Measuring Software Costs: A New Perspective on a Recurring Problem

Regular outsourcing of software development signifies a need for a common understanding between procurement specialists, project managers and software engineers of individual system requirements costs, to facilitate accord on price. Dependable cost estimates are usually required by customers long before detailed analysis and design activities would normally produce this information. A number of estimation techniques have evolved to produce cost information at an early stage in the software life-cycle, however estimation continues to contribute to budget blowouts. Contemporary techniques for costing specifications described as use cases are increasingly challenged as the size and complexity of the system expands. In addition object-oriented representations of requirements fail to directly map into structures used by project managers, leading to ongoing comparisons of value that are subjective and often unrepresentative of final project expenditure. Flaws in the current application of cost estimation methodologies have been identified and a possible solution is proposed. 1. Current Interests Curtin University of Technology, ADI Limited, Defence Materiel Organisation (DMO), Total Metrics, and other partners have applied to the Australian Research Council (ARC), as part of the November 2003 Linkage Project round, for funding for a project to take a multidisciplinary approach to the costing of software development. The results of this application are not yet known, but the partners are keen to progress this research in advance of the ARC decision, for the proposed project: “Optimising Value for Money (VfM) in software development through standardised cost metrics in the procurement, development and project management life-cycles”. This project aims to develop the principles by which the ongoing costs can be estimated from specifications based on use cases and can then be precisely mapped into a Work Breakdown Structure (WBS) used by the project management to control a software engineering life-cycle. The team believes that this research will allow organisations to set budgets with greater confidence and more accurately determine progress during software contracts. It has significant implications for procurement specialists purchasing software development. Problems encountered during a contract by customer contract managers and supplier project managers have often been introduced into the contract through suspect decisions on budget, scope, schedule and choice of supplier made at early stages of the procurement process. The validity of many VfM decisions would come under serious scrutiny if an identical analysis were to be repeated subsequent to contract closure. A reduction in the frequency of invalid cost estimates and inappropriate budgets would contribute to more successful software projects and improve the VfM proposition for bespoke software development in comparison to the purchase of packaged software. Any ‘buy’ decision over a ‘build’ decision should be based upon the business merits rather than any associated contract management problems.