|
COPYRIGHT 2001 Institute for Operations Research and the Management Sciences
As demand for online network services continues to grow, service providers are looking to meet this need and avail themselves of business opportunities. However, despite strong growth in demand, providers continue to have difficulty achieving profitability, customer churn remains high, and network performance continues to draw complaints. We suggest that strategic business planning for network services would benefit from a systems thinking approach that analyzes the feedback effects present in the underlying business process. These feedback loops can be complex and have significant impact on business performance. For instance, while the size of a provider's customer base depends on price and network performance, network performance is itself dependent on the size of the customer base. In this paper, we develop a planning model that represents these feedback effects using the finite difference equations methodology of systems dynamics. The model is validated by showing its fit with essential characteristics of the underlying problem domain, and by showing its ability to replicate observed reference mode behaviors. Simulations are then carried out under a variety of scenarios to examine issues important to service providers. Among other findings, the simulations suggest that (a) under flat-rate pricing, lowering price to increase customer base can hurt profitability as well as network performance; (b) under usage-based pricing, lowering price need not necessarily lead to a larger customer base; and (c) in addition to price, the customers' threshold of tolerance for performance degradation plays a significant role in balancing market share with profitability. We briefly present a prototype decision support system based on the systems thinking approach, and suggest ways in which it could be used to help business planning for network services.
(Online Services; Systems Dynamics; Business Performance; Decision Support)
1. Introduction
Networks are playing an increasing role in business, social, and political activities, fueled in part by growth of the Internet. The worldwide growth of the Internet coupled with development of user-friendly browser technology and standards such as XML (POET 1998) has led to intense interest in its use for electronic commerce (Rao et al. 1998, Keeney 1999). The Internet has also had significant impact on educational, social, and political activities worldwide (Hauben and Hauben 1997, Petrazzini and Kibati 1999). It serves as a forum for discussion of societal issues and as a means of delivering government services. It is not surprising to find that the demand for network services has exploded and is predicted to grow rapidly through the near term (Hoffman et al. 1996).
On the supply side, Internet Service Provision (ISP) has become a significant business activity attracting a variety of vendors and technologies (Lange 1998, Riezenman 1998). The quality of service and price tends to vary across providers, and these factors have a significant effect on customer satisfaction. (1) Of the roughly 4,000 Internet service providers, few are showing any profits, operating margins are getting tighter, and competition is increasing (Fattah 1998). While flat-rate pricing--"all you can use for $19.95"--may attract more customers, many ISPs do not have the financial strength to invest in new capacity to accommodate this growth in customer base. Smaller providers often face a dilemma--grow the customer base but remain unprofitable, or stay small and achieve profitability. Even large providers like Netcom and AOL face similar pressures.
The preceding observations are suggestive of the complex interactions that characterize the underlying business process. A few examples of specific interactions follow. For instance, low prices attract more customers, which generates more network traffic. As traffic levels rise, network performance drops, inviting customers to switch to competitors. To improve network performance, however, the provider must invest in additional capacity, but the lower prices have a negative impact on its ability to pay for capacity. In short, while customer behavior, network performance, and financial consequences may each be easy to characterize in isolation, there is feedback and interaction among them. This complicates the business-planning process considerably, and makes it much more difficult to gauge the impact of management decisions on business performance. An integrated view is therefore needed to gain a better understanding of the business process underlying network service provision. If such a view can be captured in a computer-based model, it can also provide the basis for ongoing decision support (Swami 1995, Tumay 1996, Liles and Presley 1996).
In this paper we develop a basic business model of network service provision using the finite difference equation methodology of systems dynamics. Because the emphasis is on capturing key interactions in the business process, individual structural components are represented in aggregated form. Simplifying assumptions and aggregations will be noted as the model is described in detail. The distinctive features of the model are:
(1) Its structural characteristics capture interactions among customer behavior, financial performance, and network operations inherent in network service provision. It is a systems model that takes an integrative view of the business.
(2) The model can be simulated to examine the dynamics generated by these interactions.
Such an integrated systems model can be expected to contribute in at least two ways. First, it offers a reasonably holistic synthetic environment within which to examine questions of importance to network service providers. For instance, effects of price changes or growth strategies may be studied with the knowledge that business performance behavior generated by the model includes the complex interactions among major functional areas. Once calibrated for a specific organization, such a systems model may also be used to provide decision support for specific business-planning activities. Section 2 develops the model in top-down fashion. Validation is carried out by first showing the fit between model structure and characteristics of the real operating environment. Subsequently, we show the model's ability to replicate actual observed reference behavior modes. In [section] 3, this validated model is used as a simulated environment within which to gain insights into issues facing service providers. Section 4 briefly presents a prototype decision support system based on this model, and discusses how it may be used in support of business planning. Limitations and extensions of the model are discussed in conclusion.
2. A Basic Business-Planning Model
We have used the well-known finite difference equation methodology of systems dynamics (SD) (Forrester 1961, Goodman 1974) to develop a basic business-planning model for network service provision. There are, or course, different ways to model the dynamic behavior of systems, each with its own strengths and weaknesses. SD is known to be particularly well suited for high-level process modeling (Wolstenholme 1990) where process scope needs to be wide, but the level of detail is somewhat aggregated. The business-planning process being studied requires precisely such high-level modeling. Further, the business process being modeled here has quantifiable variables such as price and capacity, as well as hard-to-measure variables such as customer perception of quality. The systems dynamics approach facilitates the representation of both types in one model. There are other advantages to SD associated with the process of model building in a cross-functional context (Wolstenholme 1990) that make it an attractive vehicle for building the model as a decision support tool. For purposes of this paper, it is sufficient to note that SD has the following major characteristics:
(1) SD uses stocks and flows to model organizational processes. Stocks represent accumulations in the system. Both physical and nonphysical variables can accumulate. (2)
(2) Flows connect pairs of stocks and cause changes in stock levels. They obey the laws of flow conservation--i.e., reduction in one stock results in an equivalent increase in the second.
(3) Connectors convey information only, and information flows are not conserved. They can serve to control physical flow rates.
(4) Converters are used to hold inputs, outputs, intermediate values, and to perform computations. They do not accumulate.
(5) Feedback effects can be captured in SD models and they play an important role in determining dynamic behavior.
(6) SD can model closed as well as open systems, the having no feedback.
(7) Different forms of delay and nonlinearity can be captured in SD models. Most systems have some form of nonlinearity that bounds system behavior in the long run.
SD models are generally represented graphically using standard symbols for stocks, flows, connectors, and converters. The underlying model is a system of finite difference equations. This mapping between the graphical and mathematical representations will be shown shortly. Models are simulated by iteratively evaluating the system of equations.
2.1. A High-Level Model Description
The SD model is presented in a top-down manner. The high-level description to be presented in this section shows major structural components and feedback effects among them. Each structural component will be described in detail, together with an explanation of its fit with essential characteristics of the underlying problem domain. Such justification is necessary as a first step towards showing model validity. The standard SD symbols are reproduced in Figure 1 and the high-level model appears in Figure 2.
[FIGURES 1-2 OMITTED]
To facilitate identification of components in systems diagrams, we will use superscripts S, F, Cn, Cv, for stock, flow, connector, and converter variables, respectively. To get an integrated view of service provision, we begin with new customers signing up with the provider. New customers increase its customer base. Customers generate traffic on the provider's network, experience a certain quality of service, and pay fees that generate revenue. Figure 2 captures these three basic components of service provision--customers, the network, and finances--as follows. The provider's customer base is represented by [TotCust.sup.S]. In each period, a certain number of new customers sign up while some existing customers leave, represented by the flows [NewCustomers.sup.F] and [DepartingCustomers.sup.F], respectively. The difference between these two flows represents net additional customers added in a period. Over time, if the two flows balance, [TotCust.sup.S] levels off to a steady value. When inflow exceeds outflow, [TotCust.sup.S] increases and vice versa.
Similarly, network capacity is represented by the stock [Network.sup.S]. The flow into this stock, [CapExpansion.sup.F], is the result of investments in additional capacity and other network enhancements. In this basic model, capacity is never removed; hence there is no outflow from [Network.sup.S]. Financial operations are also represented in a simple manner by a single stock, [RetEarnings.sup.S]. There is a periodic inflow of [Revenues.sup.F] driven by customer base and price, (3) and a periodic outflow of [Expenses.sup.F] associated with operational costs and reinvestments in the network. The difference between the two flows accumulates as the provider's retained earnings.
Even without further detail about the three components, it is easy to trace feedback loops in Figure 2, highlighting the complexity of business planning. Following connectors (the thin directed lines) in Figure 2, one can trace a positive feedback loop from [TotCust.sup.S], to [Revenues.sup.F], on through to [Network.sup.S] and back again to [TotCust.sup.S]. For a given price, an increase in [TotCust.sup.S] increases [Revenues.sup.F], which in turn permits an increase in [Network.sup.S]....
Read the full article for free courtesy of your local library.
|
Bookmark this article
Print this article
Link to this article
Email this article
Digg It!
Add to del.icio.us
RSS
