On-call burnout: what to fix first

Why on-call is different from ordinary stress

Most work stress fades when you log off. On-call removes that boundary. Even a quiet week carries the low-level vigilance of knowing the phone could go off, and a loud week adds broken sleep on top. Then the next morning is usually just a normal workday.

It also affects the quality of the work. Google's SRE book points out that stress hormones can "impair cognitive functions and cause suboptimal decision making," and that under stress, a deliberate approach can give way to unreflective action. In other words, a stressed responder is more likely to make an incident worse. Reducing on-call stress is a reliability concern as well as a wellbeing one.

Fix the system first

Burnout comes from chronic workplace stress, so individual coping can't be the whole answer. Start here.

Measure the load against a benchmark

Google's SRE book gives two figures worth comparing yourself to. It sets a maximum of two incidents per 12-hour on-call shift, based on its finding that an incident takes about six hours on average to handle, including root-cause analysis, remediation and follow-up. It also says no more than 25% of an SRE's time should go to on-call.

Those are Google's numbers for Google's teams, and yours may differ. Still, if your team is routinely far past them, that's your first problem to solve.

Then ask these questions

• Alert quality. In the last 90 days, how many pages required a human to act? Delete or demote the rest.
• Rotation size. How often does each person come up, and are the same few people always the real escalation path?
• Recovery time. After a night page, is it normal to start late or take time back, or does someone have to ask?
• Follow-through. Do postmortem actions actually get scheduled, or does the same incident keep returning?
• Hand-offs. Does the next person inherit a clear picture, or a mess they have to reconstruct?
• Blame. Do people feel safe saying "I made that worse at 3 a.m."?

Then help people recover

However good the rotation is, some nights will be rough. These habits are low-cost, and they're worth trying. They don't replace fixing the load.

Before your shift

• Make a plan for the week: lighter commitments, a known backup and a charged phone in a set place.
• Tell the people you live with what a page means, so they aren't left guessing at 3 a.m.

During a page

• Before you touch production, take one slow breath out that's longer than your breath in.
• Say your assumption out loud or in the channel. It forces the slower thinking that stress tends to crowd out.
• Ask for help sooner than feels necessary.

After the page

• Don't go straight from the terminal to bed. Take a few minutes to write down what's left for tomorrow, so your brain can stop holding it.
• Try a short downshift. In one small randomized trial, five minutes a day of exhale-focused "cyclic sighing" improved mood more than mindfulness meditation. That study wasn't about on-call engineers, but the practice is free and easy to test. Our two-minute reset walks you through a short version.
• Keep your screen dim and resist the urge to check the dashboards again.

After the shift

• Plan something restorative the following day, and treat it as part of the rotation rather than a reward.
• Write one line on what would have made this week easier, and bring it to your next review.

Where a workshop fits

If you've already looked at the load and your team is still running hot, shared skills can help. A live session gives people practice in staying steady under pressure and a common language for saying when they aren't. See how our engineering workshops work.