لطفا منتظر باشید ...
AN APPLICATION EXAMPLE
The example described below is about a Customer Relationship Management (CRM) application, where data mining and system dynamics tools are used. A similar approach can be extended for any other business application. Reddy (2001) talks about three basic processes in CRM: acquisition (of new customers), retention (of the most profitable customers), and enhancement (of the customer relationship). Supporting these processes can be facilitated with a better understanding of customer behavior; however, the existence of many intangible factors (e.g., customer satisfaction, product features, brand awareness, etc.) and the difficulty in measuring them adequately is a challenge for a successful application of CRM. Figure 4 shows a simplified system dynamics model of a CRM application related to a Brand A, where the processes of customer acquisition and retention are represented. Acquisition is made through sales and marketing or through the phenomena of “word of mouth,” and retention is related to a loss of customers rate.
Figure 4: Simplified System Dynamics Model of a CRM Application
The average customer satisfaction index for Brand A is calculated through surveys and other data (for example, frequency of repeated purchases); then, data-mining tools provide a set of rules or equations that define mathematical relationships between customer satisfaction and acquisition (or loss) of customers in the form of functions that can be integrated in expressions like the following examples:
(1) acquisition of customers by word of mouth = customers of brand A * potential customers to actual customers referral ratio * function for acquisition of customers by word of mouth(average customer satisfaction index)
(2) loss of customers = customers of brand A * function for loss of customers(average customer satisfaction index)
In the equations (1) and (2), “customers of Brand A” is the amount of actual customers of Brand A, and “potential customers to actual customers referral ratio” is the estimated proportion of potential customers that could be referred by actual customers. The “function for acquisition of customers by word of mouth” and the “function for loss of customers” are derived from datamining processes. In a similar fashion, another function for the acquisition of customers through sales and marketing is obtained incorporating variables related to the well-known components of the “marketing mix”: price, product, promotion, and place (Kotler, Armstrong, & Chawla, 2003). As a better understanding of the cause-and-effect relationships of the CRM process is achieved, such function can be reformulated as a new set of stock-and-flow diagrams that include, if needed, a new set of functions obtained via data-mining methods. To clarify these concepts, a simplified example is described.
We suppose that the “function for acquisition of customers through sales and marketing” involves, in addition to other marketing mix-related variables, the amount of money spent on advertising (X) and the amount of stores having Brand A in stock (Y). Through a neural network algorithm, a function of the form Z = aX + bY + c (where a, b, and c are constants) is obtained, and thus we know a mathematical relationship between such variables. However, this expression does not explain conceptually the real effect and impact of X and Y on Z. To overcome this issue, we could incorporate the principles of an approach known as “Dynamic Brand Value Management” or DBVM (Desmet et al., 1998), which takes into account the factors that influence brand value and also dynamic elements such as time delays and feedback loops. Figure 5 shows the redesigned model for the CRM application, where a “virtuous reinforcing loop” is visible: when a product sells well (see variable “sales”), more retailers are motivated to order it (see variable “stores deciding to order Brand A”), and the product sells even better.
Figure 5: Redesign of the Model in Figure 4 to Incorporate Concepts of the Dynamic Brand Value Management Framework
As more valid cause-and-effect relationships are added, the model will provide more insights for the CRM processes, and it will be possible to apply the intrinsic knowledge contained here to formulate new strategies when a change in the underlying conditions occur, and when there is not enough available data to mine.
