In seeking to improve software, companies are finding out how much is involved in measuring it. They are also learning that the more integral software measurement is to the company’s underlying business strategy, the more likely it is to succeed. or successful product and process definition, implementation, and evolution, developers increasingly look to software measurement techniques. 1-3 F We typically regard measurement as essential for project—but less so for process— management. Yet without adequate quantitative support, we can’t readily comprehend, much less master, the complexities of software structure, testability, quality, reliability, or safety. Only with measurement-based frameworks can we approach anything like an engineering ethos in activities vital to effective software production that are, all too often, ill-defined and poorly managed. Measurement programs can facilitate a true engineering approach to software development, and specifically software product and process improvement, giving developers a competitive advantage over those who pursue more traditional approaches. Measurement programs can also ensure that welldefined measurement frameworks emerge, along with rigorously defined measures and meaningful data. (We prefer the term measures rather than metrics, as metric connotes a generic distance measure in the mathematical and physical sciences.)

This article illustrates the automation of the most expensive phases of a Goal Question Metrics (GQM) measurement process. In particular, I describe the use of a specific tool for the definition, collection, analysis, and feedback of measures, and the techniques to interface such a tool with a configuration management system and metrics tools. (See the "Research and Practice" sidebar for more background information.) A measurement program at Pirelli Cavi, the Italian division of Pirelli Cables and Systems, has employed these techniques to automate the extraction of reliable and process-conformant data for the development of a Telecommunications Management Network (TMN) product. Although I focus on the GQM methodology, this work also provides general indications that apply in other contexts. Why GQM? The GQM 1 is a systematic technique for developing measurement programs for software processes and products. The GQM process is based on the idea that mea

Success factors for measurement programs as identified in the literature typically focus on the `internals' of the measurement program: incremental implementation, support from management, a well-planned metrics framework, and so on. However, for a measurement program to be successful within its larger organizational context, it has to generate value for the organization. This implies that attention should also be given to the proper mapping of some identifiable organizational problem onto the measurement program, as well as the translation back of measurement results to organizational actions. In this paper, we present a generic process model for measurement-based improvement, which does cover the latter issues as well. We describe a number of common uses for measurement programs in software organizations, from which we derive additional `external' success factors. In addition, we propose a number of activities that organizations can use to implement value-generating measurement programs.

Many frameworks for implementing software measurement exist, ranging from collections of success factors to maturity growth models. One may ask to what extent these guidelines increase the chance of a successful measurement program. To aid in answering this question, we introduce a generic process model for measurement-based improvement. We use this model as a reference model to compare a number of existing software measurement implementation frameworks. From these assessments we conclude that the guidelines given by these frameworks provide a considerable amount of support for the basic activities needed to implement measurement programs. However, we also observe that the guidelines hardly provide any guidance to guarantee successful usage of the measurement program. 1 Introduction Briand, Differding, and Rombach [3] state that `Despite significant progress in the last 15 years, implementing a successful measurement program for software development is still a challenging undertaking...

Measurement is a necessary prerequisite for software process improvement. However, few guidelines exist for systematic planning of measurement programs within software projects. We advocate beginning with goal-oriented measurement as expressed in the GQM paradigm. When applying GQM, meaningful refinement of measurement goals requires a deep understanding of the organization's software development processes. Software process modeling can act as a means to gain this understanding. We provide guidelines that state how software processes should be modeled in order to support the introduction of measurement in a software project. The process models are used to derive non-intrusive data collection procedures that satisfy the needs of multiple measurement goals and thereby minimize the data collection overhead. Finally, we present the lessons learned from jointly applying software process modeling and measurement technologies in two industrial software projects. 1. Introduction Continuous sof...

Effective management of requirements is of vital importance for any software development effort. This activity spans from software procurement to project management, to the technical development activities.

Success factors for measurement programs as identified in the literature typically focus on the `internals' of the measurement program: incremental implementation, support from management, a well-planned metrics framework, and so on. However, for a measurement program to be successful within its larger organizational context, it has to generate value for the organization. This implies that attention should also be given to the proper mapping of some identifiable organizational problem onto the measurement program, and the translation back of measurement results to organizational actions.

Software Process Improvement (SPI) initiatives induce organizational change, by introducing new tools, techniques and work practices. Organizations have to address acceptance issues such as resistance to change, compatibility and fear of adverse consequences. Social psychology literature includes the Technology Acceptance Model (TAM) and the Theory of Planned Behavior (TPB), which study such adoption issues and predict intention to use and actual usage of workplace technology. Some constructs of these models could be applied to software organizations to make it easier for them to counter the initial resistance and to assimilate process improvement into the work culture. To increase applicability of these models to the SPI context, some additional constructs are proposed, by taking into account organizational culture, the impact of changes caused by SPI and the unique characteristics of software developers.

According to the terms of Cooperative Agreement #NCCW-0040,

Software measurement is directly related to software production processes. Measurement programs should be flexible and easily adaptable to the features of each organization. Moreover, in the context of small software businesses, measurement should be feasible in spite of the lack of resources. In order to produce measurement systems that can be developed and that face the needs of the dynamic reality of small businesses, we propose the development of a meta-system that is capable of instantiating measurement systems that are suited to each organization. Our measurement systems are based upon the selection and customization of the data schemes and upon the definition and customization of reports. The meta-system that we will develop will be tested at INFOGENE, which is a software business incubator that is supported by the Pontifcia Universidade Catlica do Rio de Janeiro (PUC-RIO). The incubator offers plenty of useful situations for testing a measurement system's flexibilit...