Full citation: Kirk, D., & MacDonell, S. (2009) A systems approach to software process improvement in small organisations, in Proceedings of the 16th European Software Process Improvement and Innovation (EuroSPI) Conference. Alcala, Spain, Delta/Publizon, pp.2.21-30. A Systems Approach to Software Process Improvement in Small Organisations Diana Kirk and Stephen G. MacDonell SERL, Auckland University of Technology Private Bag 92006, Auckland 1142, New Zealand {dkirk, stephen.macdonell}@aut.ac.nz Abstract There is, at the present time, no model to effectively support context-aware process change in small software organisations. The assessment reference models, for example, SPICE and CMMI, provide a tool for identifying gaps with best practice, but do not take into account group culture and environment, and do not help with prioritisation. These approaches thus do not support the many small software organisations that need to make effective changes that are linked to business objectives in short time periods. In this paper, we propose a model to support such change. We base the model on an analogy of 'software system as human' and suggest that we can apply the idea of human health to help identify business objectives and improvement steps appropriate for these objectives. We describe a 'proof-of-concept' case study in which the model is retrospectively applied to a process improvement effort with a local software group. Keywords: Software process improvement, process modelling, systems approach 1. INTRODUCTION Existing models for software process improvement (SPI), for example, CMMI (Chrissis et al. 2007), ISO/IEC 15504 (SPICE) (ISO 2006) and ISO/IEC 12207 (ISO 1997), have been criticised by researchers and practitioners as being limited to large, traditional organisations and failing to provide the necessary guidance for small software groups (CaterSteel 2001, Grunbacher 1997, Huack et al. 2008, McCaffery et al. 2008). As software groups comprising fewer than 25 persons represent a majority in Europe, Ireland, Canada, Brazil and elsewhere (Laporte et al. 2008), this is seen as a major issue. Some characteristics of smaller organisations that may affect the success of SPI adoption include more informal management and planning, greater need for flexibility, a human-centric culture (Laporte et al. 2008), 'flatter' structure with imprecisely-defined responsibilities, a lack of exposure to standards and limited funds (Grunbacher 1997). Such characteristics imply an environment where dependence upon individuals is high. It has also been observed that existing models state which processes should be in place, but provide no guidance as to which to implement first (Chen et al. 2008). This means that the need of small organisations to prioritise according to business objectives (Aaen 2003, Laporte et al. 2008) is not supported. We have earlier suggested that a model to support activity selection during software projects must take into account the business-related objectives for the software project and have proposed that a more holistic and flexible approach to process selection involves a change in focus from 'defining activities' to 'selecting activities to meet objectives' (Kirk 2007). In this approach, focus is on the whole system i.e. is not limited to considerations of cost and quality but rather includes consideration of human-related factors. For example, the owner of a small software organisation may be extremely interested in retaining and increasing the knowledge of developers and so may consider 'developer knowledge' to be of importance. This understanding may inform his choice of process and he may, for example, choose informal reviews over unit testing as a means of meeting objectives.