Software Development Estimation Techniques in Industrial Contexts: An Exploratory Multiple Case-Study

Abstract

Software Effort Estimation is one of the most challenging aspects in the software development life cycle. Recent empirical studies in the area of software development estimation indicate the presence of two models for effort estimation: (i) Formal, and (ii) Expert Based (Informal). The IT sector in Palestine is one of the most promising and constantly growing sectors. Nonetheless, studies addressing effort estimation approaches and techniques within the Palestinian IT sector are still highly missing. Therefore, we were motivated to conduct a qualitative study to increase our understanding about how industrial teams approach software effort estimation and to explore the challenges they are facing. Our investigation started with a survey that targeted software professionals, and then we conducted multiple-case study approach involving four different software development companies in Palestine. Results show that: (i) around 25% of cost overrun in software projects is due to inaccurate estimations; (ii) expert based estimation models are the mostly applied models especially within agile environments; (iii) a potential advantage can be achieved when formalizing the process of expert based models by having guidelines and checklists; (iv) accuracy of effort estimation is largely affected by team experience, domain knowledge, and requirements clarity; (v) companies working with outsourcing model do have better effort estimation accuracy than companies working in local market. Based on our findings, we highlight areas that require further investigation.