Nobody reads a procedure during an incident. The team either applies decisions made in advance in the procedure or improvises. If, when encrypted files are seen on the file server at 3 a.m., it isn’t clear whom the on-call person will call, whether to shut the server down and who will assess the KVKK notification, then there’s no procedure. That’s why an incident management procedure isn’t a theory text; it’s a document that makes clear who reports what and when, how the incident is prioritized, to whom the decision escalates and which record is kept. Below we explain how to write, section by section, a procedure that meets the ISO/IEC 27001:2022 Annex A controls, and how to tie KVKK and customer notification obligations into the process.
Controls the procedure must meet
In Annex A, incident management is a chain of controls that follow one another. Each section of the procedure must correspond to at least one of these controls:
| Check | Topic | Where it appears in the procedure |
|---|---|---|
| 5.24 | Incident management planning and preparation | Roles, team, tools, contact list |
| 6.8 | Information security event reporting | Reporting channels |
| 5.25 | Assessment and decision on events | Classification and prioritization |
| 5.26 | Response to incidents | Escalation and response steps |
| 5.28 | Collection of evidence | Evidence collection and retention |
| 5.27 | Learning from incidents | Post-incident review |
| 5.5, 5.6 | Contact with authorities and special interest groups | External notification |
The procedure doesn’t have to contain all the technical response steps. Separate response playbooks are prepared for scenarios such as ransomware, account takeover or data leakage, and the procedure refers to them. The procedure stays short. The playbooks are detailed. For procedure structure and writing rules, see the guide to writing procedures and work instructions.
Definitions of event, weakness and incident
If definitions are vague, the same case is recorded as an “incident” by one team and a “fault” by another; at year end the statistics say nothing. Write this distinction clearly at the start of the procedure:
- Information security event: An observed situation that may be relevant to security; for example, a rise in failed sign-in attempts, or a report of a suspicious email.
- Information security incident: An assessed event or series of events that compromises, or is highly likely to compromise, the confidentiality, integrity or availability of information.
- Weakness: A gap not yet exploited; for example, noticing that a shared folder is open to all users.
- Personal data breach: Personal data being obtained by others through unlawful means; it gives rise to a separate notification obligation under Article 12 of KVKK (Law No. 6698).
Not every event is an incident, and the assessment is a decision step; the procedure must show this clearly. The decision and its justification are recorded. The auditor particularly wants to see the justification for events deemed “not an incident”; records with an empty justification field are the first finding.
Reporting channels and the initial record
Annex A 6.8 requires personnel to report observed or suspected events through appropriate channels in a timely manner. Channels must flow to a single point and be simple for the user:
- The service desk system, with a “security incident” category in it.
- A phone number and on-call rota for outside working hours.
- A report-suspicious-message button in the email client.
- Automatic alerts from monitoring systems (SIEM, EDR, firewall).
The user instruction shouldn’t exceed five steps: write down what you saw, take a screenshot, don’t turn off the device, don’t intervene yourself, report it. Leave the decision to disconnect the device from the network to IT, not the user. A mistake you may come across is the number on the on-call rota belonging to an employee who has left; test the rota by calling it every three months. Fields to collect in the initial record: time of report, reporter, affected system or asset, observed situation and the person opening the record. Writing the affected asset with its inventory ID speeds up impact assessment.
Classification and prioritization table
Prioritization decides what the team responds to and how fast. Write the table with criteria concrete enough that no debate is needed during an incident. The structure below is an example; set times and criteria according to the organization’s business impact analysis.
| Priority | Example criterion | Initial response (example) | To be informed |
|---|---|---|---|
| P1 – Critical | Critical business process stopped, widespread encryption, suspected personal data leak | Immediately, 24/7 | Top management, crisis team, legal |
| P2 – High | Suspected administrator account takeover, malware on a single critical server | Within a few hours, including outside working hours | ISMS manager, IT manager |
| P3 – Medium | Malware on a single user device, phishing click | During working hours | IT team lead |
| P4 – Low | Blocked attack attempt, suspected policy violation | Scheduled | In the monthly report |
Priority is set by two axes: business impact (how many users, which process, which data class) and potential to spread. Malware that starts on a single device becomes P1 within minutes if the same local administrator password is used on all clients. Priority can be raised during an incident; the procedure must say so, and that every change is recorded with its justification.
Escalation and external notification obligations
The escalation section answers “who is informed” by role, with deputies, not by name. The criterion for moving to crisis management for P1 incidents is defined here; how crisis communication is run is the subject of the crisis management and communication plan guide.
External notification obligations aren’t researched during an incident. They sit as a ready table in the annex to the procedure:
- KVKK: Paragraph five of Article 12 of the law requires the data controller to notify the data subject and the Board as soon as possible if personal data is obtained by others through unlawful means. The Personal Data Protection Board’s decision 2019/10 interprets “as soon as possible” as no later than 72 hours from learning of the breach for notification to the Board. The current decision and notification form should be verified on kvkk.gov.tr.
- Customer contracts: Many customer contracts, supplier security specifications or customer requests under TISAX define their own notification time. List these times per contract and assign a responsible role.
- Sector regulations and insurance: Notification requirements in sector regulations the organization is subject to and in the cyber insurance policy also go in the same table.
Evidence collection and the incident record
Annex A 5.28 requires procedures for the identification, collection, acquisition and preservation of evidence. If an incident may end in legal or disciplinary proceedings, the integrity of evidence must be provable. Write these rules into the procedure:
- An affected system isn’t shut down or rebuilt before a decision is made; memory and session information may be lost.
- For every item collected, who took it, when, from where and by what method is recorded (chain of custody).
- A hash of the collected files is taken and written into the record.
- Evidence is kept in an environment with restricted access and limited write permissions.
Evidence that can be lost goes in a well-meant “let’s format it and restore right away” intervention. The server is rebuilt clean, but how the attacker got in is never learned. Built-in tools are enough to export the security log from a Windows server and take its hash:
wevtutil epl Security D:\IR\INC-2026-031\SRV01-Security.evtx
Get-FileHash -Path D:\IR\INC-2026-031\SRV01-Security.evtx -Algorithm SHA256 |
Format-List Algorithm,Hash,Path
On Linux, the same is done with sha256sum file > file.sha256. Check the log size too; a Security log left at its default size overwrites old records quickly on a busy server, and the start of the incident doesn’t stay in the record. Forensic work such as disk and memory imaging needs expertise; the procedure must state which external team is called in that case. Technical detail on which data must be preserved without deletion is in the guide preserving forensic evidence after a cyber incident.
The incident record contains the timeline, decisions taken and their justifications, priority changes, external notifications and a closure summary. The record’s retention period is defined in the records list; for the method, see the article records list and retention periods.
Post-incident review
Annex A 5.27 requires knowledge gained from incidents to be used to strengthen controls. Make a review meeting mandatory for P1 and P2 incidents; for the rest, a periodic combined review may be enough. The review questions the process, not the person, and produces written answers to these questions:
- When did the incident start, when was it noticed, and what determined the gap in between?
- Which step in the procedure didn’t work or was missing?
- Which control could have prevented or detected the incident early?
- Which corrective actions were opened, with which owner and which date?
- Is there an assessment in the risk register that needs updating?
The first question usually gives the most uncomfortable answer. If the time between the start and detection is long, the problem isn’t the response but the monitoring. The outputs become inputs to the corrective action process and management review. For continuity scenarios specific to production environments, the article IATF 16949 cyber attack contingency plan is a complement. To set up the procedure together with the response plan and notification flows, see cyber incident response plan and incident notification processes, and to tie personal data breach obligations to organizational measures, KVKK technical and organizational measures.
Frequently asked questions
Is every information security event an incident?
No. An event is an observation that may be relevant to security; an incident is an event determined, after assessment, to have compromised the confidentiality, integrity or availability of information. The assessment decision and its justification are recorded.
How quickly must a data breach be notified under KVKK?
The law says “as soon as possible”; the Personal Data Protection Board’s decision 2019/10 interprets this as no later than 72 hours from learning of the breach for notification to the Board. The current rule should be verified on kvkk.gov.tr.
Should the incident management procedure include technical response steps?
The procedure defines roles, prioritization, escalation, and evidence and record rules. Technical steps specific to scenarios such as ransomware or account takeover are kept in separate response playbooks, and the procedure refers to them.
For which incidents should a post-incident review be held?
Hold a separate review meeting for critical and high-priority incidents. Low-priority incidents can be reviewed together periodically; recurring patterns come out in that review.