A cybersecurity incident report is the formal record of what happened, what it affected, what it could still affect, and what the organization is doing about it. It is written for two audiences at once: executives who will make decisions from it today, and auditors, regulators, or lawyers who may read it a year from now. That is why its first job is to separate what is confirmed from what is only possible.

This article lays out the sections that matter, in the order they are read, and shows how each one is filled using a worked example: a university student portal with two critical vulnerabilities found by an external penetration test.

Who the report is for

NIST SP 800-61 Rev. 3, published in April 2025, treats incident response as part of the organization’s wider cybersecurity risk management under the NIST Cybersecurity Framework 2.0. The report is where response meets governance. Leadership cannot prioritize, fund, or disclose without it. External parties may have a claim on it too: U.S. federal agencies, for example, must notify CISA within one hour of identifying an incident under the Federal Incident Notification Guidelines, and the fields CISA asks for are a useful model for anyone.

Write for the executive first. Technical detail belongs in the report, but after the facts the executive needs, not before.

The sections, in order

Sections of a cybersecurity incident report and what each must answer
SectionWhat it answersExample from the student portal case
Report headerDate of the report, author, approver, and who is notifiedDated February 16, 2026. Written by the Information Security Manager, approved by the CISO, notified to the President, CISO, and security team.
SummaryWhat happened, in three or four sentences, with confirmed facts onlyAn external penetration test found injection and broken access control in the student portal. The internal team validated both. No exploitation confirmed; investigation ongoing.
Incident descriptionDiscovery date and time, discovering and validating parties, systems and users affected, how long the issue persisted, resolution statusDiscovered February 10, 2026 at 14:15 EST. Three application servers and one database. About 18,000 users. Present for about six months since the last portal update. Temporary mitigations in place; full remediation due by March 16.
Technical detailWhat the vulnerability or attack allowed, in plain languageInjection allowed manipulation of backend database queries. Broken access control allowed privilege escalation and modification of other user accounts.
Impact and potential impactConfirmed harm, then possible harm by categorySeven categories: data compromise, system damage, financial loss, public relations, regulation (FERPA), security policy, and unknown risks.
Information sensitivityWhat classes of data were exposed, including an explicit unknownRestricted or confidential data (student PII, grades, enrollment, account data) and a stated unknown category for systems not yet identified.
Remediation planActions, owners, and timeframes, fastest risk reduction firstParameterized queries in two weeks; code review and monitoring in one month; RBAC in two to three months; least privilege in six months; testing before all future updates.

Separate confirmed, possible, and unknown

The most common failure in incident reports is blending the three. A report that says “student data was exposed” when the finding is “a vulnerability that could expose student data existed for six months with no confirmed exploitation” will drive the wrong disclosure decision and will not survive a regulator’s questions.

The fix is structural. State the confirmed facts in the summary and incident description. Put potential harm in its own section, by category, so that the reader sees regulatory and reputational exposure next to the technical risk but not mixed with it. Then record what is unknown as its own line item. In the example, “unknown risks: undiscovered exploitation or lateral movement to other internal systems” and an “unknown” sensitivity class keep the report honest about the limits of the investigation. An honest unknown is also a work item: it tells the reader what the investigation still has to close.

Impact categories that executives recognize

CISA’s notification guidelines ask reporters to classify functional impact, information impact, and recoverability. Those three are a good minimum. For an executive audience, widen the list so that each decision-maker sees their own exposure:

  • Data compromise: which records and which people.
  • System damage: corruption, outage, or loss of service, and for how long.
  • Financial loss: remediation cost, penalties, delayed operations, extra staffing.
  • Public relations: reputational effect, particularly when the organization’s own brand depends on trust.
  • Regulation: which laws apply. In the university example that is FERPA, the U.S. law protecting education records. In healthcare it would be HIPAA; in payments, PCI DSS.
  • Security policy: which internal standards were not enforced, such as secure coding or access control.
  • Unknown risks: what the investigation has not yet ruled out.

The remediation plan: sequence by speed of risk reduction

A remediation plan is not a list of everything that should improve. It is a sequence. Put the actions that remove the most risk soonest at the top and give structural work the time it needs. In the example, parameterized queries close the injection path in two weeks, while role-based access control and a least-privilege model get two to six months. Each action carries a timeframe. Because the vulnerabilities dated from an application update, the plan ends with a standing control: security testing before every future release.

Two details make a plan credible. Every item has an owner, even if the owner is a role rather than a name. And the plan distinguishes temporary mitigation already in place from the permanent fix still to come, with a date for the latter.

A checklist before you send it

  • Could a reader say, from the summary alone, whether data was confirmed stolen? If not, rewrite the summary.
  • Does every number have a source: a log, a test result, a system inventory?
  • Is “unknown” written down somewhere, rather than left implicit?
  • Does the regulation section name the law, not just “compliance”?
  • Does each remediation item have a timeframe and an owner?
  • Is the report dated, authored, and approved, so it can stand as a record?

Frequently asked questions

What should a cybersecurity incident report include?
A dated header with author and approver, a factual summary, the incident description with discovery time and affected systems and users, plain-language technical detail, confirmed and potential impact by category, the sensitivity of the data involved, an explicit list of unknowns, and a remediation plan with owners and timeframes.
Who writes the incident report?
Usually the incident lead or the information security manager, with approval from the CISO or equivalent. The approver’s name on the report is part of what makes it a formal record.
How soon should an incident report be written?
An initial report should go out as soon as the facts are validated, with the investigation marked as ongoing. Federal agencies in the U.S. must notify CISA within one hour of identifying an incident. Update the report as facts change rather than waiting for the investigation to end.
Is an incident report the same as a breach notification?
No. The incident report is the internal record. A breach notification is a legal communication to affected individuals or regulators, triggered by specific laws such as HIPAA, state breach laws, or sector rules. The report is the evidence base for deciding whether a notification is required.

Sources