Onboarding & Teamwork
An On-Call Shift Handoff: Transfer Current Risks, Open Incidents, and Next Actions
Transfer on-call coverage with current incident state, expiring mitigations, approved access, confirmed alert routing and explicit acceptance. Follow a fictional failed-coverage branch into a reconciled shift transfer.
An on-call shift handoff should transfer current operational responsibility with an explicit incoming acknowledgment, not just deliver a list of links. Put open incidents, customer impact, temporary mitigations, expiring exceptions, alert routing, and the first next actions at the top. Confirm that the incoming person can access the required systems through approved permissions, and use the established coverage process if they cannot accept the shift.
This LandOffer workplace guide develops a fictional handoff with an incident still open and a receiving engineer whose permissions need repair. All systems, observations, approval records, coverage changes, and checks are invented examples. No production action, paging test, access change, or real incident response was performed while preparing this article.

Separate shift coverage from incident command
A shift change can happen while several kinds of responsibility remain active. The incoming on-call engineer may receive alerts and carry assigned operational actions, while an incident commander continues coordinating the response. A service owner may retain approval for a mitigation change. Do not silently transfer all three roles because the rota changed.
Google's Managing Incidents chapter describes an explicit incident-command handoff, firm acknowledgment, and communication of the change to the response team. Its example concerns incident command specifically. Apply the need for clarity without assuming that every on-call shift change also changes the commander.
In this fictional case, Vince is the outgoing primary engineer for the IndexBridge summary-processing service. Greta is scheduled to become primary at 17:00 UTC on October 27, 2026. Drew remains incident commander for open incident I-42. Beatriz is the service owner responsible for approving changes to its temporary mitigation. These roles are stated separately in the handoff.
This is an operational shift transfer, not a general project handover or an ordinary cross-time-zone task relay. The important question is who currently holds coverage and can respond through the team's approved process, including when the scheduled receiver cannot take responsibility as planned.
Prepare a current-state packet before the boundary
Lead with the open incident, nearest expiry, actual coverage owner, and first assigned check. Link the authoritative incident record and current runbook, explain what each contains, and timestamp the summary. A dashboard captured earlier in the shift is useful history but may no longer describe the service at handoff.
GitLab's public on-call handover procedure includes shift activity, notable incidents and actions, issues that may continue or recur, and priorities for the incoming team. Its cadence, report workflow, and assignment conventions belong to that employer. They are examples of operational information to check, not a universal rota or reporting requirement.
Include what changed during the shift and what remains unverified. “Errors improved after a mitigation” is different from “the cause is fixed.” A warning that has been quiet because a suppression is active is different from a warning that has stopped occurring. Keep suppression status and mitigation limits visible so quiet notifications do not become a false recovery signal.
Keep credentials out of the packet. Point to approved access routes and named owners, and verify the receiver's actual ability to use the required role. Check the required current operator role and validity window, in addition to confirming that login works.
A filled outgoing shift record with an open incident
The following packet is fictional revision 1, prepared at 16:40 UTC on October 27, 2026. References such as I-42 and runbook revision 6 are illustrative labels, not live links or operational instructions to execute.
| Handoff item | Current fictional record | Incoming action or boundary |
|---|---|---|
| Coverage and roles | Vince is current primary until an approved successor accepts; Greta is scheduled for 17:00. Drew remains I-42 incident commander; Beatriz owns mitigation-change approval | Confirm the actual primary, authorized permissions, and alert destination before announcing the transfer |
| I-42 impact | New summary processing is delayed for some workspaces. Existing completed summaries remain readable in current checks. Affected workspace count and cause are not confirmed | Keep I-42 open; consult the timestamped incident state rather than claiming all service functions are healthy |
| Temporary mitigation | An approved ingestion restriction is active. Its authorization expires at 18:00 UTC; reassessment with Drew and Beatriz is scheduled for 17:40 | Track the expiry and decision owner. Do not assume permission to renew, remove, or change the restriction |
| Alert exception | A duplicate-warning suppression expires at 17:20 UTC. The underlying warning condition may still exist | Review the exception's scope and expiry in the approved record; quiet notifications do not prove recovery |
| Current references | Incident state revision 4; service runbook revision 6; mitigation approval record M-8; alert exception A-3 | Check these current references are accessible. An older runbook is marked superseded |
| Work already attempted | A previous worker restart did not remove the delay. No repeat attempt is currently assigned | Preserve the result and coordinate further operations through Drew, rather than repeating a discarded branch without new evidence |
| Expected change | A separate configuration rollout is paused while I-42 is open; its owner is aware | Do not interpret the shift boundary as authorization to resume the rollout |
| Pending follow-up | Next incident observation checkpoint 17:25 UTC; mitigation reassessment 17:40; exception expiry 17:20 | State who owns each next check after coverage is reconciled |
The packet says what is active, what is expiring, and what the receiver may not infer. It deliberately avoids providing commands for changing the service or exception. Those actions require the actual authorized runbook and response process, not an example in a blog article.
Repeat both expiry records in the readback: A-3 for notifications and M-8 for the mitigation authorization. The warning suppression ends at 17:20, while the mitigation authorization ends at 18:00. A receiver who notices only “temporary exception” could miss one of them or assume they share an owner. Separate the records and repeat their exact times during the readback.
Let the receiver question the handoff before accepting
During the fictional 16:45 handoff conversation, Greta asks: “Which responsibility moves at 17:00, and does Drew remain commander? I can read incident revision 4 and the dashboards, but my production operator role shows an expiry of 16:55. I cannot confirm the permissions required for primary coverage after that time. Is there approved cover while Beatriz checks the role?”
Vince confirms that Drew remains commander and records the permissions problem. Greta has discovered a barrier to acceptance, not volunteered to bypass it. Read-only visibility can support investigation, but it is not equivalent to holding the authorized operator responsibilities required by this fictional rota.
The question changes the plan before the boundary. At 16:50, coverage lead Nestor invokes the existing substitution process. Leon, the designated backup, confirms availability and current approved operator access, and accepts primary coverage from 17:00 until the next explicit transfer, within his agreed 17:00–18:00 cover window. Nestor records the temporary assignment and approved alert route. Greta accepts only read-only support while her access issue remains open.
At 17:00, Leon's acceptance is announced in the response channel and entered in the coverage record. Vince leaves primary coverage at that confirmed boundary; the team has not silently required him to remain indefinitely or pretended that Greta's schedule entry established working access. If no authorized backup had accepted, Nestor would have continued the established coverage escalation rather than marking the shift transferred.
Google's Being On-Call chapter discusses on-call response responsibilities, escalation, and response expectations agreed by teams and service owners. Its numerical response examples are specific to its context. Use your own approved coverage and escalation rules; neither this case's times nor a published example establish a universal response SLA or obligation to work beyond an agreed shift.

Reconcile access, routing, and the final readback
The following is the completed fictional reconciliation record. It shows what changed and what remains pending; the described access and routing checks were not performed by the article's author.
17:10 — Approved access repair: Beatriz completes the role-renewal process through the existing access system. Greta confirms that the required operator role is now valid for her assigned shift and that she can open runbook revision 6 and approval records M-8/A-3. No passwords or tokens are copied into the handoff.
17:15 — Coverage and alert-route check: Nestor and Greta confirm the current primary assignment and the approved paging destination in the coverage system. The fictional record also references the team's authorized notification-delivery check. This establishes the route within the example; it is not a live test receipt or permission to send an unsolicited production page.
17:18 — Explicit transfer from Leon to Greta: Greta says: “I accept primary coverage now under the confirmed operator role and alert route. Drew remains incident commander. I-42 is still open, the duplicate-warning suppression expires at 17:20, and I own the 17:25 observation checkpoint. I will bring the mitigation state to Drew and Beatriz at 17:40, before its 18:00 authorization expiry. I will not renew it on my own.”
Leon acknowledges the transfer. Nestor records 17:18 as the actual handoff time and informs the response channel. The original 17:00 schedule entry remains history; it is not rewritten as though Greta accepted then. Leon's temporary primary period was 17:00–17:18, within the cover window he accepted.
17:20 — Alert exception expiry: Greta records that the duplicate-warning suppression has reached its configured expiry and reviews the current warning through the runbook's approved route. The incident record retains the underlying condition as a monitoring concern. The end of the exception is not recorded as resolution of I-42.
17:25 — Observation checkpoint: New summary processing remains delayed, though the current observations show improvement relative to the earlier checkpoint. The affected count and cause are still unconfirmed. Greta adds the observation to incident state revision 5 and leaves I-42 open. The 17:40 reassessment and 18:00 mitigation-authorization expiry remain future responsibilities; no extension or final recovery is claimed.
The fictional record confirms Greta's 17:18 coverage acceptance while leaving I-42 and the mitigation reassessment open. The handoff succeeded in identifying an access gap and assigning actual coverage. It did not establish that the service was repaired, that all alerts would arrive, or that no risk remained.
Use a failed-coverage branch instead of an assumed receipt
If an incoming person cannot be reached, cannot access required materials, or declines because they are not the assigned receiver, keep the transfer visibly unaccepted. Use the established coverage lead or escalation route to obtain an authorized replacement. Do not solve the record by assigning someone without their knowledge or handing them another person's credentials.
The appropriate interim arrangement depends on the team's staffing, permissions, and coverage policy. This article cannot determine whether an outgoing engineer extends, a backup takes over, or a manager changes the rota. The case shows one approved substitute, with a bounded window and an explicit later transfer, rather than prescribing an arrangement for every employer.
During an active incident, tell the response team who currently owns each operational role and which transfer is pending. Do not leave two engineers both believing the other is primary, or both making uncoordinated changes because they think they share responsibility. Questions about command, production authority, and customer messaging should follow their respective owners.
Finish with a current record and named next checks
At shift end, update the packet's snapshot and preserve revisions so later readers can distinguish the planned boundary from the actual acceptance. Link the open incident, active exceptions, mitigation authorization, current runbook, and coverage record. Keep the 17:40 reassessment owner and 18:00 authorization expiry prominent in the final open-work record.
For your next handoff, ask the receiver to read back the open risk, next action, permission boundary, and owner of any expiring mitigation. Record acceptance only after the required coverage checks, using the actual process your team has approved. Leave unresolved incident work visible for the next shift. More workplace guides are available through LandOffer.
Sources and Further Reading
- Google SRE: Managing Incidents — live incident state and explicit command handoff; original SRE book context.
- Google SRE: Being On-Call — assigned operations, escalation, and team-specific response expectations.
- GitLab: On-call handover — shift records, continuing incidents, and incoming-team priorities; modified March 5, 2026.


