# Can your human support team absorb a shared service failure?

By Sean Everett · October 6, 2026

Model simultaneous customer exceptions, protect urgent work and communication capacity, and rehearse the backlog that remains after the underlying service recovers.

Topics: Service operations · Incident readiness · Automation

## Model the shared event and its human workload

Choose a dependency whose failure could affect many customers together. Estimate the work that would reach people: repeated contacts, individual workarounds, approvals and recovery checks. Use ranges for the number of affected customers and handling time, then compare the demand with available skills and shifts.

Atlassian’s review of its April 2022 outage describes disrupted customer-contact records and support queues, followed by substantial manual validation of restored sites. The case illustrates how a shared incident can impair support tools while also creating new human work. It provides no universal staffing ratio.

Include a degraded-tooling scenario. Check whether staff can find approved contacts, record customer-specific impact and coordinate work when the usual portal or identity service is unavailable. Count restoration validation separately from incoming messages so an apparently quiet queue cannot hide unfinished customer work.

## Allocate attention using explicit service priorities

Atlassian’s service-management guidance distinguishes business impact from urgency, including how soon the impact matters. Apply that distinction to the exception queue. Agree the conditions requiring immediate attention before an incident and make the reason for a case’s priority visible to the team.

Within those constraints, consider the likely useful resolution, effort and skills required. Keep uncertain estimates visible and allow review of a poor ranking. Track aging cases and protect an escalation path so difficult work is not repeatedly displaced by easier tasks that improve completion counts.

Separate a shared incident update from work that genuinely requires an individual response. Link related contacts to the common event while preserving distinct customer needs. Make clear which requests wait on the same dependency and which a responder can usefully advance now.

## Rehearse communication and the recovery backlog

Atlassian’s incident handbook assigns technical and communication responsibilities separately and calls for planned handovers as teams tire. Reserve capacity for both responsibilities in the rehearsal. Staffing assumptions should identify who can actually take each role and what other work they must stop.

Run a timed tabletop exercise with arrivals continuing during the outage, slower handling under degraded tools and a second shift taking over. Observe unanswered urgent cases, repeated investigation, missed updates and the work remaining after technical restoration. Use the exercise to revise the plan, without presenting simulated capacity as demonstrated production performance.

Agree how customers will hear the current impact, unresolved uncertainty and next update time. Continue tracking individual workarounds and validations until the promised customer operation is usable again. Close the exercise with specific changes to staffing, tooling or service scope and a subsequent check that those changes work.

## Related reading

- [Atlassian: April 2022 outage review](https://www.atlassian.com/blog/how-we-build/post-incident-review-april-2022-outage)
- [Atlassian: Impact, urgency and priority](https://support.atlassian.com/jira-service-management-cloud/docs/how-impact-and-urgency-are-used-to-calculate-priority/)
- [Atlassian: Incident response roles and communication](https://www.atlassian.com/incident-management/handbook/incident-response)
- [Measuring capacity behind routine service delivery](https://evergences.com/insights/service-scalability/)
- [Examining reliability and restoration evidence](https://evergences.com/insights/software-reliability-recovery-evidence/)

Source: https://evergences.com/insights/correlated-service-surge-capacity/