Requirements Eng (1996) 1:75--87 9 1996 Springer-Verlag London Limited Reqt/irements Engineering Requirements Capture and Analysis : A Survey of Current Practice Prodromos D. Chatzoglou and Linda A. Macaulay Department of Computation,UMIST, Manchester, UK This paper presents the findings of a detailed survey of 107 projects in which the iterative nature of requirements analysis was explored in economic terms. The survey was conducted from the point of view of the project manager. The results indicate that half of the projects take three or more iterations to complete the requirements, that the use of methodologies and project characteristics affect the number of iterations, and that in half of the projects the number of iterations planned was different from the number actually carried out. The paper concludes by attempting to explain the relationship between the economics of the requirements process and the number of iterations through a spiral model of requirements capture and analysis. Keywords: Requirements; Survey; Current practice; Project size; Systems development economics 1. Introduction Lubars et al. [1] undertook a number of in-depth interviews with practitioners in order to find out how organisations currently define, interpret, analyse and use the requirements for their software systems and products. One of the 30 or more questions asked was 'What connection is there between requirements analy- sis and planning?' The results indicated that ' Manage- ment somehow estimates the cost of individual features. Nobody we spoke to knew how this was done, although many suggested that an informal, experiential approach was adopted and not formal cost estimation models.' Correspondance and offprint requests to: P. D. Chatzoglou, Depart- ment of Computation, UMIST,PO Box 88, Manchester M60 tQD, UK. Email: EChatzoglou@mac.co.umist.ac.uk;lindam@sna.eo. umist.ac.uk This paper presents an attempt to find out more about the 'informal, experiential' approach by asking practitioners specific questions about their experience of planning the requirements stage of a project. The investigation took the form of a two-stage mail survey. In both questionnaires respondents were asked to give their answers in relation to the last project they were involved in. In the first questionnaire questions were asked about the whole of the system development process. One of the outcomes from this questionnaire was that the requirements capture and analysis (RCA) process was reported by most respondents as being an iterative process. This finding confirmed reports in the literature about the iterative nature of requirements [2-5]. The notion of iteration is very well presented by Boehm's Spiral model for systems development process [6,7]. This model follows an approach to system development which is different from the other development models (for example, the traditional waterfall model [8]; rapid prototyping [9]). It shifts the management emphasis from development of products to assessment of risk and explicitly calls for evaluation as to whether a project should be terminated [10]. It reduces all development stages to a cycle of four activities and thus development becomes a repeated application of the same normalised activities to create preliminary ver- sions of the final product [11]. Each cycle is completed by a review involving the people or organisations concerned with the product. Thus the second ques- tionnaire focused on exploring the iterative nature of requirements. This study briefly : | explains the iterative nature of the RCA process in economic terms; 9highlights the differences in the number of iterations of the RCA process in projects with different characteristics;