Proemion optimizes team and system boundaries for flow with CodeScene and Conflux
How IoT leader Proemion used CodeScene and guidance from Conflux to evolve healthy team and system boundaries for flow.
Proemion is the leading telematics platform provider for OEMs, offering a seamless integration of hardware, software, and connectivity, ensuring continuous business success for its customers. With a fully integrated telematics stack providing platform monitoring, analysis, prediction, remote diagnostic, and access capabilities, Proemion makes it easy for businesses to manage fleets and boost efficiency. After joining Battery Ventures in 2022, Proemion expanded in 2024 by acquiring Trendminer, a company specializing in industrial process and asset monitoring.
At Proemion, around 60 engineers across 10 teams are continually working on generating, collecting, and presenting key operational data via their flagship DataPortal product, and these teams are growing with customer demand. Umberto Nicoletti, Head of R&D at Proemion, realized that to support the company’s continued sustainable growth of their software engineering capability, it would need to adopt principles and practices informed by fast flow, such as Team Topologies.
In particular, it became clear that as the number of engineers increased, it was vital to establish good boundaries for flow, enabling teams to work with empowered autonomy whilst still contributing effectively to the whole solution. In 2024, after watching a talk from Conflux CEO, Matthew Skelton on untangling software delivery, Proemion began to use CodeScene to assess codebase health.
“I decided to trial CodeScene after having seen it recommended in the ‘untangling’ presentation by Matthew Skelton at the DORA community in 2023”, said Nicoletti. He needed to understand why some teams were struggling with delivery compared to others. Were some teams blocked more often, were delays due to code quality, or some other problem? Nicoletti commented, “A qualitative assessment with the respective Team Leads suggested that there was room for improvement, however, we could not pinpoint exactly what the obstacles were.”
Initially, Nicoletti used the concept of counting “blocking waits” to assess team independence (as outlined in Matthew Skelton’s talk), however, their findings were counterintuitive: the team with the highest delivery performance reported the most blocking waits. After some discussion, Nicoletti realized that a high-maturity team is typically more aware of its surroundings and limitations, and more readily able to report blockers as a result.
When they used CodeScene to compare the context of different teams, it provided clarity and pinpointed the problems faced by the lower-performing team. “CodeScene on the other hand told a story that fit my perception and that of the team lead perfectly”, said Nicoletti. “CodeScene showed us that the underperforming team exhibited a cluster of challenges: it was the only team with decreasing code quality, a few significant refactoring hotspots, knowledge loss due to staff leaving, and had low team effectiveness (we had scaled up the team size but failed to reap the full benefits of it).”
CodeScene helped the leadership at Proemion identify the causes of lower team performance around software delivery and provided actionable recommendations for addressing the problems, such as refactoring steps, hotspot analysis, and code health trends.
Using team analysis in CodeScene to map to Team Topologies principles
CodeScene provides powerful capabilities to detect and visualize team boundaries based on the patterns of activity in code repositories, drawing on concepts explored in the book ‘Team Topologies’, co-authored by Conflux CEO, Matthew Skelton. Any mismatches between the software architecture and the team architecture are easy to spot using CodeScene, via its team dynamics features.
Nicoletti used these features to assess the effectiveness of team boundaries at Proemion, in particular looking for cross-team dependencies. The analysis showed that a change from a waterfall to a stream-aligned model 18 months before had been successful in minimizing cross-team dependencies. “CodeScene gave us confidence that we did not have a major problem with blocking waits, due to the earlier reorganization inspired by Team Topologies”, said Nicoletti.
CodeScene’s team alignment explorer demonstrates that the model change from waterfall to stream-aligned reduced the communication channels outside of the team.
In this case, the circle on the team on the left-hand side represents an individual who moved from the bottom right team to the left, and Codescene correctly portrays the strong ties to their former team.
Comparing CodeScene to other code quality tools (like SonarCloud)
Led by Nicoletti, the engineering team at Proemion also used CodeScene alongside SonarCloud for a “multi-level” approach to code quality assessment.
“We found that CodeScene and SonarCloud complement each other when it comes to assessing codebase health”, said Nicoletti. “CodeScene focuses on how code quality affects team dynamics (and vice-versa), whereas SonarCloud focuses on ‘single line’ code quality, without any real context about the social or organizational aspects. With that in mind, they are both helpful, because SonarCloud helps with static code analysis on each single change and is therefore useful for the developers in their day-to-day work.”
CodeScene’s dashboard shows steadily increasing code quality and team alignment. This data matches the perceived team(s) performance as they work on the codebase that has been powering the Proemion telematics platform for over 15 years.
“In contrast, CodeScene’s focus on organizational dynamics is relevant for an audience such as team leads, engineering managers, and CTOs who can influence organizational structure and team boundaries”, explained Nicoletti. “Thanks to CodeScene, we discovered that we’re well set up from a Team Topologies perspective, which helped us to identify that the problem with the lower-performing team seemed to be down to the team not scheduling time to tackle the hotspots. This is something that other tools more or less in the same space (like SonarCloud) fail to capture and show as clearly as CodeScene does.”
CodeScene provides actionable insights for reducing time-to-value
Thanks to guidance and suggestions from Conflux, the R&D department at Proemion found practical approaches to assessing and remediating lower performance in their software delivery teams. The use of CodeScene for social code analysis was crucial to giving clarity on the nature and location of the problems.
“CodeScene backs the perceived team performance problems with data and provides actionable next steps”, said Nicoletti. “Codescene excels in revealing the dynamics and factors that affect software delivery performance through clear, actionable diagrams.”
