An integrated approach based on execution measures for the continuous improvement of business processes realized by services

Abstract

Context

Organizations are rapidly adopting Business Process Management (BPM) as they focus on their business processes (BPs), seeing them to be key elements in controlling and improving the way they perform their business. Business Process Intelligence (BPI) takes as its focus the collection and analysis of information from the execution of BPs for the support of decision making, based on the discovery of improvement opportunities. Realizing BPs by services introduces an intermediate service layer that enables us to separate the specification of BPs in terms of models from the technologies implementing them, thus improving their modifiability by decoupling the model from its implementation.

Objective

To provide an approach for the continuous improvement of BPs, based on their realization with services and execution measurement. It comprises an improvement process to integrate the improvements into the BPs and services, an execution measurement model defining and categorizing several measures for BPs and service execution, and tool support for both.

Method

We carried out a systematic literature review, to collect existing proposals related to our research work. Then, in close collaboration with business experts from the Hospital General de Ciudad Real (HGCR), Spain, and following design science principles, we developed the methods and artifacts described in this paper, which were validated by means of a case study.

Results

We defined an improvement process extending the BP lifecycle with measurement and improvement activities, integrating an execution measurement model comprising a set of execution measures. Moreover, we developed a plug-in for the ProM framework to visualize the measurement results as a proof-of-concept prototype. The case study with the HGCR has shown its feasibility.

Conclusions

Our improvement vision, based on BPs realized by services and on measurement of their execution, in conjunction with a systematic approach to integrate the detected improvements, provides useful guidance to organizations.

Introduction

The complexity and size of the current systems to support the business of organizations has grown considerably in recent years, thus increasing the difficulty in managing them properly. Business Process Management (BPM) [1], [2], [3] is becoming increasingly important for those organizations which need to gain a better insight into the way their business processes (BPs) are executed. BPM helps organizations to manage their BPs, assisting them in checking that their outputs are maintained in the range defined as successful with respect to the business goals of the organization. BPM provides the means for guiding and supporting the modeling, implementation, deployment, execution, and evaluation of BPs in an organization, based on the BP lifecycle [1], which establishes the main phases and activities that organizations have to carry out in their efforts to manage BPs: Design & Analysis, Configuration, Enactment and Evaluation of BPs. In the Design & Analysis phase, BPs are first identified and modeled and then validated and verified; in the Configuration phase BPs are implemented in the chosen technology, and then tested and deployed in the selected platform. The Enactment phase involves the execution and monitoring of BP instances and the registration of execution data in execution logs. Finally, in the Evaluation phase execution log data is evaluated to provide insight into the real execution of the BPs using, for example, process mining techniques. The boom in the use of BPM to support the BP lifecycle introduces both a fresh approach to business management and new challenges for the undertaking of management efforts. One of these challenges is that without proper guidance for applying BPM, the results are barely predictable, with the drawbacks that this entails.

According to the Gartner Group [4] “organizations carry out BPM projects in order to improve one or more business processes”. This places BP improvement as the number one motivation for BPM. In the same survey, the top five business goals identified include: improving customer satisfaction, improving the quality of BPs, reducing costs, improving BP agility, and supporting continuous process improvement. Measurement activities are implicit in the lifecycle, but to guide the integration of execution measurement through the whole lifecycle we must define measurement activities explicitly. This will help obtain insight into the real execution of BPs, knowledge that is needed for BP improvement.

The implementation of BPs by means of different systems in the organization also affects the way business people can perform the defined activities and how they perceive the software support provided. The traditional vertical vision of Information Technologies (IT) for implementing BPs, based on sections or areas of the organization, made the modification of BPs a challenging activity that required the integration of various heterogeneous information systems. These changes are not easy to introduce, in general, due to the implicit implementation of the BPs in the systems supporting them. Among other disadvantages, this often leads to more time being spent than initially planned. In addition, it may result into high costs and unfulfilled expectations of the business area regarding the functionality offered by the implemented BPs [5], [6]. Service Oriented Computing (SOC) [7] provides the basis for defining services that can implement parts of BPs (activities, sub-processes) or even a BP as a whole, by introducing an intermediate layer of services between BP definition and their implementation by means of different technologies. This approach helps bridge the so-called business-systems gap caused by different views and expectations between the business and the IT side when introducing changes into BPs. The service layer enables us to separate the specification of BPs in terms of models from the technologies implementing them, thus improving their modifiability by decoupling the model from the implementation in the technology selected [5], [6]. The approach provides the basis for integrating changes with minimum impact, both at the BPs level and at the system level, allowing for the organizational agility needed to respond to new demands or corrective measures [5], [6].

In this article we present an approach for the improvement of BPs which extends the traditional BP lifecycle [1] with measurement and improvement activities. We have called this BP Continuous Improvement Process (BPCIP). Our proposal includes a model that integrates execution measures for BPs with services implementing them in a comprehensive way; we have called this model BP Execution Measurement Model (BPEMM). In previous work we presented an initial definition of the BPCIP [8], which we have extended significantly by redefining the complete BPCIP lifecycle based on the feedback from the conference and by completing the set of execution measures to be integrated into the BPEMM.

Both the BPCIP and the BPEMM are part of a larger framework called MINERVA [9], [10], which we defined with the aim of providing an integrated approach to support the continuous improvement effort in an organization based on the realization of BPs by services with a model driven approach. Our purpose was also to enable BP execution measurement, as far as both BPs and services are concerned. The framework is organized in three dimensions: conceptual, methodological and tool support. Previous contributions described the MINERVA framework as a whole [9], [10], the conceptual [11] and the tool [12] dimensions, as well as part of the methodological dimension: the BPSOM methodology and the model-driven approach [13], [14]. We will not deal with these aspects in detail here. Contributions of this paper concern the support for measurement and improvement of BPs that are part of the methodological dimension of MINERVA. What this article contributes therefore, is as follows:

• The redefined BPCIP lifecycle, based on the BP lifecycle [1] and extended with measurement and improvement activities, as well as the elements to carry out the defined activities, such as roles and input and output artifacts.

• The complete set of integrated execution measures defined in the execution measurement model BPEMM, along with the addition of a “cube” presentation for the tridimensional taxonomy we have defined to organize the measures.

• A proof-of-concept prototype of the ProM BPEMM plug-in, to support the visualization of the results of the execution measurement of measures defined within the BPEMM, as well as to demonstrate the feasibility of our definitions.

• The validation of our approach by means of a case study undertaken in the Hospital General de Ciudad Real (HGCR), with the help of business experts.

“Measurement is the first step that leads to control and eventually to improvement. If you can’t measure something, you can’t understand it. If you can’t understand it, you can’t control it. If you can’t control it, you can’t improve it.” H. James Harrington (Harrington, 1991) [15].

Continuous process improvement refers to a status in which the organization is continuously analyzing the way it carries out its business. Its goal is to find improvement opportunities for the performance of the organization’s BPs [16], [17]. Measurement of BP execution provides the basis for analyzing the real behaviour in the organization. It helps to detect deviations from planned behaviour, as well as to discover improvement opportunities for the BPs. Once improvement opportunities are detected, organizations need to define changes in the BPs that will lead to a new improved version of these for better achieving the business goal set by the organization. Execution measurement then becomes the enabler towards understanding and controlling the real occurrences of BPs in the organization, establishing an ongoing BP improvement culture [16]. To improve processes continuously, several aspects have to be taken into account. A key one refers to the creation of an organizational improvement context, which comprises a number of elements including business and software teams committed to the improvement initiative and a systematic approach for conducting improvement efforts, the explicit specification of BP models and the software implementing them, a definition of measures collected during BP execution, or techniques and tools to enable the evaluation of the collected execution logs. One key statement of process management is that quality of products and services is largely determined by the quality of the processes used to develop, deliver and support them. An effective process is capable of bringing people, tools and methods together into an integrated whole which produces the expected outcomes [16].

An improvement approach has to support the identification of process deficiencies and provide guidance for introducing improvements in a systematic way. To carry this out, measures of the BP, activities, performance, resources, cost and results have to be defined, implemented, collected and analyzed on a regular basis [16]. Organizations in different domains such as software, manufacturing, marketing, banking, and finance share similar problems. These could include, for instance, overworked staff, thanks to poor estimating and planning, or excessive rework. There may also be a lack of consistent and stable processes, often with multiple ways to do similar things, or no sound basis for measurement and management. A foundation for organization-wide approaches and solutions may be lacking, and disappointing results from automation might exist. Other problems may arise, such as mixed results when applying approaches like Six Sigma or Business Process Reengineering (BPR), or improvements that are too localized and sub-optimal from a global business perspective [16].To meet these challenges, an improvement approach has to provide specific artifacts, along with a systematic way to support and guide the improvement effort in the organization [16]. It is not enough to provide measures and the means to analyze them, including tool support; it is essential to align the measures that are related to business strategy and goals for the entire organization with the ones that are specific for each BP. In a mature organization there is an objective quantitative basis for analyzing problems occurring in BP operation. Therefore, when changes are needed to address these problems the different options are understood as are the overall effects and consequences of choosing a particular option. Historical documented data is available and estimates and plans are based on this data, helping to achieve the expected results for aspects that include cost, schedule, performance and quality. A systematic improvement effort helps organizations move along the path from an immature status to a mature one [16], [17].

Business Process Intelligence (BPI) [18], [19] focuses on the collection and analysis of information from BPs to support decision making, and includes terms such as Business Activity Monitoring (BAM) and (Business) Process Mining (PM). BAM refers to the real-time monitoring of BP execution, focusing mainly on showing performance indicators. Several BI products exist, but they typically have a “data-centric focus”, and are “unaware of the processes the data refers to” [19]. This approach is based on traditional Online Analytical Processing (OLAP) engines that enable an analysis of the OLAP cubes. That analysis makes it possible to drill down and up among the defined dimensions (typical dimensions are regions, periods, and products) to see the consolidated information we wanted to obtain, and to “dice” the cube to view the multidimensional data from different angles [19]. Process mining, on the other hand, takes a BP execution event log as input and extracts information about its real execution. This makes three different types of process mining possible: discovery, to find the corresponding BP model, conformance, to check if the execution corresponds to the existing BP model, and enhancement, to extend or improve an existing BP model with its execution information [19]. This means that not only the control-flow of the BP can be mined from the event log; so also may the organizational perspective (to analyze the relation between resources and activities), the time (to show performance information, bottlenecks, resource utilization), or the decisions (to discover rules based on decision points in the BP) [19]. As stated in [19], “although in recent years the tool support for the approaches mentioned has matured considerably, most tools are not really “intelligent” and do not provide any process-mining capabilities. The focus is on querying and reporting, combined with simple visualization techniques showing dashboards and scorecards”. In contrast, the ProM¹ framework is one tool, among others, that does support process mining and which at the same time is available for free use [19]. ProM is based on several plug-ins that use data from the execution of BPs registered in event logs to discover BP models, to check conformance with existing BP models, and to extend existing BP models with execution information. However, few plug-ins provide information for operational support [20], [21]. There is, moreover, a general lack of an integrated view that provides the business area with the complete picture needed to analyze the execution of BPs in the organization, while simultaneously including information on the system execution.

According to [22], several improvement initiatives have become popular over recent decades. These include Just in Time (JIT), TQM (Total Quality Management), Lean manufacturing, BPR (Business Process Reengineering), and Six Sigma. Approaches such as TQM and Six Sigma are based on an analysis of business processes using different techniques, as well as on integrating improvement opportunities, based on an improvement cycle (PDCA, DMAIC); all these approaches are used by many organizations. Other life-cycle models such as IDEAL were developed specifically for software process improvement, based upon the Capability Maturity Model (CMMI), in widespread use by software organizations. In [23], [24], [25], [26], several aspects of the use of Six sigma, TQM, CMMI and IDEAL are presented and discussed, along with results from the application of these initiatives. One finding of such studies is that “like TQM, Six Sigma requires a strong incorporation of the corporate control system to enable companies to objectively measure and monitor their long-term development and monetary outcome of TQM using statistical techniques” [27]. The reference to the use of CMMI and ISO models is also interesting; these are difficult for small and medium organizations (SME) to apply because of the complexity of their recommendations and the large investment needed in terms of time and resources.

The outline of the problem we have presented above, along with the background we have sketched out, led us to identify some research challenges for our work. These refer to the lack of an integrated approach that would help organizations to improve their BPs based on their implementation with services. There is also a need for an execution measurement approach that is guided by a systematic way of integrating the improvement opportunities found in the BPs. Many existing approaches are too general, or address only one of the problems; for example, they may tackle only the definition of execution measures for collecting data about BP execution, or techniques for analyzing execution measures, or improvement processes to guide the improvement effort (c.f. Section 6). When trying to put all the views and elements together, improper integration might lead to unwanted results. We have been working with the Hospital General de Ciudad Real (HGCR) since 2007, in a project to introduce BPM in the organization. In this project, a selected set of BPs were modeled with BPMN, and we then assessed the quality of the BP models obtained, as well as the process followed to generate them [28]. This led to several interesting findings, for both the HGCR and the BPM community. Many of the problems detected in the HGCR inspired us to work on a different approach for the management and improvement of BPs. Moreover, the hospital wanted to select a BPMS that would enable it to initiate a program for implementing and executing the selected subset of BPs. It also wished to introduce and institutionalize an approach that would further guide the modeling, implementation, execution, evaluation and improvement of BPs in the organization.

Taking all this into consideration, we concluded that having an integrated approach to carry out improvement efforts in organizations in a systematic way could be very useful. Such an approach would need to bring together (1) the realization of BPs by means of services, as well as (2) the definition, implementation, collection and calculation of execution measures for both BPs and service execution. This would have to be presented in an integrated view, showing key execution measurement results. Moreover, the measurement approach would be embedded within a continuous improvement approach, defined as simply as possible, so as to be easily integrated into the organization’s culture and way of doing business. That would allow organizations to find improvement opportunities for both the BPs and the services implementing them, as well as to perform the integration of improvement of both BPs and services in a systematic way. They would thereby achieve the overall business goals of the organization, as well as the ones specific to each BP analyzed. With all the above needs in mind, we formulated the following research question:

• “How to set up a continuous improvement cycle for business processes implemented by services in organizations, based on BP execution measurement?”

Given the breadth of the question, we looked for more in-depth answers, entering into more specific aspects derived from the initial question:

• what measurement and improvement activities are needed in the BP lifecycle when BPs are realized by services ? (i.e., research question RQ1)

• what measures are appropriate for obtaining information about BP and service execution within the organization? (i.e., research question RQ2)

• what tools should be available to support the definitions in RQ1 and RQ2? (i.e., research question RQ3)

• are the proposals made in RQ1, RQ2 and RQ3 appropriate and useful for use in organizations ? (i.e., research question RQ4)

The measurement and improvement activities that we have included in BPCIP constitute the answers to RQ1 and are set out in detail in Section 2. The answer to the second question is the execution measurement model BPEMM, where we organize all the execution measures in a taxonomy, which is detailed in Section 3. The third question relates to the tool support we have defined; the answer to this is set out in Section 4. Finally, the answer to the fourth question corresponds to the validation of the proposal and we present this in Section 5.

To address the research objectives introduced previously, we applied a combination of different research methods. In the first place, we carried out a systematic literature review at the beginning of our research work, to collect existing proposals related to our research question (cf. [29]). Moreover, execution measures presented as part of RQ2 were distilled by means of an extensive literature review that was guided by experts in the subject, as well as by a systematic review about BP measures that had been carried out [30] in the context of a related research work. After that, we followed design science principles for the development of our proposal, as suggested by [31]. We worked closely together with business experts from the quality group of the HGCR, conducting several workshops for requirements elicitation to ensure that the proposals addressed real business needs. The proposal was then developed by means of several iterative cycles. In an initial cycle we defined the problem, reviewed existing literature and defined the main elements of the general MINERVA framework. For each of the three types of elements defined in the framework (conceptual, methodological, tool support) general cycles were carried out. This enabled us to answer the sub-questions and to define the different components of the framework, based on those answers. We created a set of artifacts to address the business requirements related to the research question of the proposal we are presenting here. These artifacts are: (i) a method description (i.e., BPCIP), (ii) an execution measurement model (i.e., BPEMM), and a (iii) tool chain. To validate the appropriateness of the artifacts we designed, a case study that was conducted at the end.

The added value of our integrated approach based on execution measures for the continuous improvement of BPs realized by services when compared with existing proposals can be stated as follows: firstly, the BPCIP improvement approach includes only a minimal set of key activities, roles and artifacts for guiding the BP lifecycle with explicit measurement and improvement activities. Our case study showed that this makes it easily understandable and easy to integrate into the organization’s culture and way of work. Other approaches, such as Six Sigma, define several roles and activities on the basis of a statistical analysis of BP behaviour; that can be difficult to integrate in organizations. Similar considerations apply for CMMI [26]. Secondly, the guidance provided by the BPCIP is supported by the execution measures integrated in the BPEMM model. Although several execution measures exist in literature, we selected the most important ones for measuring the execution of BPs and services and integrated these into our proposal. This created a centralized model which allows both business and IT to refer to the same definitions. This in turn serves as the basis for guiding the defined measurement and improvement activities. Although the BPMM, CMM, and CMMI models also provide measurement guidance to reach the fourth and fifth levels, with a statistical analysis focus, they do not give an explicit definition of measures and or measurement, or a roadmap for the execution of improvement activities [17]. The integration of BP execution with the execution of the services implementing those BPs, as defined in our proposal, provides a complete vision of the real operation in the organization. This can be presented both to the business and IT area enabling the discovery of improvement opportunities. Another added value of our proposal refers to the BPEMM ProM plug-in we have developed to support the processing of the BPEMM execution measures. It shows the measurement results in a graphical and easy-to-use way, which will be accessible to the ProM community and organizations wishing to evaluate it. In relation to this, twelve open-source process engines were assessed in how far the data needed for the calculation of the BPEMM measures is registered by the tools, allowing integration with ProM for the analysis.

The rest of the article is organized as follows: in Section 2 we present the BP Continuous Improvement Process (BPCIP) we defined, describing its disciplines, activities, roles and phases. In Section 3 we set forth the definition of the BP Execution Measurement Model (BPEMM) along with the specification of the execution measures we integrated, and in Section 4 there is an explanation of the tools provided to support the proposal. Section 5 presents a case study to demonstrate the applicability of the approach, and in Section 6 of related work other proposals are discussed. Finally, Section 7 discusses the conclusions we have drawn and outlines future work.