Supporting Material is available for this work. For more information, follow the link from
the Table of Contents to "Accessing Supporting Material".
Table of Contents
Go Back
A DECISION SUPPORT SYSTEM (DSS) FOR EVALUATING
OPERATIONS INVESTMENTS IN HIGH-TECHNOLOGY BUSINESS
Adolfo CRESPO MARQUEZ
Department of Industrial Management, School of Engineering. University of Seville
Camino de los Descubrimientos s/n. 41092 Sevilla, SPAIN
Telephone: +34 954 487215; FAX: +34 954 486112
e-mail: adolfo.crespo@esi.us.es
Carol BLANCHAR
Executive Manager, Analysis of Operational Investment Alternatives, Conexo, Inc.
1909 Magdalena Circle # 75, Santa Clara, CA. USA
Telephone: +1 408 248 4845; FAX:+1 408 248 5578
e-mail: carol@conexo.com
Abstract
The evolution in the way that businesses approach markets has been a frequent
literature topic in the last few years. In the high-tech industry, even the most successful
companies have been mainly focused on the features of their products and processes, trying
to develop their technology to gain a price / performance advantage, and thereby protect or
increase market share. However, this approach is disconnected from their beliefs about what
target customers really care about, nor does it is consider which of those underlying
assumptions are most critical to business growth in share, revenue, and profit. This paper
proposes a Decision Support System to connect customer value to business targets, providing
scenarios to show the customer responses and business results that will enable future funding,
with optimization techniques to compare alternatives.
The first step is for business planners to characterize their target market by
formalizing what are often informal but deeply held beliefs about what drives their customers'
purchase decisions. They create a list of attributes that together define customer value, the
basis on which customers in the target market compare and select from competing products.
With that attribute list, planners sometimes are able to go further and segment their market by
grouping customers together who put top priority on the same attributes. This system
dynamics model connects planned investments to expected improvements in the customer's
perception of those critical attributes, (relative to the competition), and thus increase sales,
revenue, and market share.
Keywords: Decision Support Systems, System Dynamics Models, Investments Evaluation, Product
Life Cycle Management, Supply Chain Management, Channel Strategy, New Product Introduction,
Solution Selling, Solution Integration, Product Awareness, Pricing.
* Corresponding Author.
1. Introduction
Decision Support Systems (DSS's) are tools that an organization uses to support and
enhance decision-making activities [1]. Early use of decision support analysis were marketing
decision support systems (MDSS), defined [2] as a coordinated collection of data, system,
tools and technology, with supporting software and hardware by which an organization
gathers and interprets information from business and environment and turns it into a basis for
marketing action.
Within the field of marketing, Higby and Farah [3] found that in the US, 32% of the
companies have installed some form of marketing DSS’; In the Netherlands, Van Campen et
al. [4] estimated the penetration of decision support systems in marketing at 37%’. The fact
that current formal marketing plans incorporate information resources in 95.2% of the firms,
compared to incorporation in 76.2% of the firms' strategic business plans [5], illustrate about
the importance of MDSS at present .
Companies and business planners have recognized the strategic importance of MDSS and
are stepping up their investments in information technology for marketing [6] Adoption of
MDSS is higher in companies with consumer products compared to industrial (business-to-
business) products companies, and in companies with more market information available [7].
Their objective is to support a decision making process which is primarily a matter of
reasoning (using the mental models of the manager) and analogizing (based on stories about
similar events retained in mind). For instance, Van Bruggen et al. [8] found that managers
who use a DSS are less inclined to anchor their decisions on earlier decisions compared with
managers who do not use the system. Similarly, these authors found that the incorporation of
model-based results into a DSS is especially beneficial. Prominence effects, overconfidence
and other biases are reduced for managers who use model-based DSSs relative to managers
who do not. In literature we find that although the applicability of some marketing models to
real-world problems has been questioned [9], there have been many examples of successful
marketing model applications (see for instance [10] and [11]).
' Based on a survey among 212 executives.
? Based on a survey of 525 companies with over 10 employees and marketing manager present.
Beyond marketing, others of these model applications are within the new products area
[12], trying to understand the dynamics between changing demand and the entry and exit
behaviors of competitors in the market place. These works model demand and number of
competitors simultaneously and empirically investigates some high-tech markets. Still other
models try to bridge between new product introduction and marketing to understand the
relationship between the number of competitors and the rate of technology diffusion [13], or
to tie conceptual design in a new product introduction with cost modeling and marketing
considerations [14].
In this paper, however, we go further to model product design and marketing innovations
to anticipate and explain the way collaborative teams, both within firms and between partner
businesses, may gain and retain customers in a very competitive high-tech marketplace. The
model also considers the expected response of a changing set of competitors. In this work we
pay special attention to the characterization of the customer behavior, and we use system
dynamics to build our simulation model*. The simulation model confirms through team
review that we have captured the behaviors that explain their customer segment response to
changes in product attributes and price, creating collective understanding of the existing
business environment, and able to be validated by historical data when available. This can be
transformed into a DSS model by examining the impact on share, revenue, and profit from
engineering and manufacturing changes made to product attributes and prices, as well as
changes made to influence the customer's perceptions, given what we believe to be true about
the business dynamics. We show an example of the model used as a framework for a scenario
(simulation) where business planners may explore specific product improvement strategies.
> System Dynamics is a methodology that was born at the MIT in the late fifties. Developed by Jay W. Forrester,
it is focused to the observation of the behavior patterns, instead of concrete events, of the systems. System
dynamics models help to understand the relationship between behavior patterns and system structure. Problems
related to system behavior, can be then solved by changing the system structure. The models are constructed
using cause-effect relationships among the variables. Frequently, relationships may result in feedback loops
involving different endogenous model variables. The feedback loops are deeply studied in system dynamics,
their gain, delay and dominance (of one loop among other) will explain the observed system behavior patterns.
At the same time, feedback loops help to make visible different system variables life cycles (technology cycle,
business cycle, industry cycle, etc.), and also may represent managerial behavior over time conveniently. The
initial focus was on the application of SD to management issues, but was soon extended to the analysis of
environmental, social and macro-economic problems (see for instance[15]). In [16] can be found a collection of
early papers. Since the mid-eighties, there has been renewed interest in applying SD to business policy and
strategy problems. This interest has been facilitated by the availability of new, user friendly, high level graphical
simulation programs (such as ithink, Powersim and Vensim). Easily accessible books describing the SD
approach (for example, [17] [18] and [19]) have also played a key role.
The simulations calculate expected results in the context of current competitor investment
and response, and planners can choose strategies to best meet business (financial and
operational) targets and forecasts.
The rest of the paper is organized as follows: In Section 2, we characterize high-tech
business planning today, with multiple dimensions of business organization, solution
architecture, channel strategy, and changing segment needs, and with each dimension
changing over time. In Section 3 we introduce a model using the System Dynamics
methodology, proven effective for quickly simulating and understanding dynamic, non-linear
behavior as a basis for collaborative decisions. Sections 4,5 and 6 are devoted to the
explanation of the consumer purchasing behavior, financial and investments sub-models
respectively. Example simulation results are presented in Section 7, generated from both
quantitative and qualitative data inputs, a critical requirement in today's fast-changing global
marketplace, and we also suggest a generic scenario as a starting point. This Section also
explains how this model can rapidly be transformed into a Decision Support System for
collaborative planning, along with some optimization capabilities to answer several possible
questions with the purpose of improving business planning under different scenarios. In
Section 8 we discuss our results to date and managerial implications of the high-tech
Business Decision Support System. Finally, Section 9 concludes the paper with a summary of
our findings and some useful directions for future research.
2. A characterization of high-tech business planning process complexity
Business planning within a high-tech environment is both dynamic and complex, with
a critical need for nonlinear, relational input and mathematical rigor. This is especially the
case where planners and decision-makers must rely on subjective and potentially biased data
[20], and where data sources span across cultures and languages.
Relational input is important where projections of both market demand and competitive
position are essential inputs to strategy ([21]; [22]). There is simply not a large enough
sample of good data to get statistically valid outcomes on the basis of projections from past
trends and patterns, nor are there controlled representative data sources, to support
correlations or regression analysis.
For all these reasons, planners increasingly turn to simulations to build confidence and
consensus in selecting operational investments to improve or protect market share, revenue,
and profit for global high-tech businesses. Adding the ability to analyze decisions in light of
the impact on share, revenue and profits, turns the simulation model into a decision support
system.
The reader needs to understand that many high-tech planners are more interested in share
as a business metric than either revenue or profit’. This is closely tied to the fast pace of
technology and product life cycles, and the increasing difficulty of trying to gain market
share as the market matures. In addition, market share is tracked and reported in trade and
investment publications and watched closely by investors and analysts looking for visible
short-term results to publicized strategy.
As a first step in the introduction of a model that can meet these needs, let’s summarize
the unique characteristics of the high-tech marketplace:
— Volatile, uncertain markets with great pressure on managers for near-term market share
and/or financial performance (In U.S. high-tech programs and product lines may be
funded for a period of time in spite of poor financial results if they prove themselves,
quarter by quarter, able to capture and hold share in strategic markets).
— Multiple planning dimensions, including technology path, product architecture, delivery
chain, alliances, channels, and services.
— Little historical data, due to technology adoption rates, reorganizations, mergers and
acquisitions, globalization, and new channels for order and distribution.
— Isolated groups of expert knowledge, each with their own language and systems.
- Absence of a single view of the possible impact of an investment, especially when results
are scattered across space and time, well beyond the scope of any single enterprise planning
system.
4 Notice that market share may be widely used, but can be sometimes a very poor performance metric. Absolute
sales volume could be preferable, since it is directly traceable to customer gains and losses. For instance, 90% of
a tiny market could contribute less to earnings than 25% of a large market.
3. A general overview of a model for a high-tech business and marketplace
Figure | is a representation of numerous planning team dialogues about the way
business grows when it offers a valuable product to an existing market. The diagram links
operational investment, conditioned by policy, to business revenue growth over a financial
year. In this way, financial constraints are introduced into the model. Obviously, the higher
the growth at a reasonable margin, the greater the level of investments that are available for
the next year.
This simplified diagram does not show all the exogenous and endogenous factors that
condition results over time, and that are included in this model for a valid simulation. For
many reasons business planners know that over time it takes more dollars of investment to
maintain the same level or to grow share the same amount (this, of course, does not apply to
all cases, e.g. if a big rival has failed, the firm may be able to grow or sustain share with less
expenditure), and the model indeed shows diminishing returns over time, depending on a
number of factors. Most importantly, the model clearly shows why "doing nothing" is almost
never a good decision for a high-tech business, and helps a business that has enjoyed great
success in the past to act aggressively to protect its position for continued profit and growth..
Incremental investments are represented in this model as completely variable, even
though volume ramps up or down would surely affect the return on fixed costs. We do not
include a fixed costs component simply because none of the financial or strategic planners
among the companies we worked with have done so. Industry practice is to build fixed costs
into overhead rates as part of labor, material, and overhead in internal part costs, or priced
into purchased parts, and are not visible to our clients nor used by them when they evaluate
and compare business plans.
Revenue &
revenue
growth ae)
Profit & profit contribution Sales
Allowable investments @s
Marketshare
Price attributes ,
perception Perceptior’
of value
Non price attributes —
perception
Figure 1. General Model Overview (original team design)
The allowable change in spending level corresponds to an expected changed value of
specific attributes. Note that the investment cycle is a consequence of corporate policy and
regulated periods to report results and commit resources, where external economic cycles and
market occur at their own pace. The model recognizes those delays between a change in
spending and a resulting improvement in customer value and sales growth.
Business planners try to further group their customers in segments within the target
market, according to the relative importance the buyers place on one or the other of the
attributes that drive their market overall.
In a scenario, investing to improve product attributes drives positive change in
customer perceptions, which are assumed by business planners to drive each competitor's
share in each segment of the overall market, and of course the related financial results.
In terms of confirmation and validation, the general model structure that we present in
figure 1 was synthesized and refined with commercial and consumer business managers,
systems analysts, critical part contract managers, financial executives, and experts in high-tech
workforce collaboration. The results are represented by the three sub-models that we show in
figure 2.
The financial model is set by reporting rules, the investment model by budget and
targeting practices, and a value index computed from quality relative to price has gained wide
acceptance and general industry use. In this paper, the authors compute the value index in a
manner that takes advantage of the capabilities of system dynamics for the benefit of fast moving
high-tech industries.
Marketshare
Investments
model
Price attributes
perception Perceptioit
of value
Non price attributés
Purchasing behavior model
Figure 2. Sub-models overview
4. Modeling customer purchasing decisions
"Purchasing" here represents the customer's decision to buy, and purchasing behavior
is the customer response to perception of value relative to the competition. How does the
customer perception of product quality and price attributes impact market share for this
business and for its competitors? In this section we will try to study this problem by
formalizing the relationship among the variables involved. Before proceeding with the model
development and discussion, we first describe the notations and definition of the main in the
purchasing behavior model variables as follows:
Subscripts:
competitors, including this business
segments grouped by the most important attributes
quality attributes
price attributes
Input: Customer Perception of each Competitor
Qac;; ; perceived quality attribute i of the competitor / in t
Pacj\, : perceived price attribute k of the competitor j in t
Qab', _. baseline perception of quality attribute i for all competitors in ¢
Pab‘, . baseline perception of price attribute x for all competitors in t
Input: Expected Impact for each Competitor in Each Segment
Qaj..;_ : competitor j impact on value for customers of the s segment and through
the quality attribute 7 in ¢
Pay fe competitor j impact on value for customers of the s segment and through
the price attribute & in ¢
Calculations: Basis for Comparison between Competitors
ICP;’, : Index of customer in segment s perception of competitor j
Calculations: Result of Investment Conditioned by Share (Reach)
Iq; . elasticity of the quality attribute i for segment s
Ips : elasticity of price attribute k for segment s
Pcs/; _. presence of competitor j in segment s in ¢
TCI", : total competitor index in segment s in ¢
Output: Market share change in units of solution product
MSH;",, market-share of competitor j in segment s in ¢
MST;,; : market-share trend of competitor j in ¢
The model can now be explained as follows: a purchaser (it could be a consumer, but
also a technical or procurement manager) will most likely select a product according to
widely-held perceptions about its quality (Qac;, ) and price (Pac}',) attributes. Examples of
quality attributes include reliability, ease of purchase, scalability, network friendliness,
service availability, and connectivity. Examples of price attributes include rebates,
promotional discounts, cost per instance of use, and channel discounts.
Once the purchaser establishes these preferences for the products of the different
competitors, we can define the baseline perceptions as follows:
Qab',= MIN ; (Qac;,), with j = 1,..., N ()
Pab‘,= MIN ;(Pac;‘,), with j = 1,..., N (2)
Next, we can formalize how much each attribute is able to impact on the value
provided by the product to the purchaser, as follows,
Oajs'= (Qac}:/ Qab\ Tq. (3)
Pa;.',= (Pac},/ Pab‘ yp, (4)
In equations (3) and (4) we assume that a purchaser in a segment will pay special
attention to the attributes of the product most important to that segment. This concept is
formalized through an index of elasticity for each price and quality attributes: Tps and Iq,’
respectively (each elasticity value is calculated through the model calibration process, and
then its value is maintained for the rest of the simulations). Switching costs and other factors
may cause customers to be less responsive to changes in some attributes — this is represented
in the model as the inherent elasticity of a quality or price attribute in a particular segment.
Once the impact on the value provided by each attribute of the product is calculated,
we can formalize an index that compares the value provided by each competitor’s product, as
follows:
L M
1P;.= | Loa, TT Pay (6)
it kt
Assessment of these indexes is not difficult since customer perception of their
products is tracked somehow by most firms [23]. After that, the model simulates behavior for
a given business by showing that the model generates correct changes in individual
competitor market share for changes in value (relative to the competition), which can be
validated by historical data. It is our main assumption that we can thereafter estimate the
share by defined segment for each of the competitors by comparing their customer perception
indices, and by assessing their presence in the marketplace (Pc,’;), as follows:
y
TCI';= Y) Pes's xICP} (6)
ja
MSH;*;= (Pes), XICP,)/TCI*, (7)
Presence of the competitors in the market has to do with their reach in each segment.
Market reach can vary from very monopolistic to very competitive, or even an almost non-
existent reach in any segment.
Equations (6) and (7) are therefore introduced to model competitor market share in a
market where competitive effects are differentially and asymmetrically distributed. Notice
how this model can be considered as a simple attraction model [24] based on the hypothesis
that a competitor market share is equal to its attraction relative to all others (equation 7). In
our case, competitor’s attraction in a segment is estimated by (Pc,’, xJCP;',).
The purchasing behavior model presented here was designed by modeling teams, as
presented in Figure 3, where three important competitors (or competitor proxies, where a
proxy defines a competitive strategy) were considered.
Share here represents the percentage of target market segment sales that can be
expected to flow to each competitor over a given time period, knowing that all the factors are
continuously changing and influencing each other during that time. Overall market size
remains exogenous to the model.
The leverage over time from successful product improvements is shown by the
increasing slope of growth curves over time, typically in the shape of an "S" curve, ramping
from accumulating assets and then tapering off from the effects of diminishing returns.
Presence of COMPETITOR
SharqMSH), Product
j j Presence of
Purchasing Behavior ‘TCL Index<——— COMPETITOR 2 Proxy
— per Segment and Product
Presence of COMPETITOR
Quality Attributes Price Impact on ‘3 Proxy per Segment and
Quality Attribute Impact on Value (Qa) ‘Value (Pa) Product
Elasticity — \ (
Quality Attribute
Baseline Perception Price Attribute Baseline
(Qab) Perception (Pab)
Quality Attribute Price Attribute
Perception per Perception per
cave ee Competitor (Pathg
COMPETITOR 3 Price
COMPETITOR 1 COMPETITOR 3 Attribute Perception Over
Attribute Teeciton Over Attribute Percepeion Over COMPETITOR 1 Price Time
ime Attribute Perception Over
Time
COMPETITOR 2 COMPETITOR 2 Price
‘Attribute Pereeption Over Attribute Perception Over
‘ime
Time
Figure 3. Original team design of the purchasing behavior model
for three competitors.
5. Modeling financial implications of strategy.
How does a product and market strategy impact business revenue? How is revenue
over time linked to the product’s price attributes and profit? To answer these questions, we
set out the variable equations formalization process, after first describing the notations and
definition of the main financial model variables:
Subscripts:
j=1,...,N competitors,
s=1,...,S segments by shared customer purchase priorities (as available)
Input: Segmented Market Data
Te; ; total potential unit sales in ¢
Ss", : size (% of the Tc,) of the segment s in ¢ (Note that this is not a model of building
and creating a market or individual segments, but of capturing and holding segment
share within the strategic market as it grows or shrinks over time, by these
exogenous values.)
Tcs*, ; total potential customers of the segment s in f
Sj*, ; unit sales of competitor j per segment s in ¢
Input: Business Financial Targets / History Allocated to this Solution Product
Sdj;, _. standard discount (% of list price) of competitor j in f
Md jj; .margin discount (% of list price) of competitor j in ¢
Mt;*, : market share (weighted by segment) trend of competitor j in period t
LP ;, +: competitor j list price in ¢
LPi;; : competitor j list price increase in period ¢
LPd ;; : competitor j list price decrease in period ¢
Calculations: Solution Revenue
R’, : revenue of competitor j in period ¢
Input: Cost Ratios
SGA", : selling, general and administrative expenses of competitor / in period t
Cc’, : cost of sales of competitor j, in period ¢
T’, : taxes of competitor j, in period ¢
Output: Bottom-line for operations and product planners
GP’, : gross profit of competitor j, in period t
Calculation and Output: Bottom line for financial planners
NOP’, : net operating profit of competitor j, in period t
COS! : cost of sales factor for competitor j as a percent of revenue
TAX _ : tax factor as a percent of net operating profit
We will now use nonfinancial measures as drivers of financial performance indicators,
which is an assumption considered in many examples of current research in this area’.
The main equation links market share to revenue and profit by reproducing a pro-
forma income statement of the business. (In the equations, we include the competitor and
index (/) to maintain the ability to track more than one competitor financials according to the
model possibilities. All of the businesses we worked with require pro forma statements to
also show associated market share, with as much back up information about target segments
as possible — either as a % goal to be achieved over time that has been set by corporate, or as
the assumed result of the planned operational targets tied to business projections.
* For instance Ittner and Larcker [25] have shown how for 2.491 customers of telecommunications firms,
customer satisfaction indexes could be correlated to revenue levels, retention and revenue changes of the firms
In addition, working with business controllers led us to incorporate sales discounts for
channel incentives (Sd; & Mdj;), cost of sales (COS ’) and tax (TAX) factors, extending
operations targets for individual programs to show front-end investments and contribution to
shareholder metrics. To meet corporate planning guidelines, the business case usually has to
project market share, revenue, and profit metrics, with details for the next 4 quarters and
summary data over three years, Once the unit sales per segment is calculated in equations (8)
and (9), equations (10) to (14) formalize the income statement.
Tes *,= Te, x Ss *; (8)
S*,= Tes, * MSH;"; (9)
s
RI = YS 58, % LP j,X(1-( Sdy+Md,)) (10)
sal
Ci, = R/,x COs! (11)
GP/,= R4,- (Ci,+ SGA4,) (12)
T/, = GP4,x TAX (13)
NOP!, = GP#,- Ti, (14)
In our experience, the financial model is conceived by business planners as shown in
Figure 4. The list price strategy is influenced by the market share trend of the business. For
example, as a matter of pricing policy, a constraint was inserted in one scenario that raised or
lowered the list price if market share projections fit defined gain or loss criteria.
over time. They conclude that their results offer qualified support for recent moves to include customer
satisfaction indicators in internal performance measurement systems and compensation plans [27].
14
Share, Price, Revenue & Profit
Segment Size (Ss
4
Share (MSH) Share Trend
(MST)
Total Potential Potential Customers
Customers(Te) per Segment (Tes)
Price Decrease (LPd)
Standard
Unit Sales (S)
Pico A rt
Margin
Discount(May-———— Revenue ® __iTist Price (Lp)
COS factor <SGA>
a
COS) Cost of Sales (GP Gross Profit GP)
Price Increase (LPi
Tax Factor (TAX)
Net Operating
Taxon) Profit (NOP)
a
Figure 4. Original team design of the financial model.
6. Modeling allowable investments
Planning and tracking targets throughout the fiscal year means calculating the rate of
investment that the business should direct toward a given market opportunity in order to
reach its profit goals.
How do we set up a policy to determine the rate of spending we can accomplish?
What variables should drive decisions about continuing or changing program investments?
Again, in order to answer these questions, we first describe the notations and definition of the
main financial model variables:
Subscripts:
competitors
Calculations: Changes in Financial Variable Values
Rg!, _: revenue growth of competitor / in period t
SGAg’ : growth of selling and general administrative expenses of competitor / in period ¢
Cg’, : cost of good sold growth of competitor j in period ¢
Tg’, : taxes growth of competitor j in period ¢
PC’, profit contribution of competitor j in period ¢
ICF! : investments constraint factor in competitor j
Where channel strategy requires incentives in the form of discounts and payments,
those costs are added to the computation of net sales as a deduction to compute revenue.
In the example scenario that follows, we represent an existing product, and therefore
we assume that conditions to increase investment map closely to changes in the financial
variable values. we first define those value changes in equations (15) to (18), where the
growth in revenue, cost of sales, SGA expenses, and taxes are calculated.
Rgi=R4, Ry (15)
Col =Ci-Cl (16)
SGAg!, =SGA/,-SGA%.1 (17)
TeaT-T14 (18)
Profit contribution growth is defined as the difference between projected revenue
growth, and the sum of the accumulated growth in the other three variables (see equation
(19)).
PCa! =Rg/i— (Ces + SGAg' + Tr) (19)
Finally, SGA expenses for the next year are calculated by considering the profit
contribution, revenue growth and other factors. To illustrate how this is done, let’s see the
example, base in a real case, presented in table 1.
Company X, Profit and Loss Statement
YRI YR2
K$ % Rev KS % Rev || Growth $ | Growth%
Sales| 2000.00 3000.00 1000.00
Standard Discount 840.00 42% 1.260.00 42% 420.00
Margin Discount 60.00 3% 90.00 3% 30.00
Cost of Sales| 825.00 75% 1,320.00 80% 495.00 60%
Gross Profi 275.00 25% 330.00 20% 50.00 18%
Net Operating Profi 534 99 21% 280.00 | 17% 50.00 22%
Before Taxes|
Tax Factot 69.00 6% 84.00 5% 15.00 22%
Net Operating Profi
162.00 15% 196.00 | 12% 35.00 22%
After Taxes}
Table 1. Numerical example for the determination of investments in the model.
Here we show how the model can be used to set target spending levels by mapping the
pro-forma statement ratios, the proposed spending to increase specific attributes, and the
expected returns from a strategy specifically engineered to influence a target segment. In the
example in table 1, profit contribution of Company X, in year 2, could be calculated as
follows:
PC yr2 = Rg yro— (Cg yr2 + SGAg yr + Tg yr2) = 550 (495+6+15)=35 K$ > 0
In Company X, growth in profit contribution is therefore positive, and revenue growth
(in %) is more than three times SGA growth (in %) during the last year (50% >14%). This
seems to be an optimal proportion for Company X to increase its spending. Suppose, for
instance, that when the above conditions are fulfilled, the company grows SGA expenses by
half ((CF=1/2) of the revenue growth (in%), then SGA yr3 would be calculated as follows:
SGA ypr3 = SGA yp (1+ICF( Rg! yp2/ R/yr1))=50(1+0.5(550/1100))=50(1.25)=75 KS.
Then, this example would be formalized as a policy constraint in our model as shown
in equation 20:
SGA, (I4ICF( Rg',/ Ri), if PCgi,>0 and (Rgi,/ R/,.1)>3 *( SGAg!,/ SGA‘)
SGAI 1 = (20)
SGA’, , Otherwise
Again, the investments model in our example, drawn from actual planning scenarios,
was represented with the planning teams as shown in Figure 5, where we find a balance loop
that shows how the rate of growth in profit contribution conditions the growth of the SGA
expenses, while ICF, Rg and SGAg, limit that growth®.
Investments Constraint Factor (ICF)
Investments
_ Growth (SGAg)
Revenue Growth (Rg) = hs | Pa Contributign Growth ive
Taxes Growth (Tg)
cos i Cc
<Revenue (R)> <Cost { Sales (C)> <Taxes (T)>
Figure 5. Original team design of the investments model.
° Again, policy could depend on other variables according to specific business and market conditions. See, for
instance, comments about share in 3" and 4" paragraphs of section 2.
18
7. From an operational investment model to a Decision Support System.
Sections 4, 5 and 6 of this paper showed details within the three sub-models included
in figure 2, containing our general model overview. With respect to system dynamics
modeling, we note the importance of causal diagrams. They helped identify feedback
mechanisms in the sub-models, to visualize how these could impact the way business grows.
The possibility to shift from solely numerical data to a graphical representation provided the
opportunity for dialog and eased mutual understanding, especially for people playing
different roles within the business planning process.
The model as shown allows us to study many product and go-to-market scenarios,
with enough rigor to quickly focus on key assumptions and to build confidence and
consensus when changing business plans.
7.1 A sample of the Decision Support System applied to a business.
We now present an example of the model's use, based on a real instance where
planners considered a possible strategy to improve three product attributes (see Figure 6).
Given a stable organization and product architecture, (rare in high-tech business), we
first validated the model with history across three years, 1995-1998, for consumer products
sold through resellers in a mature market in which the firm was dominant. In that ideal but
unusually stable case, with the product attributes that customers hold most important, and the
attribute elasticity in each segment, we were able to replicate the market response to attribute
investments during those years with reasonable accuracy. Using the model for this scenario,
planners wanted to know whether or not to continue the same rates of spending increase for
the same three attributes over the next three years, as shown in Figure 3, assuming that these
would result in the same kind of increase in value perceptions for those three attributes. They
wanted to know what kind of business results they could expect and why, in order to justify
the rate of spending they would fund to meet their growth objectives.
Best ——t-— +--+ 4
Base “2-2-2 agg
Perception of Quality Att.[Scalable]
1.25 "1
1 —
Perception of Quality Att.[Network friendly]
1,25 can
1 2-t
Perception of Quality Att.[Easy to Purchase]
2
1.75
1.5
1.25 a
1 gre
1995 1998 2001
Time (year)
Figure 6. Base and best case of attribute improvement.
The scenario results shown below in Figure 7 tell us that there is an incremental gain
in market share in all three market segments we are considering, but that the gain is greatest
in the segment called “small”, where we may reach close to a 5% gain in market share.
Segment “Soho” responds very little, and there is only a small gain in segment “Medium”. In
calculating the profit impact, the model uses input from planners about segment growth,
share size, competitor strategy, segment elasticity, and expected price or cost changes, all
happening at the same time. The planning team, which includes marketing, engineering,
finance, supply chain management, and division executives, reviews the scenario output to
understand the results, confirm underlying assumptions, and agree where they would redirect
spending.
In addition to being able to "drill down" to underlying causes, the scenario shows the
net operating profit the business could expect, if those investments are accomplished, given
the corresponding yearly increase in SGA (see Figure 8). The business planners used this
outcome analysis to adjust their planned investment program.
20
i a a a a
Base -2"~ Y DnaY onetime eer
Marketshare per Segment[Soho]
90
85
80 oar ama ae a
75
70
Marketshare per Segment[Small]
60
57.5
55 ]
52:5
50
Marketshare per Segment[Medium]
60
575 ttt
55
52.5
50
1995 1998 2001
Time (year)
Figure 7. Base and best simulations for market share improvement (notice that last two
graphs are in different scale than the first one).
Best t + t + t t 1 Best t t + t + t
Base Base Sr ree
Net Operating Profit SGA Increase
200 M 6M
150M 45M
100 M 3M
50M 1.5M
0 0
1995 1998 2001 1995 1998 2001
Time (year) Time (year)
Figure 8. Base and best simulations, expected NOP (after taxes) and
increase in SGA per year.
21
7.2 From a simulation model to a Decision Support System
Now we give an example of how this model can take advantage of optimization
techniques to compare alternative attributes investments, converting the model into a fast
Decision Support System. A modified Powell method is used to carry out these optimizations
properly. This is a direct-search numerical optimization technique which does need not to
evaluate the gradient, and which is very suitable for the analysis of dynamics of complex
nonlinear control systems. This technique is well known among direct-search methods, to
derive a very fast convergence”
In this example, we use the model to select an investment focus for the next three
years. From the list of attributes that represent the decision factors for solution customers, we
compare the impact of improving each attribute by a planned percentage. Each column in
Table 2 evaluates the choices according to a specific criteria. The table is used one column at
a time, with each one representing a different planning scenario.
The table ranks the attributes according to each criteria, while in the last rows, the
associated percentages tell us how much better the highest level of ranking is than each of the
lower level ranks.
In order to optimize their investments, the planners first define the criteria for
optimization to be used in each scenario, the calculation represented here by the column
headings. Planners normally consider more than one criteria, and usually include both
financial and market-based metrics, representing various business objectives for the planning
period. They then define optimization variables for the ranking calculations (first column).
Later, we can ask the model, by using multi-parametric optimization (considering cumulative
evaluation of the payoff), which attributes should be the spending priority to best meet the
business criteria shown in the column heading.
7 The basic idea behind Powell's method [26] is to break the N dimensional minimization down into N separate
1D minimization problems. Then, for each 1D problem a binary search is implemented to find the local
minimum within a given range. Furthermore, on subsequent iterations an estimate is made of the best directions
to use for the 1D searches. Some problems, however, are not always assured of optimal solutions because the
direction vectors are not always linearly independent. To overcome this, the method was revised [27] by
introducing new criteria for formation of linearly independent direction vectors; this revised method is called
“The Modified Powell Method”.
22
0)
3.
N
5
g
s
a.
Ey
o
<.
=
se
o
2
s
2
i=
z
3
Es
a
Ss
EF
5
3
&
Ss
2
i
Share in Revenue &
SVE SEPP Segment x Product Y Share
Attributes
Reliability 1 2 6 4 2 1
Easy to purchase 1 2 6 3 2 2
Scalability 2 2 6 3 2 3
Network friendly 2 2 6 3 2 3
Service abailability 2 2 5 3 1 3
Connectivity 3 2 4 2 3 3
Plug and play 4 3 2 4 4 4
ete 5 4 4 4 4 4
Importance
(1) more than ( 2) 7% 14% 20% 13% 12% 5%
(1) more than (3 ) 14% 20% 60% 15% 20% 16%
(1) more than ( 4 ) 78% +% 260% +% +% +%
(1) more than (5 ) 545% 340%
(1) more than (6) 400%
Table 2. Example of output from the Decision Support System
(not case described above).
The way to read the table is as follows: if we pursue maximizing Revenue, first
choice for attribute investment should be Reliability and Easy to Purchase, if you increase
perception of either of those attributes by a targeted percent (which you assume you will do if
you spend according to your plan), your results in terms of revenue will be 7% higher than in
Scalability, Network Friendly, or Service Availability, and 14% higher than in any
attribute with a 3, and so on.
The model here becomes a powerful and flexible planning tool. You may even
explore multiple objectives (see last column that includes both Revenue & Share). The same
planning team might go on to explore other investments goals, like increasing the presence in
various segments. This constitutes a Decision Support System that adds clarity and rigor to
targets and product / program strategy.
23
8. Managerial implications
As we mentioned and characterized in section 2, managers in high-tech markets face
unique challenges.
Respond to Market-Driven Demand
Business planners represent the needs of engineering, marketing, sales, order,
delivery, support, and service teams. They face changes driven by technological advances,
volatile demand, global competition, emerging standards, and significant uncertainty about
what drives their customer's decisions to buy.
This decision support system views the business as a dynamic feedback system to:
1. Sense an opportunity matched with an ability to respond with value
2. Create value — balancing features and price — and communicate that value to
customers in a target segment
Grow with the market, faster than the competition
4. Create early barriers to entry for emerging markets
Confidently redirect resources based on changes in customer purchasing behavior,
competitor investment, and the payback that can be expected from the required
additional investment
Segment According To Customer Purchase Priorities
Wherever markets are segmented by customer value and buying behavior, decision
makers may use this model to compare expected financial returns on alternative investments
that appeal to some segments more than others. Investments that affect a specific attribute
have different implications for each segment, with results for share, revenue, and profit that
also reflect external changes in size of that targeted segment and of the market demand
overall.
24
Specific investments considered by teams with whom we have done these analyses in
the past include: reseller discounts, pricing strategy, one-to-one relationship marketing
programs, advertising to raise target customer awareness, new channel development, new
product and technology introductions, introduction of non-branded offerings, forward
contracts to secure critical part supply, and collaborative communication backbones for
demand and fulfillment chains.
Focus on the Vertical Dimension of Business Planning
There is only one "product" in our model, but in high-tech sectors like telecom
infrastructure or medium business manufacturing, the end "product" is a solution, i.e.
multiple component products with different cost structures bundled for this market to meet
this set of attributes.
Financial targets usually represent product businesses selling into numerous markets,
where go-to-market, sales, service, and channel investments are treated as programs, charged
with achieving specific market objectives. Although current financial data usually comes to
us as product business targets, most critical investment decisions must also consider the
impact of changes in attributes and customer perception of value for a solution which will
determine its success or failure.
Traction from Precise Go-To-Market Strategy
Initiatives to improve business performance are directed toward specific solution attributes:
© Quality attributes are improved by investments to improve features,
performance, power requirements, footprint size, integration, customization,
delivery, localization, scalability, interoperability, quality, channels, and
alliances.
o Price attributes are improved by investments in aggressive sourcing, parts
availability, risk management, order and forecast management, channel
incentives, discounts, rebates, advertising, web-based collaborative
infrastructure, and synchronized product upgrades.
25
The critical assumption, that your planned spending will indeed increase customer
perception and impact sales as you expect, needs to be confirmed as quickly as possible. In
addition to mining existing market research, our planners tended to gain confidence through
immediate action guided by the decision support system, with a rapid "pilot", limited in scope
and carefully observed to measure and confirm perception and response. Thus, for today's
high-tech businesses, strategy and tactics tend to merge, each informing the other in a rapid
exchange between precise action and useful learning.
9. Conclusions and further research.
In this paper we have described how we have used simulation models to support
product and marketing investment decisions. We have presented a high level model structure,
and described formalization of three sub-models for purchasing behavior, financial results,
and investments for growth. We have shown how the model is used in business planning to
explore a specific problem, and given one example of the model's value as an “engine” of a
Decision Support System.
The Decision Support System that we have defined takes into account the horizontal
and vertical metrics that together define success for current high-tech businesses, matching
each investment strategy to specific attributes of customer value and business results. At the
same time, we incorporate within the model structure other important characteristics of high-
tech markets that are just emerging but will soon be factors in business investments decisions.
System dynamics simulations greatly improve analysis of go-to-market strategies,
integrating customer knowledge with simulations to analyze spending trade-offs in features,
services, support, integration, channel incentives, pricing, and advertising. The payback over
time is the shown in the output from this formal system dynamics model, a powerful DSS
tool offering the opportunity to compare strategies for a segmented market, under different
scenarios, with customized metrics.
26
References
[1] Bhatt G.D., Zaveri J. 2002. The enabling role of decision support systems in organizational
learning. Decision Support Systems. 32(3). 297-309.
[2] Little J.D.C. 1979. Decision support systems for marketing managers. Journal of
Marketing. 43. 9-26.
[3] Higby M.A., Farah B.N. 1991. The status of marketing information systems, decision
support systems and expert systems in the marketing function of U.S. firms. Jnformation
and Management. 20. 29-35.
[4] Van Campen P.A.F.M., Huizingh K.R.E., Oude Ophuis, P.A.M., Wierenga, B. 1991.
Marketing Decision Support Systemen bij Nedelandse Bedrijven. Eburon. Delft.
[5] Li E., McLeod, Jr R.. Rogers J.C., Marketing information systems in Fortune 500
companies: a longitudinal analysis of 1980, 1990, and 2000. /nformation and Management.
38 . 307-322
[6] Shaw J., Subramaniam C., Tan G.W. and Welge M.E, 2001. M.J. Shaw, C. Subramaniam,
G.W. Tan and M.E. Welge , Knowledge management and data mining for marketing.
Decision Support Systems. 31 , pp. 127-137
[7] Wierenga, B. & Oude Ophuis, P.A.M. 1997. Marketing decision support systems:
Adoption, use and satisfaction. International Journal of Research in Marketing. 14.. 275-
290.
[8] Van Bruggen, G.H., Smidts A., Wierenga B. 1998. Improving decision making by means of
marketing decision support system. Management Science. 44. 645-658.
[9] Simon H. 1994. Marketing science’s pilgrimage to the ivory tower. In: Laurent, G., Lilien,
G.L., Preas, B. (Eds.). Research Traditions in Marketing. Kluwer Academic Publishing,
Boston, MA. 27-43.
[10] Little, J.D.C., Lodish, L.M., Hauser, J.R. and Urban, G.L., 1994. Commentary. In:
Laurent, G., Lilien, G.L. and Pras, B., Editors, 1994. Research Traditions in Marketing,
Kluwer Academic Publishing, Boston, MA, 44-51.
[11] Parsons, L.J., Gijsbrechts, E., Leeflang, P.S.H. and Wittink, D.R., 1994. Commentary. In:
Laurent, G., Lilien, G.L. and Pras, B., Editors, 1994. Research Traditions in Marketing,
Kluwer Academic Publishing, Boston, MA, 52-78.
[12] Kim N., Bridge, E., Srivastava, R.k. 1999. A simultaneous model for innovative product
category sales diffusion and competitive dynamics. Jnternational Journal of Research in
Marketing. 16. 95-112.
[13] Bridges E., Ensor K.B., Thomson J.R. 1992. Marketplace competition in the personal
computer industry. Decision Science. 23(2). 467-477.
27
[14] Vollerthun, A. 2002. Design-to-Market: Integrating Conceptual Design and Marketing.
Systems Engineering. 5(4).315-326.
[15] Forrester J.W. 1975. Collected papers of Jay W. Forrester. Wright-Allen Press.
Cambridge. Massachusetts.
[16] Roberts E., (Editor).1984. Managerial Applications of System Dynamics, MIT Press,
Cambridge.
[17] Senge P. 1991. The fifth discipline: The Art and Practice of the Learning Organization. ,
Double Day, New York.
[18] Morecroft J.D., Senge P., Forrester J.W. 2000. Modeling for Learning Organizations.
Productivity Press. Portland. Oregon.
[19] Sterman, J. (2000) Business Dynamics: Systems Thinking for a Complex World.
McGraw-Hill. Irwin.
[20] Hogarth R.M., Makridakis S. 1981. Forecasting and Planning: An evaluation.
Management Science. 27(2), 115-138.
[21] Day G.S. 1990. Market driven strategy: Processes for creating value. The Free Press. New
York, NY.
[22] Schnaars S.P. 1991. Marketing Strategy: A Customer-Driven Approach. The Free Press.
New York, NY.
Back to the T
[23] Ross J. and Georgoff D.. 19 —_—_—_=iuuwductive and Quality Issues in
manufacturing. The State of the Industry. Industrial Management. Jan-Feb. 3-5, 22-25.
[24] Carpenter G.S., Cooper L.G., Hanssens D.M. and Midgley D.F. 1998. Modeling
Asymmetric Competition. Marketing Science. 7(4). 393-412.
[25] Ittner C.D. and Larcker D.F.1998. Are Nonfinancial Measures Leading Indicators of
Financial Performance? An Analysis of Customer Satisfaction. Journal of Accounting
Research. 36(suppplement). 1-31.
[26] Powell M. J. D. 1964. An Efficient Method for Finding the Minimum of a Function of
Several Variables Without Calculating Derivatives. Computer Journal. 7(2). 155-62.
[27] Powell M. J. D. 1968. On the calculation of orthogonal vectors. Computer Journal. 11(2).
302-304.
28
Back to the Top
CC BY-NC-SA 4.0