If you open a 93-control SoA and can’t answer “which controls alert us during an incident?” in five minutes, your control set is sitting there as a list; it isn’t being managed. One of the most visible differences between ISO/IEC 27002:2022 and the 2013 version is that every control is assigned five attributes. These attributes let you filter the 93 controls from different angles, report to management in different views and build a common language with frameworks such as NIST CSF. This guide explains the attributes for ISMS managers and security architects who manage the control set, and shows with examples how to make them useful in the SoA and the control inventory.
The relationship between ISO 27001 and ISO 27002
ISO/IEC 27001 is a requirements standard; certification is done against it. Its Annex A gives only a short control statement for each control. ISO/IEC 27002 is the guidance standard: with the same numbering, it contains each control’s purpose, implementation guidance, other information and attribute values. Conformity with ISO 27002 isn’t audited in certification; but when the interpretation of a control is being discussed, both organizations and auditors open ISO 27002.
Attributes appear only in ISO 27002 and are entirely informative. No attribute value makes a control mandatory or takes it out of scope. They’re labels. They’re used to view the control set according to your own needs. For the 2022 and 2024 changes in general, see the article ISO 27001:2022 and the 2024 amendment.
The five attributes and their values
In the standard, attribute values are written with a “#” sign; a control can take more than one value for the same attribute.
| Attribute | Values | The question it asks |
|---|---|---|
| Control type | #Preventive, #Detective, #Corrective | Does the control act before, during or after an incident? |
| Information security properties | #Confidentiality, #Integrity, #Availability | Which property does it protect? |
| Cybersecurity concepts | #Identify, #Protect, #Detect, #Respond, #Recover | Which stage of the cybersecurity cycle does it belong to? |
| Operational capabilities | 15 values (e.g. #Governance, #Asset_management, #Identity_and_access_management, #Continuity) | Which operational team or capability is it related to? |
| Security domains | #Governance_and_Ecosystem, #Protection, #Defence, #Resilience | Which high-level security domain does it serve? |
The cybersecurity concepts are based on the five concepts in ISO/IEC TS 27110. The full list of operational capabilities: #Governance, #Asset_management, #Information_protection, #Human_resource_security, #Physical_security, #System_and_network_security, #Application_security, #Secure_configuration, #Identity_and_access_management, #Threat_and_vulnerability_management, #Continuity, #Supplier_relationships_security, #Legal_and_compliance, #Information_security_event_management and #Information_security_assurance.
The values of a few controls show how to read the structure:
| Check | Control type | Cybersecurity concept | Operational capability |
|---|---|---|---|
| 5.7 Threat intelligence | #Preventive #Detective #Corrective | #Identify #Detect #Respond | #Threat_and_vulnerability_management |
| 8.13 Information backup | #Corrective | #Recover | #Continuity |
| 8.15 Logging | #Detective | #Detect | #Information_security_event_management |
| 8.16 Monitoring activities | #Detective #Corrective | #Detect #Respond | #Information_security_event_management |
Examples of filtering controls by attribute
The practical value of attributes shows when you keep the control inventory as a table (Excel, a SharePoint list, a GRC tool) and filter by these columns. Typical questions:
- “Which controls will alert us during an incident, and what state are they in?” List those with control type #Detective together with their implementation status in the SoA. Detection controls declared applicable but with weak evidence show the blind spots in the incident management chain. In the field, 8.15 usually tops this list: logs are collected, but nobody reads them and the retention period is left to whenever the disk fills up.
- “What is our ability to recover after ransomware?” Intersect controls with #Recover and #Respond values with the #Availability property. The small set that results is the controls a business continuity exercise should test.
- “Which controls does the IT infrastructure team own?” Filter by the #System_and_network_security and #Secure_configuration operational capabilities and assign control owners.
- “Is our control set piled up on preventive controls?” Count implemented controls by control type. If the number of detective and corrective controls is low, prioritize the incident management and continuity areas. A system that only prevents is blind at the first breach.
If you keep the control inventory as a CSV, you can split cells that contain more than one value and work with a simple filter. The example below uses the standard Python library to list #Detective controls that aren’t yet fully implemented (column names and rows are for illustration):
control;title;control_type;cyber_concept;status
8.15;Logging;#Detective;#Detect;Partial
8.16;Monitoring activities;#Detective #Corrective;#Detect #Respond;Implemented
8.13;Information backup;#Corrective;#Recover;Implemented
The content above is the controls.csv file; the script below reads it:
import csv
with open("controls.csv", encoding="utf-8") as f:
for r in csv.DictReader(f, delimiter=";"):
types = r["control_type"].split()
if "#Detective" in types and r["status"] != "Implemented":
print(r["control"], r["title"], r["status"])
The same logic can be set up in Excel with a “contains” filter or in a SharePoint list with a multi-select column. The real issue is spelling discipline: choose values from a fixed list instead of free text. If one cell says “#Detective”, another “Detective” and a third “#detective”, the filter silently returns incomplete results and nobody notices.
Attribute views in management reporting
Top management won’t read a 93-control SoA. An attribute-based summary of the same data, however, makes decisions easier. Views that can be used for management review or quarterly reports:
| View | Rows | Audience |
|---|---|---|
| Security domain summary | Four security domains × implementation status | Top management |
| Cyber cycle view | Identify–Recover × number of open actions | IT management, CISO |
| Team view | Operational capability × control owner | Team leaders |
| CIA view | Confidentiality, integrity, availability × number of risks | Risk owners |
In these views a number on its own says nothing. It must be traceable which controls sit behind each cell and which evidence the implementation status is based on. Otherwise the view turns into a colorful dashboard that can’t be defended in an audit.
Using attributes to map to other frameworks
The cybersecurity concepts attribute is the shortest way to talk to the NIST Cybersecurity Framework. But there’s a difference: NIST CSF 2.0 defines six functions (Govern, Identify, Protect, Detect, Respond, Recover), while ISO 27002’s cybersecurity concepts consist of five values and have no Govern equivalent. When linking governance controls to the CSF Govern function, you can use the #Governance and #Legal_and_compliance values in the operational capability attribute and #Governance_and_Ecosystem in the security domain.
Attribute mapping is a rough pre-classification; it doesn’t replace one-to-one mapping at control level. When deciding whether a CSF subcategory is met, the implementation guidance of the relevant ISO control and your evidence must be assessed together. We cover building a profile on the CSF side in the article building a NIST CSF 2.0 profile on our cybersecurity site; for the framework itself, see the NIST CSF page.
Defining your own attributes
ISO 27002 explicitly states that organizations can define their own attributes. In addition to the five standard attributes, adding these columns to the control inventory is useful in most organizations:
- Regulatory mapping: KVKK, NIS2, TISAX (VDA ISA) or a customer contract clause. It avoids duplication when managing more than one compliance program with the same control set.
- Location or scope unit: In multi-site organizations, which sites the control applies to.
- Evidence source and frequency: Which system produces the control’s evidence and how often. For example, a monthly patch compliance report from WSUS or the endpoint management console.
- Degree of automation: Manual, semi-automated, automated. Manual controls need their operating effectiveness tested more often.
- Maturity: On the organization’s own scale; target and current levels in separate columns.
When you define your own attributes, put the value lists under document control and write down their definitions. If what “partial” implementation status means isn’t defined, everyone who reads the report understands it differently.
Mistakes you may come across when using attributes:
- Treating an attribute as the justification for selecting a control: The logic “this control is #Corrective, we have few incidents, let’s exclude it” can’t be defended in the SoA. The selection justification always comes from the risk assessment.
- Changing values locally: Changing the standard values because they’re “more accurate for us” breaks comparison with external frameworks. If a different classification is needed, keep the standard values and add a new attribute.
- Mistaking numbers for maturity: Having many “implemented” controls in a security domain doesn’t show that the domain is effective. Operating effectiveness is tested separately.
An attribute-based control inventory doesn’t replace the SoA; it must be kept consistent with the selection justification and implementation status in the SoA. For the SoA structure, see the ISO 27001 SoA guide, and for theme-based interpretations of controls, the article Annex A organizational controls. To analyze your current control set by attribute, see our ISO 27001 gap analysis service, and to map controls to risks, our risk analysis and SoA service. The official source for the attribute table is the ISO/IEC 27002:2022 standard.
Frequently asked questions
Are attributes mandatory for ISO 27001 certification?
No. Attributes appear only in ISO 27002 and are informative. The certification audit is carried out against the requirements of ISO 27001; using attributes is the organization’s choice.
Do we need to add attribute columns to the SoA?
No. The mandatory content of the SoA is the controls, the justification for selecting them, implementation status and justification for exclusions. Keeping attributes in a separate control inventory and linking it to the SoA by control number keeps the SoA simple.
Are the cybersecurity concepts identical to the NIST CSF functions?
Five of them match by name: Identify, Protect, Detect, Respond and Recover. NIST CSF 2.0 also includes the Govern function; ISO 27002 has no direct concept equivalent to it, and governance controls are linked through the other attributes.
Should we show the auditor attributes we’ve defined ourselves?
It isn’t mandatory; but attributes such as control ownership, evidence source and regulatory mapping help you get to evidence quickly in the audit. If you’re going to show them, the value definitions must be documented and up to date.