Unlocking high engineering pace: Prevent unplanned work

Every sprint a team can lose a quarter of its capacity to outages, urgent bugs, and emergency fixes, and the symptom is obvious: planned features never ship and morale erodes. The article quantifies the impact, showing that 30-40% of engineering time spent on unplanned work translates directly into 60-70% of expected output, feeding a perception that the organization can't keep its promises.

The author proposes a three-category model to make sense of the chaos. "Control" covers incidents the team creates-production outages, defects, infrastructure failures, and data quality problems. "Influence" includes problems that arise from the wider environment, such as vulnerable dependencies, vendor breaking changes, and urgent business requests. "Accept" is anything truly external, like regulatory deadlines or zero-day exploits, which can only be communicated and managed.

For the controllable slice, the article lays out concrete expectations: observability that catches issues before customers notice, automated testing that embeds quality, staged rollouts with rapid rollback, and disciplined safeguards even under pressure. Reducing technical debt, tightening acceptance criteria, and enforcing code-review rigor turn many incidents into preventable events, directly boosting velocity and predictability.

The piece ends with a simple playbook: measure unplanned work for a month, categorize each incident, set quarterly targets to eliminate the Control-category work, allocate capacity to address Influence-category problems, and communicate openly about Accept-category events. By turning fire-fighting into data-driven improvement, teams regain focus, improve morale, and deliver predictable business value.