most situations this requirement specifies that
the Information must be selectively generated — management Is simply Incapable
of assimilating reams of paper - and must be based on accepted measures - out-
put must relate directly to management conceptions of processes occurring in
t he mon i to red env I ronmen t .

In order to meet this requirement an Information system must be based on
explicit models of the environment provided by management. In most instances
management's Initial system definition is stated In qualitative "business
terms". Before meaningful specifications can be established this frequently
vague and ambiguous initial statement must be refined and restated in explicit
terms. Factors considered relevant in the decision environment must be defined
and differentiated from those to be excluded.
Management Understanding

Management must be Involved In this quantitative specification of system
boundaries. They must understand and accept the conceptual structuring of
system requirements in terms sufficiently explicit to define the measures and
analytical procedures to be encompassed by the system. If this level of
communication Is not achieved, It may be Impossible for those concerned with
system formulation to develop a system which will be used.

But, It may be argued, this means involving management In wholly
unacceptable detail. "Management Is appropriately concerned with the big
picture. It is unrealistic to expect them to become Involved In questions of
measurement." The response to this objection Is to reiterate the area of
eventual system application — matters of company policy. It is difficult to

Page 8.

conceive of a point In the dec is Ion -making process at which involvement Is more
warranted than In Insuring precise description of the decision structure —
unambiguous system specification.

The process of explication often uncovers the not altogether surprising
fact that various members of management have different Implicit conceptual
models of the decision environment. Making these models explicit removes the
ambiguities which permit vague words to mean different things to different
people. Alternative representations are proposed and necessitate the creation
and validation or rejection of more than one model.

As with any other specialized tool the Information system must be carefully
designed to meet the specific requirement of the craftsman who will use it and
the user must understand its function and capabilities. There is no such
thing as a generalized Information system. It Is difficult to conceive of a
more specialized and highly segmented market than that for management Informa-
tion. The products which have gained acceptance In this market have been one
of a kind special orders produced with careful attention to the needs and
preferences of the ultimate user.
The Disaggregated Data File

At the heart of every successful information system is a disaggregated
data file - a file in which information is maintained in detailed time
sequence as it is generated. As new Inputs are received they are maintained
along with existing data rather than replacing or being combined with existing
information. New data are not combined with old to form sums, averages, or
aggregate distributions. As a result, structural biasing through aggregation
which destroys much information value is avoided.

Figure 3 illustrates the concept of a disaggregated data file based on
data from invoice records. The disaggregated customer file contains the name,
address, demographic, and financial experience records for particular consumers.
Each transaction is recorded in chronological order in the file so that eit any

Page 9.

point in time It is possible to recreate the company's interactions with each
consumer over time. In a similar manner the product file is organized to
reference a detailed chronological sales record.

The importance of a disaggregated file rests in part on the evolutionary
process through which successful Information systems develop. Although an
Snformation system may initially be designed to perform strict limited
functions, as management gains experience these functions change. !f data
are initially structured (aggregated) to meet first stage requirements, later
modification of system functions necessitate costly file reorganization.

The existence of a disaggregated file facilitates system evolution. Given
access to detailed chronological data the manager is able to test new concepts
and ideas against historical data asking the question "what would have happened
if we had used these criteria in our monitor system - performed this analysis
to isolate actionable situations?". In the first stages of system development
it is simply impossible to anticipate the direction of later advancement.
Aggregate data files may preclude highly profitable system modification. The
disaggregated data file provides the flexibility which is the prerequisite of
intelligent system evolution.
Design for Evolution

Successful Information systems are designed to permit expansion and change.
As Indicated above, the disaggregated data file is a key element in system
flexibility. . In addition, data files must be designed to permit expansion.
Variable, rather than fixed record length file structures and self-expanding
file constructs are basic to the well planned system.

As management gains experience in working with well organized and accessible
data they become increasingly interested in and prepared to use more advanced
analytical procedures. The system's analytical structure must not preclude
this advancement. Programs must be organized to permit experimental use of new

' Page 10.

techniques as well as the permanent incorporation of additional capabnities
as part of the standard system configuration.

Characteristics of Managers
Who Implement Successful Systems

In view of the key role which management plays In the specification and
design of information systems, it is not surprising to find that the managers
who are successful in developing and using management Information and control
systems share common attributes which distinguish them from the average
admin Istrator ,

There Is no industry bias In the population of successful information
system users. Successful systems have been designed and Implemented in consumer
and industrial product and service companies as well as financial Inst 5 tut ions .
The members of this heterogeneous group may, however, be distinguished from
their colleagues on the basis of the following characteristics which they share:

1. Belief that corporate growth is limited by their ability to assimilate
and act on information.

2. Willingness to consider change.

3. Willingness to become personally involved in explication.
k. Willingness to test preconceptions.

1 3 4 5 6 7 8 9 10 11

Online LibrarySloan School of ManagementReference papers on market oriented management systems → online text (page 1 of 11)