When tracking development performance, it is important that:
Metrics are simple, fully automated, and standardized
Performance can be benchmarked against industry peer groups
Setup is minimal (requiring connecting no more than a Git provider)
The most important metric is New Features per Engineer, a proxy for productivity. Elite companies release more new features to market than their peers.
While a balance of bug fix and refactor is needed to maintain and improve quality, companies should optimize for shipping more new features on a sustained, long-term basis.
Consider an example: Acme, a retail company, that recently raised Series D funding and is expanding into new markets and geographies. Acme is hiring quickly in the engineering department, but productivity is dropping at an accelerating rate due to ineffective leadership and a lack of standardized processes. They are accumulating technical debt and it now takes weeks instead of days to release new features to market.
Acme is scaling up inefficiently.
Devasphere, a competitor to Acme, is also hiring, but less quickly as they invest in new processes and tools. For instance, they built out a top-notch onboarding experience for new developers, rolled out AI coding tools across their organization of over a hundred developers, and re-organized their product and engineering teams to better support expansion into new markets and geographies. It takes less than a day to release a new feature.
Devasphere is scaling up efficiently.
Devasphere also implemented a Development Observability platform so that they have visibility into the ROI of their investments. They present engineering metrics to their board, which has confidence that they are building a world-class software development team.Argon, a global retailer that operates in the same markets, was recently acquired by a private equity firm after a couple years of stagnant growth. They are focused on reducing costs across sales, marketing, and engineering.
Argon shifted most of their engineering team offshore while maintaining quality after determining that those teams operated with a higher capital efficiency (a lower cost per new feature) than onshore teams.Argon is scaling down efficiently: as they scale down, their productivity per engineer is improving.
Similarly, understanding the stages of lead time — the time it takes to commit, review, test, and merge a feature — can help you identify bottlenecks in the delivery process.
Companies with a culture of fast, iterative development generally have a faster lead time, while slower lead times usually indicate the presence of bottlenecks, delays, and less automation.
Measuring and improving development performance is a continual process. By surfacing data at every level of the organization — including investors, heads of engineering, department heads, managers, and individual contributors — companies can create feedback loops that drive organizational change.
For instance, heads of engineering may track quarterly changes in efficiency and adjust hiring plans accordingly, while managers may use weekly reports to track their team’s work in progress.
Measuring engineering performance works best when combined with a culture of trust and transparency. A key consideration is protecting individual privacy, which enables companies to create a workplace that values psychological trust and safety, a key driver of innovation and organizational learning. Choosing the right metrics and implementing them in the right way is the key to setting teams on the path to high performance.
Measuring development performance is essential for tech-enabled companies that want to stay competitive in today’s fast-paced market.
Companies that have visibility into development performance — including how their development teams are performing relative to industry peers — and create a culture of continuous improvement are most likely to win markets.