The risk assessment is finished, twenty high risks are listed, the meeting minutes are signed. Six months later you look at the open actions and none of them has an owner, a date or a budget. The risk assessment says which risks matter; the risk treatment plan sets what will be done about them, by whom, by when and with what resources. ISO/IEC 27001:2022 clause 6.1.3 defines this step from selecting treatment options through the Statement of Applicability (SoA) to risk owner approval; clause 8.3 requires the plan to be implemented and its results recorded. Below we explain how to build a traceable plan that doesn’t stay on paper.
Risk treatment options
ISO/IEC 27005:2022 defines four basic options. In practice they may go by different names; what matters is that the meaning, not the name, is clear:
| Option | What does it do? | Typical example |
|---|---|---|
| Avoid | Stops or changes the activity that gives rise to the risk | Completely banning the processing of customer data on personal devices |
| Reduce (modify) | Lowers likelihood or impact with controls | MFA, network segmentation, immutable backups |
| Share (transfer) | Transfers part of the impact to another party | Cyber insurance, contractual liability, managed services |
| Accept (retain) | Knowingly takes on the risk as it is | Putting a low-impact vulnerability under monitoring |
More than one option is used together for a single risk. In a ransomware scenario, controls that lower likelihood (email filtering, patch management), controls that lower impact (offline or immutable backups, restore testing) and an insurance policy that shares the financial impact can all be in the same plan. A single-option plan is usually an incomplete plan.
Decision order for choosing options
When deciding on the option, ask these questions in order:
- Is it within the risk acceptance criteria? If so, and the authorized person approves, acceptance is enough. The criteria should have been defined in the risk acceptance criteria work.
- Is the activity really necessary for the business? If not, avoidance is often the cheapest and most effective option. Closing an unused external access channel is easier than trying to protect it. Port forwards on the firewall that have stood open for years with nobody owning them are a typical example in the field. Close them and the risk goes away.
- Is the cost of the control proportionate to the expected impact? Reduction requires comparing the control’s cost with the possible impact. The comparison can be qualitative but must be recorded.
- Can the impact be transferred financially? Sharing transfers only the financial impact. Legal liability, customer trust and operational disruption stay with the organization. Even when a service is outsourced, risk ownership and accountability continue to sit with the organization.
- What happens until the control takes effect? If the permanent fix will take months, a temporary compensating measure (extra monitoring, access restriction) must appear as a separate row in the plan. The gap in between is a risk too.
Fields the plan table must have
The risk treatment plan can be a separate document or be kept as columns of the risk register produced by the information security risk assessment. Whatever the format, every action must have these fields:
| Field | Explanation |
|---|---|
| Risk No / option | The linked risk and the treatment option chosen |
| Action | A concrete task description with a verifiable result |
| Related control | Annex A or internal control reference |
| Owner | The person carrying out the action (may differ from the risk owner) |
| Resources / budget | Approved budget line, labor or “existing resources” |
| Target date | End date along with milestones |
| Expected residual risk | Target L × I when the action is complete |
| Status and evidence | Planned / in progress / complete / delayed + evidence link |
The action sentence decides the plan’s quality. “Access security will be improved” isn’t an action; nobody can say when it’s finished. “MFA will be made mandatory on all accounts with VPN access; the exception list will be approved by the ISMS manager” is an action, and its completion can be verified. An action that can’t be verified doesn’t close.
Sample plan rows
| Risk / action | Check | Owner / date | Residual target |
|---|---|---|---|
| R-07 Ransomware: move backups to immutable storage outside the domain | A 8.13 | System administrator / Q1 | 3×5 → 2×3 |
| R-07 Ransomware: quarterly restore test and record | A 8.13 | System administrator / ongoing | — |
| R-11 Supplier remote access: convert permanent VPN accounts to on-request access | A 5.19, A 8.5 | Network owner / Q2 | 4×4 → 2×4 |
| R-15 Customer drawings on personal devices: remove BYOD access | A 8.1 | IT manager / Q1 | Risk closed (avoidance) |
Control numbers follow ISO/IEC 27001:2022 Annex A: 5.19 information security in supplier relationships, 8.1 user endpoint devices, 8.5 secure authentication, 8.13 information backup. Note the second row in R-07: moving backups to immutable storage is a one-off task, while restore testing is an ongoing activity. If you write both on the same row, when the first is finished the second also appears “complete”. A comprehensive assessment of supplier risk needs a separate process; details are in the supplier information security risk assessment guide.
Link to the SoA
Clauses 6.1.3 b) and c) require determining the controls needed for treatment and comparing them with Annex A to verify that no necessary control has been omitted. Clause 6.1.3 d) carries this result into the Statement of Applicability. The link is built like this:
- Every control reference in the plan must appear as “applicable” in the SoA.
- The related risk numbers can be written in the SoA as the justification for including the control; that’s the shortest traceability path for the auditor.
- A control marked “implemented” in the SoA but with an open action in the plan is an inconsistency. Fill in the implementation status column to reflect reality, for example “partly implemented, R-11 action open”.
- When a risk is closed by avoidance, the SoA justification is reviewed again only for controls chosen for that risk.
For the structure of the SoA itself and exclusion justifications, see the what is the ISO 27001 SoA guide.
Residual risk and risk owner approval
Clause 6.1.3 f) requires risk owners both to approve the plan and to accept residual risks. In practice there are two separate moments of approval:
- Plan approval: The risk owner accepts that the actions, budget and dates are adequate. Without this approval, the plan is the IT team’s well-meant to-do list.
- Residual risk acceptance: After the actions are complete, the risk is re-scored. If the actual residual risk is higher than expected, the risk owner either asks for further treatment or accepts the actual value.
That’s why the expected residual risk is written at the planning stage: it gives a reference point for measuring whether the action worked. When re-scoring, look for evidence that the control that lowers likelihood actually works. If the MFA policy was switched on but twenty accounts were forgotten in the exception group, the control is in place but not working. “Installed” and “working” aren’t the same.
Following up plan progress
Most risk treatment plans fail not because they’re badly written but because they aren’t followed up. For a follow-up setup that works:
- Use a single source: Keeping actions in a separate Excel file, in the ticketing system and in meeting notes produces inconsistency. Open the actions in your existing work tracking tool and tag them with the risk number; let the plan reference these records.
- Monthly status review: The ISMS manager reviews open actions with their owners; for delayed ones, a new date and the reason for the delay are written down.
- Escalation rule: Actions whose target date goes over the set period, or that are postponed a second time, are reported to the risk owner and top management.
- Ask for evidence at closure: “Complete” status is marked only with an evidence link (configuration screenshot, system report, approval record).
- Use indicators: Indicators such as the number of overdue actions and average closure time for high risks become inputs to management review. Indicator design is covered in the key risk indicators guide.
Mistakes you may come across
- Marking all risks as “reduce”, without ever considering avoidance and acceptance
- Writing actions that need budget into the plan without getting budget approval
- Writing a department name instead of a person as the owner
- Not re-scoring the risk after an action is completed
- Not informing the risk owner of plan changes (postponed dates, narrowed scope)
If the owner column says “IT”, that action has no owner. A department doesn’t do work; a person does.
Designing the risk treatment plan together with the SoA and setting up the follow-up routine are among the main outputs of our ISO 27001 risk analysis, risk treatment and SoA work. For the standard definitions of the treatment options, ISO/IEC 27005:2022 is the primary source.
Frequently asked questions
Does the risk treatment plan have to be a separate document?
No. ISO/IEC 27001 requires a risk treatment plan to be formulated and kept as documented information; it doesn’t specify the format. Columns added to the risk register or tagged records in a work tracking system are also enough; the action, owner, date and approval must be traceable.
Does the risk owner change when a risk is shared?
No. Insurance or outsourcing transfers part of the financial or operational impact; ownership of the risk and accountability stay with the organization. The risk owner continues to assess the adequacy of the transfer and the remaining risk.
When should a risk be re-scored after an action is completed?
After the action is closed and evidence has been obtained that the control works. A control being deployed and working effectively are different things; residual risk is updated when it’s shown that the control is actually operating.
What happens if the plan and the SoA are inconsistent?
It’s one of the points often questioned in certification audits. If a control that appears implemented in the SoA has an open action in the plan, the auditor asks which is correct. Matching the implementation status in the SoA with the current status of the plan significantly reduces this risk.