Risk management foundations
Risk Assessment: ISO 31000 Process, Techniques and Risk Matrix Limits
Risk assessment is the overall process of identifying, analysing and evaluating risk, so that decision-makers understand what could happen, how likely and how severe it is, and whether it needs treatment.
Published Topic: Risk management foundations
Risk assessment is the overall process of identifying, analysing and evaluating risk, so that decision-makers understand what could happen, how likely and how severe it would be, and whether it needs treatment. In ISO 31000:2018 it is clause 6.4, made of three steps: risk identification, risk analysis and risk evaluation, carried out against criteria agreed beforehand.
Key takeaways
- ISO 31000 sets the process; IEC 31010:2019 explains how to choose and apply the techniques [1][2].
- The 2019 edition of IEC 31010 describes 42 techniques in ten groups, from expert elicitation to value at risk [3].
- Qualitative, semi-quantitative and quantitative methods trade ease of communication against comparability and rigour [5].
- Risk matrices are useful for communication but can misrank risks; they should not drive resource allocation on their own [4].
- Assessments of complex and systemic risks need dependency analysis, compound scenarios and explicit uncertainty, not only a register [9][11].
What is a risk assessment?
ISO 31000 defines risk as the "effect of uncertainty on objectives", usually expressed in terms of risk sources, potential events, their consequences and their likelihood [1]. A risk assessment turns that definition into evidence for a decision. It answers four questions: what could happen and why (identification), how likely and how serious it is (analysis), whether it is acceptable against the organisation's criteria (evaluation), and how confident we are in the answer.
In ISO 31000 the assessment sits within a wider process: communication and consultation; scope, context and criteria; risk assessment (identification, analysis and evaluation in clauses 6.4.2 to 6.4.4); risk treatment; monitoring and review; and recording and reporting [1]. The vocabulary is standardised in ISO 31073:2022, which replaced ISO Guide 73 [7].
The disaster risk community uses a parallel definition. The UN Office for Disaster Risk Reduction (UNDRR) describes disaster risk assessment as "a qualitative or quantitative approach to determine the nature and extent of disaster risk by analysing potential hazards and evaluating existing conditions of exposure and vulnerability" [6]. Understanding disaster risk is Priority 1 of the Sendai Framework [10].
The ISO 31000 risk assessment process, step by step
- Set scope, context and criteria. Define the decision the assessment supports, the objectives at stake, the time horizon, and the risk criteria (how likelihood and consequence will be measured and what level of risk is tolerable) (clause 6.3) [1].
- Identify risks. Find, recognise and describe risks, including their sources, events, causes and potential consequences (clause 6.4.2) [1].
- Analyse risks. Understand the nature of each risk, its likelihood and consequences, the effectiveness of existing controls, and dependencies between risks (clause 6.4.3) [1].
- Evaluate risks. Compare the results of the analysis with the risk criteria to decide where additional action is needed, which may include no further action, treatment, further analysis or a review of objectives (clause 6.4.4) [1].
- Treat risks. Select and implement options and plans (clause 6.5) [1].
- Monitor, review, record and report. Keep the assessment current and make its results and assumptions available to those who decide (clauses 6.6 and 6.7) [1].
IEC 31010:2019 adds a practical layer for planning and quality: define the purpose and scope; understand the context; engage stakeholders; consider human, organisational and social factors; review criteria for decisions; manage information and develop models; verify and validate results; and run uncertainty and sensitivity analysis before applying results to decisions [3].
Risk assessment techniques: what IEC 31010:2019 offers
IEC 31010:2019, Risk management — Risk assessment techniques, is the second edition of the standard. It replaced the 2009 edition, increased "the number and range of application of the techniques", and stopped repeating concepts already covered in ISO 31000 [3]. Its Annex B describes 42 techniques in ten groups [3].
| Purpose (IEC 31010 Annex B group) | Examples of techniques |
|---|---|
| Eliciting views from stakeholders and experts | Brainstorming, Delphi technique, nominal group technique, structured interviews, surveys |
| Identifying risk | Checklists and taxonomies, FMEA/FMECA, HAZOP, scenario analysis, structured what-if (SWIFT) |
| Determining sources, causes and drivers | Cindynic approach, Ishikawa (fishbone) analysis |
| Analysing controls | Bow tie analysis, HACCP, layers of protection analysis (LOPA) |
| Understanding consequences and likelihood | Bayesian networks, business impact analysis, event tree and fault tree analysis, Markov analysis, Monte Carlo simulation |
| Analysing dependencies and interactions | Causal mapping, cross impact analysis |
| Providing a measure of risk | Toxicological risk assessment, value at risk (VaR), conditional value at risk |
| Evaluating significance | ALARP/SFAIRP, F-N diagrams, Pareto charts, reliability centred maintenance, risk indices |
| Selecting between options | Cost/benefit analysis, decision tree analysis, game theory, multi-criteria analysis |
| Pencatatan dan pelaporan | Risk registers, consequence/likelihood matrix, S-curves |
The grouping is itself useful advice: choose techniques by the question being asked, not by habit. A register and a matrix record risks; they do not analyse dependencies or measure tail losses.
Qualitative vs quantitative risk assessment
The US National Institute of Standards and Technology (NIST) gives clear working definitions in SP 800-30 Rev. 1 (September 2012) [5].
| Pendekatan | How NIST describes it [5] | Strengths | Kelemahan | Typical uses |
|---|---|---|---|---|
| Qualitative | Nonnumerical categories or levels (very low to very high) | Fast; easy to communicate; works with little data | Small range of values makes prioritisation difficult; experts may interpret levels differently | Screening, workshops, early-stage registers |
| Semi-quantitative | Bins, scales or representative numbers whose meanings are not maintained in other contexts | Combines communication with some comparison | Scores look precise but are ordinal; arithmetic on them can mislead | Composite indices, cyber and operational risk scoring |
| Quantitative | Numbers whose meaning and proportionality are maintained inside and outside the assessment | Supports cost-benefit analysis of responses | Needs data and models; results need interpretation of assumptions | Catastrophe models, VaR, reliability, insurance pricing |
The choice should follow the decision. A quantitative model is worth building when the decision involves significant money, irreversible consequences or comparison across many options. Otherwise, a well-structured qualitative assessment with explicit uncertainty can be more honest than a spuriously precise number.
The limits of the risk matrix
The consequence/likelihood matrix, or heat map, is the most widely used tool in risk management, and IEC 31010 lists it as a recording and reporting technique rather than an analysis method [3]. Its limits are well documented. Cox showed in 2008 that typical matrices can correctly and unambiguously compare only a small fraction of randomly chosen pairs of hazards, and can "assign higher qualitative ratings to quantitatively smaller risks" [4]. He concluded that matrices should be used with caution and with careful explanations of the judgements built into them [4].
Four practical problems follow:
- Range compression. A five-by-five grid puts risks whose expected losses differ by orders of magnitude into the same cell.
- Ordinal arithmetic. Multiplying a likelihood rank by a consequence rank produces a number with no stable meaning [5].
- Single-scenario thinking. Most risks have a range of consequences; a single cell hides the tail.
- No interactions. A matrix treats each risk as independent, so it cannot represent compound or cascading risks [11].
Use matrices to communicate, with defined scales and stated limits, and use quantitative or scenario methods where decisions depend on ranking.
Assessing complex, compound and systemic risks
Conventional assessments assume risks are independent and can be scored one at a time. For climate, disaster and infrastructure risks that assumption fails. Zscheischler and colleagues showed that floods, wildfires, heatwaves and droughts often arise from interacting processes, and that assessing one driver at a time can underestimate risk [11]. The IPCC's guidance for authors treats climate risk as the result of "dynamic interactions between climate-related hazards with the exposure and vulnerability" of the system affected, and extends risk to the responses to climate change themselves [9].
For these risks, the assessment should add dependency mapping, compound and cascade scenarios, and an explicit treatment of uncertainty. The IRGC framework adds a pre-assessment step to frame the problem and identify stakeholders before appraisal, and a characterisation and evaluation step that judges acceptability with decision-makers [8].
Examples: matching the technique to the question
The same process applies across sectors, but the techniques differ with the question and the data available.
- A water utility planning for drought. Scenario analysis frames dry-year sequences; a fault tree or Bayesian network traces how supply fails; cost/benefit and multi-criteria analysis compare storage, demand management and interconnection options [3].
- A hospital assessing power dependency. Business impact analysis sets maximum tolerable outages for each service; bow tie analysis shows the barriers between a grid failure and a loss of critical care; a risk register records owners and review dates [3].
- A city exposed to river flooding. A disaster risk assessment combines hazard maps, exposure of people and assets, and vulnerability, then evaluates coping capacity under likely scenarios [6]; compound scenarios combine river flow with intense rainfall or storm surge [11].
- A bank's climate stress test. Quantitative scenario models translate hazards and transition pathways into losses; value at risk or expected shortfall summarise tail exposure [3].
- An information system. NIST SP 800-30 structures the assessment around threat sources, vulnerabilities, likelihood and impact, with qualitative or semi-quantitative scales [5].
In each case the assessment is only as good as its framing: the scope, the criteria and the people consulted determine what the techniques can reveal.
Common mistakes in risk assessment
- Starting with a template, not a decision. The scope and criteria determine which risks and techniques matter.
- Confusing hazards with risks. A hazard becomes a risk only in relation to objectives, exposure and vulnerability [1][6].
- Ignoring existing controls, or assuming they work. ISO 31000 notes that controls "may not always exert the intended or assumed modifying effect" [1].
- Treating scores as measurements. Semi-quantitative scores are ordinal and should not be added or averaged across domains [5].
- No uncertainty statement. Every result should state its confidence and main assumptions [3].
- One-off assessments. Risks change; ISO 31000 requires monitoring and review [1].
How GCRI works on risk assessment
GCRI's method pages distinguish four kinds of number, each labelled differently: teaching values, illustrative applications, published benchmarks and specifications. Teaching values, such as the strengths of the thirty relations between six sectors in its coupled-risk model, are never presented as findings for a place. Published benchmarks are cited on the same line, with their scope, and are "not a forecast for any project or place". GCRI's evidence system classifies each claim as observed, officially reported, institution-reported, estimated, modelled, forecast, scenario, expert judgement or unknown, so that simulated results are never reported as observed impact.
The assessment method follows ISO 31000 and lists where a real analysis must go beyond a teaching cascade: place-specific relation strengths from outage records and operator data, thresholds and non-linear failure, restoration times and buffers, hazard footprints and asset locations, behaviour and institutions, and ranges with stated confidence. A participating country carries this out through its own National Model, under its own data terms. GCRI's systems modelling work covers model intercomparison, uncertainty and validation. GCRI does not certify or approve assessments, and its programme is at service level R0, as recorded on the programme status page.
Further reading
- ISO 31000:2018, clauses 6.3 to 6.7 [1]
- IEC 31010:2019, clauses 6 and 7 and Annex B [2][3]
- NIST SP 800-30 Rev. 1, chapter 2 on assessment approaches [5]
- Cox (2008), What's wrong with risk matrices? [4]
- UNDRR terminology on disaster risk assessment [6]
Written by the GCRI knowledge team. Last reviewed 2026-10-30. Corrections are welcome through the knowledge base report link on this page.
Istilah kunci
- penilaian risiko
- The overall process of risk identification, risk analysis and risk evaluation; clause 6.4 of ISO 31000:2018. [1]
- Kemungkinan
- The chance of something happening, whether defined, measured or determined objectively or subjectively, qualitatively or quantitatively (ISO 31000:2018, 3.7). [1]
- Konsekuensi
- The outcome of an event affecting objectives; it can be certain or uncertain, positive or negative, and can escalate through cascading and cumulative effects (ISO 31000:2018, 3.6). [1]
- Risk source
- An element which alone or in combination has the potential to give rise to risk (ISO 31000:2018, 3.4). [1]
- Matriks risiko
- A grid that ranks risks by ordinal categories of consequence and likelihood; listed in IEC 31010:2019 as a recording and reporting technique and known to misrank some risks. [3]
- Semi-quantitative assessment
- Assessment using bins, scales or representative numbers whose values and meanings are not maintained in other contexts, such as a 1 to 10 score (NIST SP 800-30 Rev. 1). [5]
- Disaster risk assessment
- A qualitative or quantitative approach to determine the nature and extent of disaster risk by analysing potential hazards and evaluating existing conditions of exposure and vulnerability (UNDRR terminology). [6]
Frequently asked questions
What are the steps of a risk assessment?
ISO 31000 defines risk assessment as three steps: risk identification, risk analysis and risk evaluation. They sit inside a wider process that starts with scope, context and criteria and continues with risk treatment, monitoring and review, and recording and reporting, with communication and consultation throughout.
What is the difference between qualitative and quantitative risk assessment?
Qualitative assessment uses descriptive categories such as low, moderate and high, which are quick and easy to communicate but hard to compare. Quantitative assessment uses numbers whose meaning holds outside the assessment, such as probabilities and monetary losses, which support cost-benefit analysis but need data and explanation. Semi-quantitative methods use scores or bins in between.
Who should carry out a risk assessment?
A risk assessment should be led by people who understand the objectives at stake and carried out with those who hold relevant knowledge: operators, specialists, affected stakeholders and, for complex risks, independent reviewers. ISO 31000 treats communication and consultation as part of the process, and IEC 31010 includes techniques for eliciting expert and stakeholder views. Accountability for accepting the result stays with the decision-maker.
What is IEC 31010?
IEC 31010:2019, Risk management — Risk assessment techniques, is the companion standard to ISO 31000 that explains how to plan an assessment and choose techniques. Its second edition describes 42 techniques in ten groups, from brainstorming and the Delphi technique to bow tie analysis, Monte Carlo simulation, value at risk and the consequence/likelihood matrix.
Why are risk matrices criticised?
Risk matrices compress continuous likelihoods and consequences into a few ordinal boxes, so risks with very different expected losses can land in the same cell and some risks can be ranked in the wrong order. Cox showed in 2008 that matrices can assign higher ratings to quantitatively smaller risks. They remain useful for communication if their scales are defined and their limits stated.
What is a disaster risk assessment?
A disaster risk assessment determines the nature and extent of disaster risk by analysing hazards and evaluating exposure and vulnerability, as UNDRR defines it. It typically covers hazard location, intensity, frequency and probability, the people and assets exposed, their vulnerability, and the capacity to cope under likely scenarios.
Sumber
- International Organization for Standardization. (2018). ISO 31000:2018 Risk management — Guidelines.
- International Electrotechnical Commission / International Organization for Standardization. (2019). IEC 31010:2019 Risk management — Risk assessment techniques.
- International Electrotechnical Commission, via VDE Verlag. (2019). IEC 31010:2019 Edition 2.0, preview (foreword, introduction and table of contents).
- Society for Risk Analysis / Wiley. (2008). What's Wrong with Risk Matrices? (Cox), Risk Analysis 28(2): 497-512.
- National Institute of Standards and Technology. (2012). NIST Special Publication 800-30 Revision 1: Guide for Conducting Risk Assessments.
- United Nations Office for Disaster Risk Reduction (UNDRR). (2017). Sendai Framework terminology: disaster risk assessment.
- International Organization for Standardization. (2022). ISO 31073:2022 Risk management — Vocabulary.
- International Risk Governance Center (IRGC), EPFL. (2017). IRGC Risk Governance Framework.
- Intergovernmental Panel on Climate Change (IPCC). (2020). Use of 'risk' concepts in IPCC assessments: Guidance for IPCC authors.
- United Nations. (2015). Sendai Framework for Disaster Risk Reduction 2015-2030.
- Nature Portfolio. (2018). Future climate risk from compound events (Zscheischler et al.), Nature Climate Change 8: 469-477.