Prepare
Incident command depends on trained people and ready paperwork, and both have to exist before the incident does. The technical preparation lives in the PIVTR-D preparation playbook. This page covers what the incident organization needs in place before anyone stands it up.
Running the program covers the standing parts: one severity scale, the commander rotation, update schedules, the review board and metrics.
People
Name at least two trained people for every seat you expect to fill, so the incident survives its first shift change, and three for incidents that run past three days. The staffing math below shows the numbers.
Training covers the role, the forms and the meetings. Weight it toward the Incident Commander, the one role nobody has day to day.
Practice in the seats. In Google's Wheel of Misfortune exercises, less experienced on-callers role-play past incidents to build skill where mistakes cost nothing.
Run tabletops the same way: replay real incidents, rotate who takes command, communications and operations, and time the first objectives and the first executive update. Four tabletops are ready to run: ransomware, a data center flood, a regional outage at a hosting partner, and a denial of service extortion that hides a customer account takeover. Each page has a Run this exercise button. It runs the exercise clock, shows each inject on a player screen for the room, records what the facilitator saw and what the room decided, and writes an after-action draft.
Staffing math
A seat covered around the clock on 12-hour shifts needs two people a day. Past the third or fourth day people need days off, and each seat needs three. Count the command and section chief seats for the type you expect, then add the group supervisors and their teams.
| Type | Command and section seats | People, first 3 days | People, beyond 3 days |
|---|---|---|---|
| 3 | 7: Incident Commander, Operations, Planning, Public Information, Liaison, Legal advisor, Safety | 14 | 21 |
| 2 | 10: all Command and General Staff, with Intelligence/Investigations separate | 20 | 30 |
With three people per seat, A, B and C, this rotation gives each person two shifts in every three days and at least 24 hours between shifts.
| Day | Day shift, 07:00 to 19:00 | Night shift, 19:00 to 07:00 |
|---|---|---|
| 1 | A | B |
| 2 | C | A |
| 3 | B | C |
| 4 | A | B |
| 5 | C | A |
| 6 | B | C |
| 7 | A | B |
Offset the Incident Commander's handover from the Operations Section Chief's by an hour or two, so someone in command always saw the end of the last shift. Agree before the incident how many days in a row anyone may work.
Authority
Write down what the Incident Commander can do without asking: isolate systems, suspend accounts, take services offline, activate retainers, spend up to a set amount, engage law enforcement. Put the limits in business terms.
Name who approves anything above them, and their backups. PIVTR-D has a starting authority matrix.
NIMS calls the executive who owns the incident the Agency Administrator, and the Incident Commander's authority comes from them through a delegation of authority. Write a standing delegation now, signed by the executive, so the on-call Incident Commander has it the moment they take command. For a large incident, or when an outside team takes over, issue one for that incident using the delegation of authority form.
Detection that reaches someone
The incident organization starts when someone is paged. Google's Incident Management Guide asks four things of an alert: it arrives in time, it covers what users depend on, it measures what users experience, and someone can act on it.
For security, read that as detections that fire early enough to contain and cover the systems and identities that matter most. They carry enough context to size the incident and route to someone who can take command. The PIVTR-D identification playbook covers the detection side.
Tools that open the incident for you
Automate the first 10 minutes so responders spend them thinking. One action should open the incident record (PIVTR-D T1) and a blank ICS 201, and create the command bridge and the out-of-band room. It should also page the Incident Commander and on-call, start a timestamped activity log, and post the first internal notice.
Google's guide also recommends automating common tasks and the first read of impact, so on-callers can work the problem.
Contracts and outside help
- Incident response retainer: how to activate it, response times, who can sign.
- Cyber insurance: notice deadline, panel vendors, what the insurer must approve.
- Counsel: who to call, and when counsel directs the investigation.
- Managed providers: which containment actions they take on their own authority.
- Law enforcement: named contacts, agreed before you need them.
Ready to run an incident
| # | Item |
|---|---|
| 1 | Two trained people named for Incident Commander, operations, communications and planning |
| 2 | Incident Commander authority written, with limits and approvers |
| 3 | Incident types mapped to your severity scale and published |
| 4 | Forms and the incident record template ready to copy |
| 5 | Out-of-band channel and accounts provisioned outside the main identity provider |
| 6 | Contact roster for internal and outside parties, kept offline and tested |
| 7 | Retainer, insurer and counsel details in the roster |
| 8 | One action that opens a new incident |
| 9 | Shift length and relief rules agreed |
| 10 | A tabletop in the last quarter, with the seats rotated |