← All articles

What Is Incident Commander? Role, Duties, and Phases

Published · 9 min read

Key takeaways

  • An incident commander owns decisions, not implementation: they declare severity, assign work, approve customer communication, and call the incident resolved.
  • The role comes from FEMA's Incident Command System and was popularized for engineering by Google's SRE incident management practices.
  • IC duties shift by phase: set severity and escalation policy before an incident, run triage and communication during it, and own the postmortem after.
  • The most common failure mode is hero mode: the moment the IC starts debugging, nobody is coordinating and the fix usually takes longer.
  • A solo founder can be the IC by keeping customer-facing duties and dropping coordination ceremony, using tooling to substitute for missing roles.

An incident commander is the single person who owns decisions during an outage: they triage the problem, assign work, approve customer communication, and call the incident resolved. On a 1-10 person SaaS team, that person is usually you, the founder, juggling fixes and status page updates at the same time.

What is incident commander?

An incident commander is the single person accountable for coordinating a technical incident from detection through resolution. They decide severity, assign tasks, approve customer communications, and declare the outage over. In a SaaS context, they are not necessarily the person fixing the bug; they are the person running the response while engineers debug and the status page gets updated.

Incident commander definition in one paragraph

The incident commander owns decisions, not implementation. They set or confirm severity, delegate work, keep the incident timeline, gate every public update, and call the all-clear. If nobody holds this role, tiny teams default to everyone talking at once while customers watch silence on the status page. The full duty breakdown lives in incident commander roles and responsibilities.

Where the role comes from (fire service and ICS origins)

The role is borrowed from the Incident Command System, a standardized command structure developed for US emergency response so multiple agencies coordinate under one leader. FEMA's ICS documentation defines the commander as the individual responsible for overall management of the incident, a structure SRE teams at Google popularized for engineering through their published incident management practices.

What are the core responsibilities of an incident commander?

The three jobs: decisions, communication, coordination

An incident commander has three jobs that hold across any incident, any team size, any stack.

Decisions. The IC owns the call: is this an incident, what severity, when to roll back, when to escalate, when to declare it over. One person decides so five people don't debate.

Communication. The IC posts updates to the status page on a fixed cadence (every 15-30 minutes during an active incident), keeps stakeholders informed, and decides what customers are told. Writing "investigating" at 3 AM is part of the job, not an afterthought.

Coordination. The IC assigns work: who debugs, who tests the fix, who watches dashboards. They track what's been tried so nobody repeats it.

What the IC explicitly does NOT do

The IC should not be hands-on-keyboard fixing the bug. Debugging and commanding use different mental modes, and trying to do both means updates go stale while you're deep in a stack trace. If your team is one person, accept that updates will lag and use tooling to compensate: StatusPulse lets you post manual incident updates from one place so communication doesn't depend on you stopping mid-fix.

What does an incident commander do during each phase of an incident?

The incident commander's duties shift with each phase of the lifecycle. Before an incident, they set severity levels and escalation policy; during it, they run triage, mitigation, and communication; after it, they own the postmortem. The Google SRE book's incident management chapter describes the IC as the single decision-maker who delegates everything else, and the phase mapping below follows that split.

Duties mapped to the lifecycle: a phase-by-phase table

PhaseIC dutyDelegate to
DetectionConfirm the alert is real, not a false positiveMonitoring
TriageDeclare severity, assign roles, start the timelineNobody, yet
MitigationPick the fix, coordinate the fixerComms lead
CommunicationApprove status page updates at a set cadenceComms lead
ResolutionConfirm recovery, close the incidentOn-call
PostmortemRun the review, assign follow-up actionsAnyone involved

Detection and triage: declaring the incident

The IC's first real job is deciding whether an outage is actually happening. Homepage checks passing while your API fails is the classic miss, which is why protocol-level probes (TCP, DNS, SMTP) catch what URL pings don't. Once confirmed, the IC declares severity, names the roles, and starts the timeline. StatusPulse logs each probe result so the declaration is backed by data, and the incident CSV export gives you the raw timeline for the postmortem.

Mitigation and communication: running the incident

Here the IC stops fixing and starts coordinating. They choose between rollback and forward-fix, set a customer-update cadence (every 30 minutes is a common rule), and approve what goes on the status page. Writing updates while debugging is how 3 AM incidents go sideways.

Resolution and handoff

The IC confirms the fix with their own probes, posts the all-clear, then hands off to the postmortem owner. Honest limitation: a solo founder is usually both IC and fixer, so the delegation column collapses to one name.

Incident commander role vs other incident roles: who does what?

The incident commander owns the incident: they declare it, set severity, assign work, and decide when it's resolved. The comms lead writes status page updates and customer emails. The ops lead (or subject-matter expert) actually fixes the thing. The scribe keeps a timeline of actions and decisions for the postmortem.

Role split in a 3+ person team

With three or more people, split it explicitly: one IC, one ops lead, one comms/scribe combined. The IC's only job is coordination. If you have a status page workflow already, the comms lead runs it while the IC tracks the probe failures and assigns next steps. Pair this with a comms toolchain; our guide to incident communication tools covers the 2026 stack: https://blog.statuspulse.ai/post/incident-communication-tools-2026-reliability-guide

Why the IC should not be the one typing the fix

The moment your IC starts debugging, nobody is coordinating. Severity calls go unmade, customers get no update, and the fix takes longer. In a two-person team this is hard to avoid, but even then, decide up front who is IC and who is hands-on-keyboard, and swap only at a clear handoff point.

Can a solo founder be the incident commander?

Yes. Atlassian's incident management docs describe the IC as the person coordinating response and communication (Atlassian), and nothing in that role requires a second human. What it requires is decisions made fast: declare the incident, post to the status page, fix or mitigate, communicate again. A solo founder can do all four, but only if the ceremony is stripped out first.

The solo IC checklist: what to keep and what to drop

Keep the parts that touch customers: declaring incidents, updating the status page, and the post-incident writeup. Drop the parts that exist for coordination at scale: role handoffs, scribe assignments, and formal war-room cadence. When you are the IC, the comms lead, and the fixer, every minute spent on process is a minute not spent on the fix. Your MTTA improves when there is nobody to wait for, but your MTTR usually gets worse, because debugging and customer updates compete for the same pair of hands. That trade-off is why solo incident response lives or dies on preparation, and why we track it with the metrics in MTTA, MTTR, and the rest.

Using tooling to substitute for missing roles

Automate the scribe: alerts, probe status changes, and timestamps logged automatically give you the incident timeline a scribe would write. Automate the comms lead: a status page that updates in one click and escalates to your phone covers the "tell customers" half of the job. Automate detection: protocol-level probes on your critical paths catch what a homepage check misses, so the 3 AM call comes from a real signal, not a false alarm.

Where StatusPulse's own incident tooling falls short for solo ICs

Honestly: we give you probes, alerts, and a status page, but we do not give you a second brain. There is no on-call rotation to share the 3 AM wake-up, no runbook automation, and no built-in post-mortem workflow, so the writeup discipline is on you. If you need PagerDuty-style escalation trees or role-based incident workflows, StatusPulse is not that tool today.

How do you train or appoint an incident commander?

Pick the person who can make a call without waiting for consensus. In a 1-10 person team that is usually the founder or the senior engineer on call, and the appointment should be written down: name, backup, and escalation order. An IC who is also the person fixing the bug is fine, as long as they hand off the fix when communication starts slipping.

Rotation and on-call schedules in tiny teams

With two or three people, rotation is simple: alternate weeks, and the person off-call is the backup IC. Write the order in your status page tool or PagerDuty schedule so nobody negotiates it at 3 AM. If you are a solo founder, your rotation is you plus a runbook that tells a half-asleep version of you what to do.

A 30-minute IC drill you can run this week

Pick a real service, kill it in staging, and set a timer. The IC must post to the status page within 5 minutes, post an update within 15, and write a 3-line incident summary after. Debrief for 10 minutes: what was slow, what was unknown? Run it monthly and the first real outage becomes a rerun, not a debut.

What makes a bad incident commander?

Hero mode: the IC who fixes instead of coordinates

The most common failure is the IC who starts debugging. The moment your incident commander is also the person typing commands, nobody is tracking scope, decisions, or comms, and the fix usually takes longer. Coordination is the job. Google's SRE book treats the IC as the role that explicitly does not touch the broken system during major incidents (Incident Management chapter, Google SRE Book).

Silent incidents: the IC who forgets the status page

The second failure mode is going dark. Customers forgive downtime; they punish silence, and updating your status page during the incident is a core IC duty, not an afterthought. If you're solo, post a one-line acknowledgement first, investigate second, then update every 30 minutes until resolved.

Incident commander duties checklist (copy-paste template)

Pre-incident

Write your severity levels down and share them. Confirm who is on-call and how they get paged. Test the alert path monthly, including the PagerDuty or SMS fallback.

In-incident checklist

  1. Declare the incident and set severity within 5 minutes of the first alert.
  2. Start a timestamped log in one shared channel or doc.
  3. Post the first status page update within 15 minutes, even if it says "investigating."
  4. Assign one person to comms so the fixer can fix.
  5. Post updates every 30 minutes until resolved.
  6. Confirm resolution on the status page and in the log.

Post-incident checklist

Within 48 hours: write the timeline, name the root cause, list action items with owners, and review whether monitoring caught it or a customer did.

FAQ

Does an incident commander need to be the most senior engineer?

No. The role rewards fast decision-making, not deep technical knowledge. Pick the person who can declare severity and assign work without waiting for consensus, which in a small team is usually the founder or the on-call senior engineer. The IC coordinates while someone else debugs, so expertise matters less than decisiveness and clear communication.

Is incident commander the same as incident manager?

The terms overlap heavily, but they differ in scope. An incident commander runs a single live incident: they triage, delegate, approve customer updates, and declare resolution. An incident manager usually owns the broader program: on-call schedules, severity definitions, postmortem follow-through, and process improvements across many incidents. In tiny teams, one person often holds both jobs anyway.

How many incident commanders should a small team have?

One primary and one backup, with the rotation written down. With two or three people, alternate weeks and make the off-call person the backup IC, so nobody negotiates roles at 3 AM. Solo founders are the only IC by default, which works if a runbook tells a half-asleep version of you what to do.

What does an incident commander do when the incident is caused by a third-party vendor?

The job is unchanged: confirm the outage is real, declare severity, and post a status page update naming the vendor and what customers should expect. The IC keeps pressing the vendor for updates on a fixed cadence and communicates each one, rather than staying silent while waiting. Resolution still gets confirmed and announced on your own status page.

Monitor it, then tell your customers

10 probes, 3 status pages, forever. No credit card.