On-call is a critical responsibility for engineering teams that run production systems. Managed well, it builds ownership and operational excellence. Managed poorly, it causes burnout, resentment, and attrition. This guide covers how to design an on-call programme that is fair, sustainable, and genuinely improves your system's reliability.
The Purpose of On-Call
On-call exists to ensure that production systems receive timely attention when things go wrong outside of working hours. But its value extends beyond incident response. A well-designed on-call programme creates a feedback loop between the engineers who build systems and the operational reality of running them. When engineers are on call for their own services, they build more reliable systems.
The goal is not to have engineers available at all hours - it is to have a clear, predictable process for responding to production issues with minimal disruption to engineers' lives. This distinction matters because it shapes every design decision in your on-call programme.
- On-call creates a feedback loop between building and operating systems
- It should be predictable, fair, and minimally disruptive to engineers' lives
- Engineers who are on call for their own services build more reliable systems
- On-call quality is a direct input to team retention and satisfaction
Designing Fair On-Call Rotations
Fairness is the foundation of a sustainable on-call programme. Every eligible engineer should share the on-call burden equitably. Rotation schedules should be published well in advance - at least one month - so engineers can plan their personal lives around on-call weeks. Provide easy mechanisms for swapping shifts when conflicts arise.
The minimum team size for a sustainable on-call rotation is four to five engineers. Fewer than that creates an unsustainable burden where engineers are on call too frequently. If your team is too small for a dedicated rotation, explore shared on-call arrangements with other teams or escalation models where a broader group provides primary coverage.
Consider time zones when designing rotations for distributed teams. Follow-the-sun models, where on-call responsibility shifts to engineers in different time zones during their working hours, can dramatically reduce the impact on any individual while providing round-the-clock coverage.
Reducing the On-Call Burden
The best way to make on-call sustainable is to reduce the number of pages. Track alert volume, false positive rates, and time-to-resolution for every on-call shift. If engineers are regularly paged for non-urgent issues, your alerting thresholds need adjustment. If the same issue pages repeatedly, invest in fixing the root cause rather than continuing to respond reactively.
Establish clear escalation paths and runbooks. When an engineer is paged at two in the morning, they should not have to think about what to do from scratch. Well-maintained runbooks that provide step-by-step resolution guidance for common issues dramatically reduce response time and cognitive load.
- Track alert volume and false positive rates for every on-call shift
- Fix root causes of recurring pages rather than continuing to respond reactively
- Maintain runbooks with step-by-step guidance for common issues
- Establish clear escalation paths so on-call engineers are never alone
Compensation and Recognition for On-Call
On-call work should be compensated. Whether through additional pay, compensatory time off, or other mechanisms, engineers who sacrifice their personal time for operational coverage deserve tangible recognition. Work with your management chain and HR to establish fair compensation policies.
Beyond formal compensation, recognise on-call contributions publicly. Acknowledge engineers who handled difficult incidents well. Review and address the systemic issues that make on-call painful. When engineers see that their on-call feedback leads to actual improvements - better alerting, more reliable systems, improved runbooks - they feel valued and are more willing to participate.
Common On-Call Management Mistakes
The most damaging mistake is normalising an unsustainable on-call burden. If engineers are regularly losing sleep, missing personal commitments, or dreading their on-call weeks, the programme is broken. Do not dismiss complaints as a lack of commitment - treat them as signals that the programme needs investment.
Another common error is exempting senior engineers or managers from on-call. This creates a two-tier system that breeds resentment and disconnects leadership from operational reality. Everyone who builds production systems should share the operational responsibility, including the engineering manager when appropriate.
Key Takeaways
- On-call should be predictable, fair, and minimally disruptive
- Reduce alert volume and false positives to make on-call sustainable
- Compensate on-call work through pay, time off, or other tangible mechanisms
- Maintain runbooks and clear escalation paths for common incidents
- Never normalise an unsustainable on-call burden - fix the root causes
Frequently Asked Questions
- Should engineering managers be in the on-call rotation?
- In small teams, yes - it demonstrates shared responsibility and keeps you connected to operational reality. In larger teams, you should be available as an escalation point rather than in the primary rotation. Regardless, you should regularly review on-call reports, understand the burden your team carries, and actively work to reduce it.
- How do I handle engineers who refuse to participate in on-call?
- First, understand their concerns. If the objection is about fairness, compensation, or unsustainable burden, address those issues systemically. If the objection is fundamental - they believe on-call is not part of their role - have a clear conversation about expectations. On-call is a standard responsibility for engineers who build production systems, and this should be established during hiring, not after.
- What metrics should I track for on-call health?
- Track pages per on-call shift, time-to-acknowledge, time-to-resolve, false positive rate, and after-hours page frequency. Also track qualitative feedback: ask engineers to rate their on-call experience after each shift. Review these metrics monthly and use them to drive improvements in alerting, reliability, and runbook quality.
Build Your On-Call Programme
Access on-call rotation templates, runbook frameworks, and incident response guides designed for engineering managers building sustainable operational programmes.
Learn More