Figure 1. Visualization of the research model and method used in each phase.

3.1 Step 1 – Exploratory Case-study

The first step of this study’s exploratory nature aims to answer RQ1, RQ2.1, RQ2.3, RQ2.4 and RQ3 and is described in the following subsections.

3.1.1 Phase 1 – Contextual analysis and design

Case studies have been long established in the software engineering field to explore how different processes and tasks are carried out in practice in the industry. Due to the exploratory nature of the study, we, therefore, employed a multiple qualitative case study approach [1] to gain a detailed understanding of how PD arises and its related adverse consequences.
First, the study was presented and discussed during a workshop with one or more participants from each of the five involved companies. Such participants were always employees with management responsibilities related to the processes, methods, and tools employed in the organizations. This phase acted both as an introduction session where each company briefly described their overall software development process together with an assessment of concerned stakeholders of the process. The goal of these workshops was to introduce the participating companies to the study, to align and equip them with relevant knowledge about the concept of PD (we showed the definition, we explained the metaphor, and we reported a few anecdotal examples) and to gather background and contextual information on each participating company in preparation for the following interviews. Further, the aim of this stage was also to communicate and describe what the ”process” term refers to within this study’s context and also to identify potential interviewees for the next stages of the study. Each workshop lasted from 30 to 60 minutes and was digitally recorded.

3.1.2 Phase 2 – Qualitative Data Collection

The data collection method was a combination of interviews together with the analysis of internal company documents. This study employed a combination of the technique of unstructured and semi-structured interviews where the questions were both formulated as general concerns and interests from the researcher about PD. However, the interviews also had several questions that were prepared and formulated in advance [2].
The interviews aimed to explore the concept of PD within each of the companies and what aspects impact and drive such debt (see Table I). We started by asking the interviewees to describe their overall software development process, both from a historical and a systematic perspective. We asked four follow-ups to learn about the aspects of the process, focusing on the characteristics described in Section 3.1.1 (according to our RQs).