Modeling Business Process Variability for Design-Time Configuration

Introduction

Some business processes tend to recur in different organizations or even in different industries. For example, process analysts often use the term order-to-cash to refer to a business process that starts from the moment a purchase order is received by a supplier, to the moment this purchase order has been fulfilled (and the supplier has received the corresponding payment). Virtually all order-to-cash processes include activities related to invoicing, delivery and payment. However, variations can be observed across order-to-cash processes. For example, an order-to-cash process for the delivery of goods (e.g. delivery of office supplies) is different from an order-to-cash process for the delivery of services (e.g. delivery of consultancy services). In the first case, there is a physical delivery that happens at a discrete point in time and the condition of the goods can be checked upon receipt. On the other hand, the delivery of a service may occur over a long period of time (say 6 months). Over this period, several invoices may be issued for the same original purchase order. Also, checking the quality of a consultancy service is often trickier than checking the quality of a box of reams of paper. Not surprisingly, the corresponding order-to-cash process models will have many differences.

But despite such differences, companies have a lot to learn from each other when it comes to analysing and re-designing their order-to-cash processes. It would be inefficient if every time a company wants to model its order-to-cash, it did so completely from scratch, without consideration for how other companies perform their order-to-cash process. In this setting, this chapter deals with the following question: How to model business processes that are similar to one another in many ways, yet differ in some other ways from one organization, project or industry to another? If we can do so, it then becomes possible to capture multiple order-to-cash processes in a single model. This combined order-to-cash process model can then be used as a starting point to derive order-to-cash process models for specific companies.

This idea is captured by the concept of reference process model. A reference process model combines a family of similar process models together. A reference process model is designed in a generic manner and is intended to be configured to fit the requirements of specific organizations or projects. Thus, it is an alternative to designing process models from scratch.

In this chapter, we will use examples taken from the film industry, in particular from the post-production phase of a screen project. Figure 1 shows two process models for screen post-production: shooting on Tape and shooting on Film. The modeling language used in this figure is BPMN (cf. Chapter X). These process models share some commonalities, represented by the first two activities. Whether the movie is shot on Tape or on Film, post-production always starts with the preparation of the footage for edit, followed by the Offline edit. After this activity, the two practices differ in the way the edit is completed - Online if the footage is Tape, or Negmatching if the footage is Film. Online edit is a cheap editing procedure that well combines with low-budget movies typically shot on Tape. On the other hand, if the movie is shot on Film on a high-budget production, it is preferable to carry out a Negmatching instead of an Online edit, as the former offers better quality results, although it requires higher costs. This represents a variability in post-production. Depending on the type of project, one option or the other will be used.

Figure 1.

A reference process model is an integrated representation of several variants of a process model

A reference process model for screen post-production may combine both of these options, by merging the commonalities and capturing the variability by means of variation points. A variation point, depicted by a special OR-gateway on the right-hand side of Figure 1, is a point in the process model in which multiple variants exist and a decision needs to be taken of which variant to use. The selection of the most suitable variant is called configuration. Once all the variation points have been configured, the reference process model can be transformed into a derived model (e.g. by dropping the variants that are no longer needed), through a process called individualization.