The concept behind the research presented in this thesis is that a generic Software Process Improvement model, based on Quality Function Deployment, can be used to help a company improve their software process in an affordable manner while also keeping business goals in focus. The ultimate aim was to develop a model which would be useful in practice for small software development companies. In the course of this project, the researcher developed, verified and validated such a model.

During the 1990s there has been a movement in the software development community towards the improvement of the software process as a means to improve the quality of software. Small software development companies are important to the Irish economy, and therefore, should keep up-to-date with global developments. Existing models such as the Capability Maturity Model, BootStrap and ISO9000, and the proposed international model SPICE, have been criticised, particularly in relation to the small company, as being cumbersome, expensive and failing to present a comprehensive improvement strategy. As a result of her previous research work, the researcher proposes a Software Process Improvement / Quality Function Deployment model based on this technique.

The author has developed one stage of the proposed model, the Software Process Matrix, which is generic for all companies using it. This matrix is mainly based on a previously existing self-assessment model. To complete and verify the matrix she uses hypothesis tests taking expert opinion into account, carried out over three iterations of the matrix. The Software Process Matrix has been validated by the researcher using action research with control groups in a longitudinal study.

The matrix has been implemented in two small software development companies in the mid-west of Ireland while also investigating the software processes in two control companies based in the same region. As a result of this research, a number of modifications to the original Software Process Improvement / Quality Function Deployment model are presented.