nova página do texto(beta)
Inglês (pdf)
Artigo em XML
Referências do artigo
Tradução automática
Enviar este artigo por email
Citado por SciELO 
Similares em
SciELO 
Permalink
I. Introduction
STRATEGIC alignment for improving the performance of an organization has received a renewed attention among academics and practitioners since the 1990s [1]. Some authors conclude that means for establishing and assessing such alignment are underdeveloped [2]. Consequently, achieving this alignment will be one of the most important management priorities in the upcoming years [3].
Several studies agree that organizational objectives are essential for leading the strategic alignment [2- 5]. According to Basili et al. [4], such objectives are not always explicitly nor clearly stated in the organizations. Thus, the verification of the achievement of those objectives is a difficult task.
Some other studies propose key performance indicators as vital factors for guaranteeing strategic alignment [2], [4], [5]. According to Cosenz [6], "what cannot be measured cannot be controlled." Consequently, strategies and assessment practices, such as the adoption of proper performance indicators, need to be implemented. In addition, according to Kronz [7], the holistic management can be achieved by collecting and analyzing performance in terms of key performance indicators (KPIs).
As a result, relationships among organizational objectives and key performance indicators are becoming more important. According to Basili et al. [5], an integrated vision of the company can be obtained by aligning and communicating all the goals, strategies, and measurement opportunities. So, all of the organizational elements can be directed to the same target. In most of the companies, performance measurement indicators have been developed; however, little effort to adequately establish links between objectives and performance indicators has been devoted [8].
Some approaches for exploring organizational objectives, KPIs, and objective-KPI relationships have been proposed. Some of them have metrics and models for measuring objective achievement. For example, Shamsaei et al. [9] propose an approach for modeling the context and measuring compliance levels based on certain rules. Pourshahid et al. [10] describe a method for decision making by including objectives, decision-making devices, and KPIs. Barone et al. [11] and Mate et al. [12] outline a method for modeling processes, objectives, indicators, and situations affecting the objectives. Strecker et al. [13] present a conceptual modeling proposal with the aim of satisfying essential requirements in the domain of organizational performance measurement. Frank et al. [14] outline a method for modeling indicator systems, based on a modeling language called SCORE-ML. Some proposals are mainly theoretical, and specifications for guiding any computational tractability are ignored, e.g., BSC (Balanced Scorecard) [15] and GQM (Goal Question Metrics)+ Strategies® [4], [17]. Thevenet [18] proposes the INSTAL method, which is aimed at aligning information systems to strategic business objectives in an organization. Doumi et al. [19] propose the modeling of strategic alignment among objectives and information systems by using indicators. Giannoulis et al. [2], [20] present the SMBSC (strategic maps and balanced scorecard) metamodel, which includes the concept of measure for supporting the evaluation of an objective performance. Finally Popova et al. [21] present a formal framework for modeling objectives, performance indicators and their relationships.
Unfortunately, the aforementioned contributions lack a wholly appropriate method for treating organizational objectives, KPIs, and objective-KPI relationships. They exhibit some drawbacks: i) some proposals are used for expressing objectives and KPIs by using informal natural language, causing ambiguity problems; ii) other proposals are used for expressing objectives and KPIs too formally inducing stakeholder understandability problems; and iii) in some cases the objective-KPI relationships are established by using personal criteria involving subjectivity problems. In the next Section we present a comparison of such contributions in order to evidence the aforesaid drawbacks.
In this paper we propose a novel method based on patterns for overcoming the previously identified drawbacks. Such a method allows us for deriving a set of candidate KPIs from organizational objectives. The method is based on pre-conceptual schemas [22], [23] for representing organizational objectives, key performance indicators, objective-KPI relationships, and the involved domain elements. Also, it uses a set of patterns for deriving candidate KPIs. As a result, a lab study is developed in order to demonstrate how the key performance indicators can be derived from organizational objectives by identifying and by applying certain patterns.
This paper is organized as follows: in Section II we show a review of the related work, and we present additional background information for facilitating a comparison among several approaches dealing with organizational objectives, KPIs, and objective-KPI relationships. In Section III we describe the proposal based on patterns for deriving key performance indicators from organizational objectives. In Section IV we provide a lab study for illustrating the main aspects of the proposal. Finally, in Section V we present some conclusions and future work.
II. Related Work
A. Objective-Oriented Proposals
Several approaches have been proposed for studying the strategic alignment in organizations. More specifically, such approaches have been focused on organizational objectives, KPIs, and objective-KPI relationships.
A first set of approaches based on GORE (goal-oriented requirements engineering) establishes the objectives as main components for contributing to the strategic alignment. Such proposals only emphasize on representing, defining, and analyzing the organizational objectives, but they disregard the inclusion of KPIs and objective-KPI relationships. Some of such proposals are presented by Kavakli [24], Engelsman et al. [25], Yu et al. [26], de la Vara et al. [27], Groner et al. [28], and Giannoulis et al. [29]. The latter formalize the strategy maps [30] by using a metamodel. The MAP methodology [31] allows for the modeling of the high-level goals and strategies of an enterprise.
B. KPI Oriented Proposals
Another set of proposals is focused on representing, defining, and analyzing the KPIs, but they disregard the inclusion of organizational objectives and objective-KPI relationships. Some of such proposals are presented by Caputo et al. [32] using SBVR (semantics of business vocabulary and business rules) for developing a KPI Vocabulary; Del-Rio-ortega et al. [33] proposing an ontology for the definition of process performance indicators; and Wetzstein et al. [34] using WSML (web service modeling language) for defining a KPI ontology.
C. Objective-KPI Oriented Proposals
Since the aforementioned set of proposals provides crucial elements for supporting the method we propose in this paper, we are mainly centered on those approaches aimed at including both organizational objectives and KPIs, as well as the objective-KPI relationships. Therefore, in this section, we review such proposals; besides, we analyze them with the previously established drawbacks in mind.
BSC (balanced scorecard) [14] includes causal linkages among objectives and performance measurements. In this context, objectives are measurable and the measures support the assessment for fulfilling such objectives.
GQM+ Strategies [4], [17] is an approach for linking organizational operational objectives and strategies from the top management level to the project level and back. Also, this approach is intended to align the business at all levels of the organization in a seamless way. Finally, GQM+ provides a mechanism for monitoring the success and failure of objectives and strategies by using measurement.
Giannoulis et al. [2] introduce the SMBSC meta-model, which allows for the integration of strategy maps and BSC by using a metamodel. Then, Giannoulis and Zdravkovic [35] present a case scenario based on i* and SMBSC.
Strecker etal. [13] introduce a conceptual model for representing an indicator system by rationalizing the process of creating, using, and maintaining indicators. Such a proposal includes the indicator-objective relation, and benefits from the reuse of MEMo (multi-perspective enterprise modeling) modeling concepts. Another proposal integrated with MEMo is outlined by Frank et al. [14]. The authors propose a modeling method for designing business indicator systems related to the organizational objectives. The modeling concepts are defined by using the performance modeling language called SCoRE-ML.
Doumi et al. [19] propose the modeling of strategic alignment among objectives and information systems by using indicators. Such indicators are associated with tasks and they are used for helping stakeholders to implement actions for achieving objectives. In addition, the indicators facilitate the verification of the accomplishment of the organizational objectives and the reorganization of the business processes.
The INSTAL method introduced by Thevenet [18] aims at aligning information systems with organizational objectives. Such a method involves metrics applied to strategic elements and measures applied to operational elements. However, in this method, performance indicators are lightly treated, and explicit objective-KPI relationships are not considered.
Two extensions of URN (user requirements notation) have been proposed. Shamsaei et al. [9] propose some rules for modeling the context and measuring the compliance level of their process. In addition, a regulation model, including policies, sub-policies, rules, and KPIs is provided. Pourshahid et al. [10] propose a method for making decisions, including objectives, decision-making devices, and KPIs into a single conceptual framework. In both URN extensions, a graphic tool is used for representing and analyzing KPIs. Basic notation of such a tool is illustrated in Figure 1. Unfortunately, KPIs are subjectively related to the objectives, and the specification of KPIs is carried out in an uncontrolled way by using natural language.
Barone et al. [11] and Maté et al. [12] use BIM (business intelligence model) for modeling the business strategy. This framework comprises the modeling of objectives, indicators, and potential situations affecting objectives. Also, some techniques and algorithms are provided for deriving values and composing indicators. Such proposals are supported by a visual editor prototype for drawing business schemas and reasoning about them. An example of this framework is illustrated in Figure 2.
Popova et al. [21] present a framework for formally modeling predicate-logic-based performance indicators and their relationships by using a performance-oriented view. LEADSTO language and the LEADSTO property editor tool are used in the modeling process. LEADSTO language is a sublanguage of TTL (temporal trace language) which enables direct temporal-or causal-dependency modeling among state properties. Besides, the relationships are formally defined by using axioms expressed in TTL. The components of the framework and the considered modeling views- process, performance, organization, and agent-can be directly related to the components of the GERAM (the generalized enterprise reference architecture and methodology). This approach is illustrated in Figure 3.
D. Comparison of Proposals
Table 1 shows a comparison of the aforementioned proposals. We use the following notation: "✓" means the proposal successfully addresses the issue; "p" indicates the proposal addresses it partially; and " x" indicates the proposal does not contemplate the issue.
We see some proposals related to the objective-KPI relationships. However, none of the analyzed approaches provides an assisted method that would systematically guide an appropriate derivation of KPIs. Moreover, some of such proposals specify objectives and KPIs either informally in natural language or simply as labels within a specific diagram, leading to ambiguity problems. other proposals express objectives and KPIs by using formal specifications that induce stakeholder understandability problems. Lastly, some proposals establish objective-KPI relationships by using only the personal criterion involving subjectivity problems. For these reasons, more research on this field is needed.
III. Our Proposal
The main weakness of the current research in this area is the lack of a systematic approach for deriving KPIs from the concepts involved in specific organizational objective domain. In this paper we propose a systematic method for deriving KPIs from organizational objectives. Thus, we can correctly map elements representing the organizational objective domain to elements representing the performance indicator domain. These indicators are useful for assessing the fulfillment of the organizational expectations. In addition, an appropriate traceability among the aforementioned elements can be established.
A method for systematically linking organizational objectives to KPIs should include the following features:
The method should provide a clear understanding of the domain where the organizational objectives are defined.
The method should support the analyst by providing techniques to perform the derivation process of KPIs in a systematic and precise way.
Accordingly, our proposal includes the following phases as a way to deal with the previously mentioned problems:
Defining a modeling process for describing the organizational objective and their domain.
Developing a methodological guideline for deriving KPIs from organizational objectives.
A. Phases
1) Modeling phase:
In this step, we represent the actual domain of the enterprise related to such organizational objective. We use the pre-conceptual schemas [22] for modeling the organizational objectives and their domain. Pre-conceptual schemas have several advantages: unambiguous syntax, integration of concepts, dynamic elements, and proximity to the stakeholder language. In addition, untrained stakeholders can understand pre-conceptual schemas [23].
Basic Concepts and Notation
Nodes
Pre-conceptual Schemas include five kinds of nodes connected to two kinds of arcs for graphically representing a model [22, 23].
Every concept has an incident connection-type arc starting or ending in a relationship. The concept node can be of two kinds:
Class-Concept is a concept that contains attributes and can be instantiable.
Attribute-Concept is a leaf concept within a Pre-conceptual Schema, i.e., an attribute of a Class-Concept.
A dynamic relationship represents actions and has exactly one incoming and one outgoing connection arcs. Such a relationship can be connected to concepts by using connections arcs. In addition, it can be connected to other dynamic relationships by using implication arcs.
A structural relationship can be either is-a or part/whole type. It is incident to concepts and has exactly one incoming and one or more outgoing arcs of connection type.
An achievement relationship can be connected to structural and dynamic relationships and concepts. In addition, achievement relationships can be connected among each other by using implication arcs.
Arcs
A connection arc connects a concept to a relationship or vice versa.
An implication arc connects a dynamic relationship to a dynamic relationship. Implication arc can be also used for connecting achievement relationships among each other.
The basic notation of the pre-conceptual schemas is illustrated in Figure 4.
2) Derivation Phase
In this phase, a set of candidate KPIs is derived from organizational objectives and their previously modeled domain. The derivation process comprises a set of patterns, which guides the analysis of the several possibilities for identifying key performance indicators. According to Martinez [36], a pattern reflects something to be used in a number of situations and, thus, has some generality. She establishes three features of the patterns: (1) the description of a pattern contains a context, which explains the intent of the pattern and suggests how it should be used; (2) patterns express solutions in a variety of ways, depending on the details of a circumstance; and (3) pattern descriptions can express architectural considerations, regardless of specific languages and design methodologies.
Consequently, we define a set of patterns to be systematically processed and to guarantee recurrent solutions in every step of the process. Such patterns specify candidate KPIs from an organizational objective model.
B. Description of the Pattern
We use a template-based on the work developed by Martinez [36] -for describing the patterns defined in this paper. Hence, the basic elements we use are the following:
Name: A sentence summarizing the pattern.
Context: A situation characterizing a problem. It describes situations in which the problem occurs.
Problem: The recurring problem arising in the context.
Structure: A detailed specification of the structural aspects of the pattern.
Solution: The strategy for solving the recurring problem.
Examples: A model intended to illustrate the pattern.
C. Patterns in the organizational objective model
The patterns proposed in this paper are focused on identifying several structures belonging to an organizational objective model that can be useful for deriving some KPIs related to such organizational objective. Then, a specific pattern will be used depending on the type of structure identified.
Proposed Patterns
The KPIs derivation process is carried out in a systematic way by identifying those relevant elements in the model suggested by the stakeholder and by analyzing such elements. In order to complete such task, we provide a set of patterns based on pre-conceptual schemas for systematically guiding the derivation process. In this section we present some examples of such patterns:
The leaf-attribute pattern: To be applied when an attribute concept -which is related to an achievement relationship- can be used for generating a set of candidate KPIs.
The dynamic-relationship pattern. To be applied when a dynamic relationship -which is related to an achievement relationship -can be used for generating a set of candidate KPIs.
Applying Patterns
The derivation phase can be initiated after modeling the domain of the enterprise related to an organizational objective. Thus, the proposed patterns can be used when some part of the pre-conceptual schema matches the pattern. In order to complete this task, a specific method for applying the proposed patterns is presented. The method comprises the following steps:
Step 1. Analysis of modeled elements.
Each of the modeled elements should be analyzed in order to determine if such an element can be considered as a candidate element. A candidate element should meet the features described in any of the proposed patterns.
Step 2. Identification of the appropriate pattern.
Once a candidate element-and the portion of the pre-conceptual schema-have been identified, the suitable pattern-type should be determined.
Step 2.1. When an attribute concept is encountered and it is linked to an achievement relationship, the leaf-attribute pattern (pattern 1) can be applied.
Step 2.2. When a dynamic relationship is encountered and it is linked to an achievement relationship, the dynamic-relationship pattern (pattern 2) can be applied.
Step 3. Derivation of KPIs.
The procedure described in the pattern solution should be followed up in order to carry out the derivation of KPIs from the modeled organizational objective.
Catalog of Patterns
In this Section, each pattern is explained in detail by using the structure defined by Martinez [36].
1) The leaf-attribute pattern.
Context
This pattern is applied when an attribute-concept is candidate for deriving a set of KPIs. An example of this structure is shown in Figure 5.
Problem
The problem relies on generating a set of KPIs from a candidate attribute-concept, which is determined by using the pattern 1. Also, the type of achievement relationship- improvement, maintenance, or accomplishment-linked to the concept should be identified. So, the appropriate KPI structure can be defined.
Structure
The elements involved in this pattern are as follows:
Attribute-concept. The concept used for deriving the set of KPIs.
Achievement relationship. One of the relations linked to the concept. Such a relationship can be improvement-, maintenance-, or accomplishment-type.
Structural relationship. One of the relations linked to the concept. The type of this relationship is part/whole.
Solution
In this paper, the achievement relationships involved in the case study are treated as improvement -type. More detailed information about achievement-type relationships is provided by Lezcano [16]. The process for applying this pattern is as follows.
A KPI is obtained by writing a function name from the list: amount of, average, maximum, minimum, number of. Then, the label of the concept is added. Also, in case of average function, the KPI is formed as follows: writing "average" followed by the label of the concept and the word per. Then, the label of the source concept of the structural relationship is added. Similarly, when the function name is amount of, maximum or minimum, the KPI is formed by adding the word per and the label of the source concept in the structural relationship. Lastly, in case of number of function, the KPI is formed by writing number of followed by the label of the source concept, in the structural relationship-by using its plural form.
Example
The application of the leaf-attribute pattern (pattern 1) is illustrated in Figure 6.
A set of candidate KPIs is derived by following the process described by our proposal corresponding to pattern 1. As a result, the candidate KPIs are:
amount of BONUSES
maximum BONUS
minimum BONUS
amount of BONUSESper EMPLOYEE
maximum BONUSES per EMPLOYEE
minimum BONUSES per EMPLOYEE
average BONUSES
average BONUSES per EMPLOYEE
number of EMPLOYEES
2) The dynamic-relationship pattern.
Context
This pattern is applied when a dynamic relationship is a candidate for deriving a set of KPIs. An example of this structure is shown in Figure 7.
Problem
The problem relies on generating a set of KPIs from a candidate dynamic relationship, which is determined by using the pattern 2. Also, the type of achievement relationship- improvement, maintenance or accomplishment-linked to the concept should be identified. So, the appropriate KPI structure can be defined.
Structure
The elements involved in this pattern are as follows:
Concept. The target concept in the dynamic relationship.
Achievement relationship. The relation linked to the dynamic relationship. This relationship can be improvement-, maintenance-, or accomplishment-type.
Dynamic relationship. The relationship containing the action verb.
Solution
In this paper, the achievement relationships involved in the case study are treated as improvement -type. More detailed information about achievement-type relationships is provided by Lezcano [16]. The process for applying this pattern is as follows.
A KPI is obtained by writing a function name from the list: amount of and percentage of. Then, the action verb is added by using its past participle. Lastly, the label of the target concept, in the dynamic relationship, is added.
Example
The application of the illustrated in Figure 8.
Figure 8 illustrates the occurrence of pattern 2, so a set of candidate KPIs is derived by following the process described in the solution Section corresponding to pattern 2. As a result, the candidate KPIs are:
IV. Case Study
In this Section, we present a case study related to the organizational objective IMPROVING SALES. In this example, the domain related to such objective is modeled. So, IMPROVING SALE is represented by the achievement relationship IMPROVING and the class concept SALE. This organizational objective is in turn decomposed into two sub-objectives: INCREASING REVENUES and INCREASING PRODUCT SELLING.
After the relevant concepts and their relationships have been modeled, we can analyze the application of the proposed patterns. As a result, pattern 1 and pattern 2 are identified and a set of candidate KPIs are derived. This example is illustrated in Figure 9.
As a result, the set of candidate KPIs related to the organizational objective IMPROVING SALES are summarized in Table 2, including the pattern applied for selecting the candidate KPI.
V. Conclusions and Future Work
Strategic alignment involves holistic control of the organization. Such control should be assisted by appropriate methods that allow us to reason about the fulfillment of the organizational objectives. KPIs have demonstrated to be effective resources for assessing such fulfillment. By analyzing some contributions, we evidenced they exhibit some drawbacks for linking organizational objectives to KPIs. In this paper, we proposed a method based on patterns for deriving KPIs from organizational objectives as an appropriate solution for overcoming the identified drawbacks.
Then, we demonstrated the application of the proposed method by using a lab study.
As a result, we experienced the advantages of using the pre-conceptual schemas as a mean for representing a particular domain. Such advantages rely on the proximity to the stakeholder language as well as the capacity for overcoming ambiguity problems. Moreover, the proposed patterns allowed for the derivation of candidate KPIs in a systematic way overcoming subjectivity problems. Thus, the use of such patterns demonstrated relevant advantages concerning the reuse, communication, and documentation of solutions to recurrent problems related to the generation of KPIs from goals.
Lastly, by applying the proposed method in this paper, the stakeholder is provided with a set of candidate KPIs intended to assess the fulfillment of the organizational objectives. Such KPIs are systematically obtained from a particular domain modeling, and they correspond with the own and specific needs an organization experiments. Since the candidate KPIs are systematically obtained, they can make aware the stakeholder about still undiscovered candidate KPIs.
Some lines of future work can be proposed: (i) the application of the proposed method on other different domains; (ii) the comparison of the results against the established indicators by specific frameworks; (iii) the full automation of the process for automatically deriving KPIs; (iv) the definition of new patterns for deriving candidate KPIs from goals; and (v) the automatic generation of source code from the candidate KPIs, to be included in the source code of the application to-be-developed.
