Quick answer: Small studios skip on-call until something breaks at 2am. A weekly rotation with clear escalation, a runbook per system, and shared compensation prevents the burnout that ad-hoc on-call creates.
An on-call rotation isn't overhead. It's the structure that lets one engineer rest while another covers - which is the only way you stay healthy as a team.
Weekly cadence
One engineer carries on-call per week. Hand off on Friday afternoon; the next person has the weekend to settle in. Shorter rotations exhaust everyone; longer ones become punishment.
Two-tier escalation
Primary handles inbound. If unreachable in 15 minutes, secondary is paged. After 30 minutes uncovered, the engineering lead is paged. No more than three tiers.
Pay for the time
On-call is work. Flat weekly fee plus per-incident escalation pay. Without compensation, on-call becomes a quiet quit driver.
Runbook per system
Each system has a runbook in the same repo as the code. New systems require a runbook before launch. Stale runbooks are flagged in monthly review.
“On-call is a system. Treat it like one.”
If you're at three engineers, you're not too small for on-call. You're at the size where ad-hoc on-call breaks the first engineer fastest.