Prioritisation

Part of Asynchronous work habits

Separating an urgent alert from an ordinary update

If every message is marked urgent, people cannot tell which one needs immediate action. Define urgency by impact and response time, then route it differently from routine status.

Desk Workflow · · 3 min read

Treat a message as urgent when its current impact calls for immediate human intervention. A deadline or a sender’s anxiety alone does not make a message an incident; if no urgent action is needed, use the ordinary update route.

At Atlassian, SEV 1 means critical, very high impact, such as a customer-facing service being down for all customers, customer data loss or a security breach. SEV 2 means significant impact, such as a service being down for some customers or a core function being significantly affected. SEV 3 means low impact, such as a minor inconvenience with a workaround.

Atlassian routes SEV 1 and SEV 2 incidents to on-call professionals for an immediate fix, regardless of the time of day. SEV 3 incidents can be handled during daytime or working hours.

Set an acknowledgement and response target so the recipient knows when an alert has been missed. One example critical-alert policy sets acknowledgement within 5 minutes, a first response within 15 minutes and status updates every 30 minutes until resolution.

Pair the target with a fallback. PagerDuty high-urgency notification rules escalate if an alert is not acknowledged; set the escalation rule for your team rather than assuming a particular delay.

Severity labels can guide routing: critical and error map to high urgency in PagerDuty, while warning and info map to low urgency. Info is for normal operational messages requiring no action; warning may indicate that action is needed to prevent an error.

Send ordinary updates to the normal project record and response window. Jira and Confluence keep work and decisions visible; a severity-based setup can route warnings to Slack and a queue, and informational messages to a log and dashboard.

For example, a SEV 1 alert about Jira Cloud being down for all customers calls for an immediate on-call response and names the incident owner and next update time. By contrast, an Info update reports a routine operational message requiring no action and goes to a log and dashboard.

A decision request can have a deadline without being an incident. Route it as a request with a due date unless its impact meets the urgent-alert trigger.

After an incident, put the final decision and timeline in the durable record. Review missed alerts and false alarms, then adjust the trigger or escalation rule where needed.

Urgent Alert vs Ordinary Update: Key Differences

  • Trigger for UrgencyImmediate human intervention required due to high impact (e.g., service outage, data loss)
  • Routing PathOn-call professionals via PagerDuty or incident management tools (e.g., Jira Service Management)
  • Severity LevelSEV 1 (critical) or SEV 2 (significant impact)
  • Action Required?Yes – immediate fix needed
  • Example ScenarioJira Cloud down for all customers
  • Trigger for Ordinary UpdateNo immediate action needed; routine status or informational message
  • Response Time RequiredDuring working hours; no strict time-bound response
  • Routing PathNormal project record (e.g., Jira, Confluence), log, or dashboard
  • Severity LevelSEV 3 (low impact), warning, or info
  • Action Required?No – may include workaround or future planning
  • Example ScenarioRoutine operational message with no impact

Atlassian Severity Levels: Impact and Response

  1. SEV 1 – CriticalHigh impact: customer-facing service down for all, data loss, security breach
  2. SEV 2 – SignificantModerate impact: service down for some users, core function severely affected
  3. SEV 3 – Low ImpactMinor inconvenience with a workaround available

Handling an Incident: From Alert to Resolution

  1. Identify SeverityAssess impact: is immediate action required? (SEV 1 or SEV 2?)
  2. Acknowledge Within 5 MinutesEnsure acknowledgment triggers alert tracking and accountability
  3. First Response in 15 MinutesProvide initial diagnosis and next steps
  4. Status Updates Every 30 MinutesKeep stakeholders informed until resolution
  5. Final Decision RecordedDocument outcome, timeline, and lessons learned in durable record (e.g., Jira, Confluence)
  6. Review & ImproveEvaluate missed alerts or false alarms; refine triggers or escalation rules

More from Prioritisation