On-Call Handoff: The 60-Second Ritual That Prevents Dropped Incidents
Handing off the pager is a 60-second ritual or a series of small disasters. The structured handoff that keeps active incidents from being dropped at shift change.
The 60-second ritual
A clean handoff is short. The departing on-call hands the next on-call a five-item summary. They both confirm receipt. The rotation tool gets updated. Done. Skipping the ritual to "save time" is how 6am pages land on someone who didn't know they were on rotation.
The ritual is short by design. Anything longer than 60 seconds creates avoidance; on-callers skip it when they're tired. Anything shorter than 30 seconds tends to miss real handoff items. The 60-second target is what makes the ritual sustainable across a year of weekly rotations.
The other reason for the short ritual: the cost of a missed handoff is dramatically higher than the cost of including it. A 6am page landing on someone who didn't know they were primary costs an hour of slow response. The 60-second ritual, done weekly, costs 52 minutes per year per engineer. The math is overwhelming.
The five items to hand off
- Active incidents: any open incidents the new on-call inherits. Include severity, current theory, current action, and the channel where the work is happening.
- Recent flaps: alerts that flapped in the last 24 hours and may flap again. The new on-call needs to know to triage these as "probably noise, but check anyway."
- Known issues: ongoing degradations the team is tracking but not actively fixing. Common case: a vendor outage we're working around, a slow-cooking migration, a feature flag mid-rollout.
- Coming changes: deploys, migrations, vendor maintenance landing on this shift. The new on-call should know what's scheduled so they can correlate alerts to known changes.
- Comms commitments: customer messages promised but not yet sent. If the previous on-call promised a customer an update by 14:00, that promise needs to land with the new on-call.
Each one is a sentence or two. The full handoff is under 60 seconds in normal weeks; longer if there's an active incident.
The discipline is to include all five even when most are "none." "No active incidents, no recent flaps, no known issues, the v2.4 deploy is finishing tonight, no comms commitments" takes 15 seconds and is a complete handoff. Skipping items because they're empty is what creates gaps; the new on-caller doesn't know if "no active incidents" means "checked and confirmed" or "didn't think to check."
Handing off an active incident
If an incident is open at shift change, both on-callers stay on the bridge for 15 minutes. The departing IC briefs, the new IC takes over the channel, the bridge has the new IC name in the topic. Without overlap, the incoming on-call walks into a hot bridge with no context.
The 15-minute overlap exists because incident context doesn't transfer in a 60-second message. The new IC needs to hear the team's current theory, the alternatives that have been ruled out, the customer comms commitments, and the next 30 minutes of planned action. Compressing this into a written handoff loses the nuance that determines whether the next 30 minutes go well or poorly.
The protocol. The current IC says "I'm handing IC at the top of the next status update." The new IC joins, listens to one full status cycle (5-10 minutes), then takes the IC role at the next cycle boundary. The departing IC stays on the bridge for another 5-10 minutes as a safety net; if the new IC freezes, the departing one is still there to coach.
Calendar discipline
The handoff is a real calendar event, every shift, recurring. Not a Slack message. Not a "ping me when you're up." A calendar event with both names attached, 15 minutes long.
The reason for the calendar event: it makes the ritual unmissable. A Slack message can be missed if the departing on-call ends their shift before the incoming one comes online. A calendar event with two attendees gets reminders, gets honoured, and shows up on both calendars.
The 15-minute slot is intentional: longer than the 60-second handoff so there's time for clarifying questions. Most weeks the slot is half-empty; that's fine. The slot exists for the weeks when there IS an active incident or a tricky known issue, and you'd rather have the time and not need it than need it and not have it.
Two failure modes
"Nothing to hand off." The handoff happens but is empty. Often used to skip the ritual; later the incoming on-call discovers a fire that was already smouldering. Always state the five items even if four of them are "none." The discipline of stating them forces the departing on-call to actually check, rather than assuming.
"Slack handoff in writing." Works in theory; in practice the new on-call doesn't read it for two hours and the smoke turns into flame. Written handoffs work as a SUPPLEMENT to the synchronous one, not a replacement. The synchronous moment is what creates accountability; the written record is what makes the handoff durable.
A third failure mode worth naming: "Aspirational handoff." The departing on-call describes what they HOPED would resolve before their shift ended ("the v2.4 deploy should be done by now"), not what's actually true. The incoming on-call thinks something is resolved that isn't. Force concreteness: "v2.4 deploy started at 22:00, expected completion at 23:30, current status is still rolling out, here's the dashboard link."
Weekend and holiday handoffs
Weekend and holiday handoffs are where most teams have their worst silent failures. The Friday-evening handoff to weekend on-call gets rushed because the departing engineer is leaving for the weekend. The pre-holiday handoff gets skipped because "nothing's going to happen anyway." Both are when the worst incidents land.
The fix: extra rigor on the boundaries. The Friday handoff includes a 30-minute slot, not 15. The pre-holiday handoff is in writing AND in person, and includes the manager as an FYI so there's a third person aware of the handoff state. The first day after the holiday includes a "what happened over the holiday" report in the standup.
The other extra discipline for these boundary handoffs: explicit escalation paths. "If anything SEV2 or higher happens between now and Tuesday, here's who to escalate to and how." Written down, linked from the handoff. The default escalation chain might not work if half the team is on holiday; the boundary handoff is when you confirm the alternative path.
Documenting rotation context
Beyond the per-shift handoff, every rotation should have a "current state" document that persists. What services this rotation owns, the SLOs, the active long-running known issues, the contact tree. The handoff updates this document; the document is what new rotation members read on their first shift.
Without the persistent doc, every new rotation member has to learn the context from scratch. Onboarding a new on-caller takes weeks instead of one shadowed shift. The doc is the difference between "we have a rotation" and "we have a sustainable rotation."
The doc lives next to the runbooks (most teams use the same wiki/Notion). The handoff post is short because most context lives in the doc; the handoff just covers what's CHANGED since the doc was last updated.
What to do this week
Three moves. (1) Add a weekly recurring 15-minute calendar event for handoffs, both names attached. The calendar discipline alone catches 80% of skipped-handoff failures. (2) Pin the five-item handoff template in your on-call channel. New on-callers reach for it; veterans use it as a checklist. (3) Audit your last 4 handoffs: did all five items get covered? If you find any "yeah, we skipped that one because nothing was happening," that's the team's pattern to fix.