An Empirical Investigation of Pull Requests in Partially Distributed BizDevOps Teams

TL;DR

This paper investigates the challenges of using pull requests for collaboration in partially distributed BizDevOps teams, highlighting barriers and factors affecting success in cross-site cooperation.

Contribution

It provides an empirical analysis of pull request barriers in distributed BizDevOps teams, emphasizing domain complexity, process differences, and employee turnover.

Findings

Pull requests work well locally but face challenges across sites.

Barriers include domain complexity, process differences, and employee turnover.

Cross-site collaboration is hindered by communication and organizational issues.

Abstract

In globally distributed projects, virtual teams are often partially dispersed. One common setup occurs when several members from one company work with a large outsourcing vendor based in another country. Further, the introduction of the popular BizDevOps concept has increased the necessity to cooperate across departments and reduce the age-old disconnection between the business strategy and technical development. Establishing a good collaboration in partially distributed BizDevOps teams requires extensive collaboration and communication techniques. Nowadays, a common approach is to rely on collaboration through pull requests and frequent communication on Slack. To investigate barriers for pull requests in distributed teams, we examined an organization located in Scandinavia where cross-functional BizDevOps teams collaborated with off-site team members in India. Data were collected by conducting 14 interviews, observing 23 entire days with the team, and observing 37 meetings. We found that the pull-request approach worked very well locally but not across sites. We found barriers such as domain complexity, different agile processes (timeboxed vs. flow-based development), and employee turnover. Using an intellectual capital lens on our findings, we discuss barriers and positive and negative effects on the success of the pull-request approach.