Incident response plan: roles, escalation, exercises

What an incident response plan should set down about roles and escalation, and why the plan has to be tested in a tabletop exercise.

Topic
IT security
Author
ARLITECH expert team
Published

In the middle of an incident there is no time to argue about who is allowed to shut down a server, or whom in management to call at dawn on a Sunday. These questions have to be settled and written down while things are calm. That is exactly what an incident response plan is for: it records the decisions before anyone needs them.

The plan describes how the organization works when something goes wrong: who leads the response, and when a decision has to be escalated. It also covers which channel the participants use to stay in touch. The technical steps (preserving logs, isolation, recovery) sit alongside it in separate playbooks. The concrete tasks of the first hours are covered in our article on the first 24 hours; this piece sets out the framework those tasks depend on.

What an incident response plan contains

Nobody reads a sixty-page document during an incident, so a usable plan is short, and it pays to split it in two: a slim core document, with appendices of one or two pages per topic alongside it. The core covers the following:

  • Scope and definitions. What the organization treats as a security event and what as an incident, and which systems and sites the plan applies to. If there is an OT environment, name it separately, because the trade-offs are different there.
  • Severity classification and the escalation rules tied to it.
  • Roles with names, deputies, and contact details, and what each role is allowed to decide.
  • Communication rules: internal channels, a backup channel, and how external communication is handled.
  • External contacts: contact details for the incident response partner, insurer, legal counsel, authorities, and key suppliers.
  • Statutory and contractual notifications: who has to be notified in which situation, and who is responsible for doing it.
  • Closure and lessons learned: who writes the post-incident report, and how the changes make their way back into the plan.

The appendices typically hold the contact list, the notification templates, the evidence preservation instructions, and short step lists for common scenarios (ransomware, a compromised account, data leakage). The plan is only usable if it also exists on paper or somewhere independent of the corporate network. If the file server has been encrypted, a plan stored on it cannot be opened either.

Severity levels and escalation rules

The severity level decides how many people get involved and how fast. If every alert goes up to management, managers get used to them over time and start ignoring them. If nothing goes up, a serious case can sit in the on-call queue for days. Three or four levels are enough for most organizations; what matters is that the classification criteria are written down in advance.

LevelTypical situationWho is notified
LowA single, contained event, for example an infected workstation that endpoint protection has isolatedOn-call operator, during normal working hours
MediumAffects several machines or a privileged account, but business operations have not stoppedThe incident lead, the head of IT
HighOutage of a critical system, a suspected data leak, or signs of domain-level compromiseThe designated member of senior management, the legal and communications leads
CriticalOrganization-wide outage or a process safety riskExecutive management, crisis team, external partners

The criteria should be measurable: the range of affected systems, the type of data affected, the impact on the business process, and whether there are signs of an active attacker presence. Working from a sense that something “seems serious enough”, two people on call would classify the same event differently.

Statutory notification obligations have to be checked regardless of level. A personal data breach under Article 33 of the GDPR, or a significant incident at an organization within the scope of NIS2 under Article 23 of the directive, may already require notification at the high level.

Easy to raise, lowered only with a reason

A proven rule is to pick the higher level when in doubt, and to let anyone working on the event raise it. Only the incident lead can lower it, with the reason recorded in writing. This matters because information is scarce in the first hours, and an alert that looks harmless can hide an extensive intrusion.

Escalation also needs a time limit. The plan sets how many minutes the person called has to acknowledge the notification, and who is next in the chain if they do not respond. Weekends and holidays need their own cover arrangements, because an incident often starts, or comes to light, outside working hours.

Roles and decision rights

In a small organization one person may hold several roles, and in a larger event several people stand behind a single role, so the plan defines roles and assigns names to them. What cannot be left out: every role has a deputy, and the contact details of both are in the plan.

  • The incident lead coordinates, sets priorities, and decides the order of the technical steps. They rarely touch a keyboard; their job is to keep the big picture.
  • Technical lead: directs the investigation and containment, working with the system administrators and external experts.
  • Business decision-maker, the designated member of senior management. This person owns the decisions that carry a business cost: shutting down a service, disconnecting a site, approving the notification to the authorities.
  • Communications lead: internal updates, clients and partners, the press.
  • Legal and data protection lead: notification obligations, contractual notifications, legal handling of evidence.
  • Scribe: records every observation, decision, and action with a timestamp. Without one, the post-incident report and the regulatory notification get written from memory.

It is worth writing the decision rights down in a table as well, one decision per row. Who can disconnect a network segment? Who can disable an executive's account? Who can bring in outside experts, and who decides whether to file a police report? If this is not written down, the decision usually lands with whoever happens to be available, or with nobody.

In OT environments, plant management decides on steps that affect the production process (isolating a segment, stopping a controller, switching to manual operation), on the recommendation of IT incident response. The plan should name who that is at the plant and who stands in on the night shift.

Communication when email cannot be trusted either

In a compromised environment, corporate email, the chat app, and video calls may all be in the attacker's view. An attacker with domain administrator rights can read email too, and watch the plans to lock them out take shape. The plan therefore has to define an out-of-band channel in advance, one that does not depend on corporate identity management or the corporate network.

What the backup channel can be

  • Phone calls and SMS to known numbers listed in the plan.
  • A separate group in a messaging app that is not used with corporate accounts, which the participants have already joined.
  • A designated conference call number or a physical location where the crisis team meets.

The backup channel is not something to open during an incident. If the group only gets created then, people suddenly find they do not have everyone's phone number, and nobody can check that it really was a colleague who joined.

Verifying who is calling

An incident is fertile ground for social engineering: people are rushing and everyone wants to help. A “quick, reset my password, I'm locked out” request may just as well come from an attacker. The plan should include a callback rule: a sensitive request is carried out only after the requester has been called back on the number in the contact list.

One voice to the outside

Communication with clients, partners, and the press stays in one pair of hands, using message drafts approved in advance. Staff know whom to refer callers to when questions come in. The plan also sets what staff may post on social media while the incident is ongoing.

Tabletop exercise: testing the plan

Until someone has tried a plan out, there is no way to know whether the contact details and decision paths in it actually work. In a tabletop exercise, participants play through a scenario in a meeting room without touching any live system. The NIST SP 800-84 guide describes it as one of the basic methods for testing plans.

How it is structured

  1. Goal and scenario. A realistic situation tailored to the organization: ransomware on a Friday evening, the takeover of an executive's mailbox, or an incident at a key supplier that also affects your own systems.
  2. Participants. Everyone who has a role in the plan, together with the designated member of senior management. Without management in the room, the business decisions never get tested.
  3. Injects. The facilitator feeds in new information stage by stage: “the backup server is unreachable too”, “a journalist is on the phone”, “the insurer is asking for the first report”. After each inject, the group states what it will do and who decides.
  4. Observer. Someone who takes no part in the decisions and only takes notes: where the group got stuck, what was missing from the plan, who did not know what their job was.
  5. After-action review. Right after the exercise, while it is still fresh. The result is a short list of fixes, each with an owner and a deadline.

The first exercises typically surface simple gaps, such as an outdated phone number or a decision nobody knows the authority for. They also often reveal that the actual time to restore from backup falls far short of what the business expects (RTO). In a meeting room these can be fixed calmly; during a live incident they can no longer be. At least one exercise a year, plus another after every major organizational or system change, helps keep the plan from going out of date.

NIST SP 800-61 as a framework

There is no need to invent the structure of the plan from scratch. SP 800-61, published by the US National Institute of Standards and Technology (NIST), is a widely cited, publicly available incident handling guide. Its second revision, from 2012, describes the process in four phases:

  1. preparation,
  2. detection and analysis,
  3. containment, eradication, and recovery,
  4. post-incident activity, including lessons learned.

The third revision, published in April 2025, aligned this model with the functions of the NIST Cybersecurity Framework 2.0. Incident response thus becomes part of overall risk management: Govern, Identify, and Protect provide the preparation, while Detect, Respond, and Recover cover handling the incident. In practice, this means it is management that has to hold people to account for maintaining the plan, assigning the roles, and feeding the lessons back in; IT cannot guarantee that on its own.

For European organizations, the NIS2 Directive (EU 2022/2555) explicitly lists incident handling and business continuity among the risk management measures. For industrial organizations, the IEC 62443 series of standards also addresses incident handling. The NIST framework can serve as a starting point for both: the four-phase structure can be adopted as is and filled in with your own roles and systems.

Related service: Incident response retainer →

Contact

Ready to secure your critical systems?

Our experts will assess where your security stands and build a roadmap that fits your organization.

ARLITECH's office in Gödöllő

Visit us

You are welcome to visit our Gödöllő office in person.

Address
Centrum Office Building, 2100 Gödöllő, Dózsa György út 13, 3rd floor, office 309
Office hours
Mon–Fri, 9 a.m. to 5 p.m.
Get directions (opens in a new tab)