Factors influencing user story estimations: an industrial interview and a conceptual model

www.ceejme.eu ww.wsb.pl/wroclaw/ceejme ISSN electronic version 2353 - 9119 Central and Eastern European Journal of Management and Economics Vol. 4, No.4, 261-280, Dec. 2016 Factors influencing user story estimations: an industrial interview and a conceptual model Marek MAJCHRZAK, Lech MADEYSKI Wroclaw University of Science and Technology, Poland Abstract: A proper estimation of time in user stories is a crucial task for both the IT team as well as for the customer, especially in agile projects. Although agile practices offer a lot of flexibility and promote a culture of continuous change, there are always clearly de need timeboxed periods where an IT company has to commit to delivering working soft-ware. Estimating time of user story implementation provides clarity and the opportunity to control the project by the management, yet at the same time, it can increase pressure on software developers. Thus, incorrectly estimated user stories may lead to quality problems including system malfunction, technical debt, and general user experience issues. The paper describes user story characteristics, reasons of user story estimation inaccuracy as well as a model of their potential impact on post-release defects in large IT software ventures, all derived from the conducted interview with practitioners in Capgemini software development company. Keywords: estimation, user story, agile, defect prediction, Scrum JEL: C92, G31, D81, D83, C53 1. Introduction In the 1995 Standish Group conducted research among IT companies (The Standish Group, 1995), they found that only 16% of the projects were completed on-time and on-budget. The last CHAOS report (The Standish Group 2015) showed the increasing rate, near to 40%, of agile projects successfully resolved. Despite the fact that the Scrum framework, at the team level, dominates (VersionOne 2015) among other agile methods and practices, one can observe growing importance of Development & Operations (DevOps) culture. The main reason for applying DevOps in the development process is the ability to release very often (i.e., daily or even several Correspondence Address: Lech Madeyski, Wroclaw University of Science and Technology, Wybrzeże Wyspiańskiego 27, 50-370 Wrocław, Poland. Email: ; Marek Majchrzak, email: © 2016 WSB University in Wrocław FACTORS INFLUENCING USER STORY ESTIMATIONS … times per day) (Puppet Labs and IT Revolutionary Press 2015) fully tested working software. To achieve that, especially in large IT ventures, one needs to introduce sophisticated testing as well as predictive analytics techniques (Buenen, Walgude 2016). Predictive analytics can be used to predict defect-prone software modules (e.g., classes). The aim of this paper is lay down the foundation for the novel kind of defect predictors, based on inaccurate estimations, that could be used to enrich software defect prediction models. Understanding the reasons of inaccurate estimation and its potential impact on software quality would also be beneficial for the practitioners. Estimating the cost of a software development project, since the very beginning of software engineering (Caminer 1958, Brooks, Frederick 1987), is one of the most crucial tasks for IT companies. Unfortunately, it is still one of the project management aspects that has to be improved. The number of studies has been conducted to understand inaccurate estimation phenomena in agile projects and possible effects on project schedule and budget. For example, Lang et al. (2011) looked at how classical problems that affect cost estimation in traditional software development projects are managed within the agile paradigm. They analysed several agile projects, and they found that estimation inaccuracy was a less frequent occurrence for these companies than in traditionally software development projects. They analysed several factors which may lead to inaccurate estimates. They found that potentially a severe threat to accurate estimates are user communication difficulties. Additionally, they notice that new people and new technologies in the team are a principal threat to produce accurate estimates. Many studies focus on project cost estimation inaccuracy. Liskin et al. (2014) analysed the single User Story (US) aspect and in particular the granularity (size). They found that the granularity of the US was seen as one of the main reasons for inaccurate estimation. Børte et al. (2012) analysed the role of social interaction in software e ort estimation. They proposed to consider the socio-cultural perspective in e ort estimation. Especially the ways in which software professionals reach a decision. They found, that during the planning meeting, US mediates between the historical practices and the use of generic and specific knowledge. Knowledge of time spent on a task allows scheduling work which can be carried out in the nearest future. It is important for project managers who can control human resources in an appropriate manner to achieve the best possible results. On the other hand, historical data from past sprints may be used by an agile team as a hint during the planning meeting to avoid incorrect user 262 FACTORS INFLUENCING USER STORY ESTIMATIONS … story (US) estimates and to mitigate the risks of sprint failure. Working under time pressure in every software project may lead to different quality problems. However, there are no studies that have analysed the impact of incorrectly estimated US on post-release defects in software. Hence, the aim of the paper is to fill this gap and to investigate the reasons of inaccurate estimate and impact on post-release defects. This paper is organised as follows: a general description of the US life-cycle and characteristics in the Section 2. Section 3 describes the conducted survey, participants and presents the results, while the discussion is in Section 4. Conclusions and future work are presented in Section 5. 2. User story characteristics In software development and especially in agile software development, a US is a high-level definition of a requirement (Ambler 2002) in the business language of the end user. User stories are a part of an agile approach which helps to shift the focus from writing about requirements to talking about them (Cohn 2004). From the business perspective, each US has a title, a description, an author, a group of stakeholders and its priority in the product backlog (PB). From the perspective of a development team, the characteristics include the size (estimate), tasks, acceptance criteria and a test plan. Each US has conversations that happen during backlog grooming and iteration planning to solidify the details. Additionally, especially in large enterprise ventures, US are grouped in the epics and themes (Cohn 2004) (see Figure 1). 263 FACTORS INFLUENCING USER STORY ESTIMATIONS … Figure 1. Themes, epics and user stories Source: based on Cohn (2004) Refinement Process and Development. In enterprise projects, a US is p (...truncated)