GRC Risk Management: A Practical Guide
How risk management fits into a GRC program: ownership, risk appetite, a worked register entry from inherent to residual, and what NIS2 actually requires.
GRC risk management is the risk half of governance, risk and compliance: the part that decides which threats to your objectives get money and attention, and which get accepted in writing. Governance sets who decides and within what limits. Compliance proves you met an external obligation. Risk management is what connects them — it turns a list of things that could go wrong into ranked, owned decisions that governance can approve and compliance can evidence.
Most explanations of GRC stop at that three-word definition. This guide shows the mechanics: who owns what, how appetite is set, how a risk moves from inherent to residual with real numbers, and what European regulation now requires. More on the discipline sits in our complete guide to GRC in cybersecurity.
How does risk management fit into a GRC program?
Think of the three letters as a loop rather than three departments.
Governance sets objectives, risk appetite and decision rights. It answers: how much of this risk are we willing to carry, and who is allowed to accept it?
Risk management identifies what threatens those objectives, scores it consistently, assigns treatment, and reports what remains against appetite.
Compliance takes the controls chosen as treatment and demonstrates, with evidence, that they operate — to auditors, regulators and customers.
The loop closes because compliance findings and control failures feed back as new risks. A control that fails its test is not only a compliance exception; it is a risk whose residual score just moved.
NIST CSF 2.0 (NIST CSWP 29, published 26 February 2024) made this structure explicit by adding GOVERN as a sixth Function alongside IDENTIFY, PROTECT, DETECT, RESPOND and RECOVER. GOVERN covers how an organisation's cybersecurity risk management strategy, expectations and policy are established, communicated and monitored, and NIST is direct that governance activities matter because they connect cybersecurity to enterprise risk management.
GRC, ERM and IRM: what is the difference?
The acronyms overlap, which causes real confusion in tooling decisions.
| Term | Scope | Primary question |
|---|---|---|
| GRC | Governance, risk and compliance as one operating model, usually anchored in IT and security | Are we governed, managing risk and provably compliant? |
| ERM | Enterprise risk across all categories: strategic, financial, operational, hazard | What threatens the enterprise's objectives overall? |
| IRM | Integrated risk management; risk viewed through business context and dependencies | What does this risk mean for the business service it supports? |
They are not competing standards. Cyber risk managed under GRC should roll up into ERM — which is exactly what CSF 2.0 subcategory GV.RM-03 asks for, requiring that cybersecurity risk management activities and outcomes are included in enterprise risk management processes.
Who owns risk in a GRC program?
Unclear ownership is the most common reason a register goes stale. Separate four roles and write them down.
| Role | Responsibility |
|---|---|
| Risk owner (business leader) | Accountable for the risk and its treatment decision; owns the residual outcome |
| Control owner (operational) | Runs the control day to day and produces evidence it works |
| Risk function (GRC team) | Owns the method, the scoring scale and the register's integrity — not the risks themselves |
| Management body / board | Sets appetite, accepts risks above threshold, reviews the portfolio |
The distinction that matters: the GRC team owns the method, not the risks. When a risk function starts owning risks, business ownership evaporates and treatment stops happening.
NIS2 reinforces this at board level. Article 20 places management bodies under an obligation to approve cybersecurity risk-management measures and oversee implementation, and Member States must ensure they can be held liable for infringements.
How do you set risk appetite?
Appetite is the sentence that lets someone say no without escalating. CSF 2.0 subcategory GV.RM-02 requires that "risk appetite and risk tolerance statements are established, communicated, and maintained".
Write appetite as a threshold on the same scale you score risks:
- Appetite — the level of residual risk you will carry without further treatment. For example: residual score 8 or below on a 5×5 scale.
- Tolerance — how far above appetite you will temporarily sit while treatment runs, with a deadline. For example: up to 12 for a maximum of two quarters.
- Escalation — above tolerance, the management body decides: fund the treatment, or formally accept and record it.
Set appetite per risk category, not one number for the whole organisation. Availability risk on a customer-facing service and data-confidentiality risk carry different tolerances in almost every business.
A worked risk register entry
Here is where most GRC guidance stops being useful. This is one row, end to end, on a 5×5 scale where score = likelihood × impact.
Risk: An unpatched internet-facing VPN appliance is exploited, giving an attacker network access.
Step 1 — inherent score. Before considering existing controls. Likelihood 4 (edge appliances are actively targeted and exploits are public). Impact 5 (network-wide access, personal data in scope). Inherent = 4 × 5 = 20. Above appetite of 8, and above tolerance of 12.
Step 2 — existing controls and their real effect.
| Control | Working? | Effect |
|---|---|---|
| Monthly patch cycle | Partially — 30-day SLA, exploits land faster | Reduces likelihood a little |
| MFA on VPN | Yes | Reduces likelihood materially |
| Network segmentation behind VPN | No — flat internal network | No reduction in impact |
| Monitoring and alerting on appliance | Yes, but tuned noisily | Reduces impact slightly |
Step 3 — residual score. Likelihood drops from 4 to 3 (MFA and patching help, but the SLA is too slow for edge devices). Impact stays at 5, because nothing implemented reduces blast radius. Residual = 3 × 5 = 15. Still above tolerance.
Step 4 — treatment decision. The register makes the gap obvious: every control addresses likelihood, none addresses impact. Treatment therefore targets impact — segment the internal network behind the VPN — and secondarily tightens the patch SLA to 7 days for internet-facing appliances. Projected residual after treatment: 2 × 3 = 6, inside appetite.
Step 5 — owner and date. Infrastructure lead owns it, segmentation delivered within two quarters, tracked as an exception at tolerance in the meantime.
Two things make this row work. Impact and likelihood are scored separately, so you can see which one your controls actually move. And the residual score is recalculated from the controls that genuinely operate, not from the ones written in a policy. CSF 2.0 asks for exactly this consistency in GV.RM-06: a standardised method for calculating, documenting, categorising and prioritising risks.
A risk register that links each risk to its controls, findings, assets and vendors keeps this reasoning visible. When the register is a spreadsheet, the link between a failing control and the risk it was supposed to treat is the first thing lost.
What does ISO 31000 add?
ISO 31000:2018, Risk management — Guidelines, published in 2018 and confirmed as current in 2023, gives the process that sits underneath the register: identifying, analysing, evaluating, treating, monitoring and communicating risk, supported by a set of principles and a framework for embedding risk management into governance and decision-making.
Its contribution to a GRC program is discipline rather than a checklist. ISO 31000 is guidance, not a certifiable requirement, and it pushes you to embed risk in decisions rather than run it as a parallel reporting exercise.
What does NIS2 require of risk management?
For entities in scope, risk management stops being discretionary.
Article 21(1) requires appropriate and proportionate technical, operational and organisational measures to manage risks to network and information systems, and says the measures must be proportionate to the entity's exposure, size, and the likelihood and severity of incidents — which is risk-based scoring stated as law.
Article 21(2)(a) requires "policies on risk analysis and information system security" as the first of the ten minimum measures. Article 21(2)(f) requires "policies and procedures to assess the effectiveness of cybersecurity risk-management measures" — so you must be able to show that your treatment worked, not just that it was chosen. Article 21(4) requires an entity that finds it does not comply to take corrective measures "without undue delay".
Supply chain sits inside the same article: 21(2)(d) covers the security of relationships with direct suppliers. In practice that means third-party risks belong in the same register and on the same scale as internal ones, which is why a third-party risk management platform should feed the register rather than maintain a separate one.
Getting this working across scoring, controls and evidence in one place is what our GRC software is built for — see ResilientX GRC in action if your register and your control evidence currently live in different systems.
How do you keep the register honest?
- Score from evidence, not memory. A control counts toward residual risk only if its last test passed.
- Cap the register. Hundreds of unranked risks means nothing is ranked. Merge duplicates into the risk that owns the objective.
- Feed it from detection. Findings from exposure management should raise or close risks automatically, not via a quarterly copy-paste.
- Filter the noise. Alert volume inflates apparent likelihood; understanding false positives keeps scoring grounded.
- Review on a cycle and on triggers. Quarterly for risks above appetite, annually for the rest, immediately after a material incident or change.
Frequently asked questions
What is the difference between GRC and risk management?
Risk management is one of the three disciplines inside GRC. On its own it identifies, scores and treats risk. GRC adds governance — who sets appetite and who may accept risk — and compliance, which proves the resulting controls operate. Risk management without governance produces registers nobody acts on.
Is GRC risk management the same as cybersecurity risk management?
Not quite. Cybersecurity risk management covers threats to information systems. GRC risk management is the operating model that also handles regulatory, third-party, operational and policy risk, and rolls cyber risk up into enterprise reporting. NIST CSF 2.0 makes the connection explicit through its GOVERN function.
How do you measure whether GRC risk management is working?
Track how many risks sit above appetite and for how long, the share of treatments delivered by their due date, the proportion of controls with current passing evidence, and the time between a finding and the register reflecting it. Volume of risks logged measures activity, not effectiveness.
Do small organisations need a formal risk register?
Yes, but proportionate. NIS2 Article 21(1) ties the measures to entity size and exposure. A 40-person company can run a credible register of 15 to 25 risks with named owners and a simple 5×5 scale. The failure mode at small scale is not an unsophisticated register — it is no owner and no review date.
Risk management earns its place in GRC when a score changes a decision. Set appetite as a number, score inherent and residual separately, and make every risk carry an owner and a date.
See ResilientX GRC in action — book a demo and bring your current register; the gaps usually show up in the first ten rows.