Using Quality Function Deployment in Software Requirements Specification

USING QUALITY FUNCTION DEPLOYMENT IN SOFTWARE REQUIREMENTS SPECIFICATION

Abstract. We present in this paper an approach to requirements specification based on quality function deployment (QFD). We discuss how techniques like QFD with accompanying tool-support can contribute to the development of high quality requirements models that are the basis for information systems that are perceived as valuable by the stakeholders of the systems. Based on the practical application of this technique on a number of information systems projects, including both traditional development projects and the implementation of packaged solutions, it is our experience that the quality of requirements specifications can be improved by using this technique, in particular when used together with groupware tools supporting the overall process. Especially, what we term semantic, pragmatic, and social quality are often affected positively. On the other hand, the QFD-technique is not sufficient alone, but needs to be combined with other RE-techniques such as prototyping, observations, and more traditional conceptual modeling to be able to cover and analyze all types of systems requirements.

1 Introduction

Information systems development is a group activity, needing the input of many people to result in innovative systems that truly support the needs of the stakeholders. Software requirements can be classified into three categories: Normal, exciting and expected .

1. Normal Requirements are what the stakeholders communicate during traditional facilitated sessions or in interviews. They cover the base functionality of the application. These requirement contribute proportionally to customer satisfaction and expectations.

2. Exciting Requirements are aspects which the users do not expect. Often exciting requirements involve innovation of the business process or new ways of handling functionality. Stakeholder satisfaction with the application can be dramatically improved through the implementation of a few exciting requirements. Failure to implement these requirements do not adversely affect overall satisfaction. Exciting requirements can be identified through a number of techniques. These include: observation, brainstorming, affinity analysis, relations diagramming and causal loop diagramming.

3. Expected Requirement are those aspects that the users assume the developers already know. Missing an expected requirement represents the greatest risk to user satisfaction. There is little benefit for implementing these