Demo

Step 3 · ISO/IEC 27005:2022, Clause 7

ISO 27005 level of risk: assessing consequences and likelihood

Analysis gives each identified risk a level, from two estimates: what the organisation would lose if the scenario happened, and the chances that it happens. It does not invent its scales: it applies the ones set with the context. The whole point is to rate in a justified and reproducible way.

Updated on · 5 min read

Clause
7, Risk assessment: analysis
Participants
Risk owners, business owners, CISO, CIO
Deliverable
Consequences, likelihood and level of each risk, justified
Guide chapters
  1. Applying ISO 27005
  2. Context and risk criteria
  3. Identify risks
  4. Analyse risks
  5. Evaluate risks
  6. Treat risks
  7. Monitor and review
  8. ISO 27001 and ISO 31000
  9. ISO 27005 and EBIOS RM
  10. Glossary

In brief

  1. Analysis consists of three operations: assessing potential consequences, assessing likelihood, then determining the level of risk. These are also the three points of clause 6.1.2 d) of ISO/IEC 27001.

  2. It can be qualitative, semi-quantitative or quantitative. The choice depends on the data available, not on a hierarchy between the approaches.

  3. Consequences are assessed with the business, on what has value; likelihood is assessed with those who know the systems, the threats and the controls in place.

  4. Each rating comes with a written justification: this is what makes the assessment "valid" and allows it to be repeated the following year.

  5. The level obtained takes existing controls into account. It will serve as the point of comparison for the residual risk after treatment.

Qualitative, semi-quantitative or quantitative

The standard allows three types of analysis. None is better than the others in itself: a quantitative analysis built on fragile figures is less reliable than a qualitative analysis whose levels are well defined.

The three types of analysis
TypePrincipleSuitable whenLimit
QualitativeLevels described in words (limited, serious, critical)Little data, need to talk with the businessDepends on the quality of the definitions
Semi-quantitativeLevels associated with values or rangesYou want to rank and aggregate without precise measurementsThe numbers give an impression of precision they do not have
QuantitativeMeasured values: amount of losses, annual frequencyReliable data history, financial trade-off to justifyRequires data that few organisations have

Assess the consequences

You assume the scenario has happened, and you ask the business concerned what it loses. The scale set with the context serves as the reading grid: length of interruption, cost, breach of an obligation, harm to people or to reputation.

  • Look at all three properties. The same event can affect availability (the workshop stops), confidentiality (drawings leak) and integrity (machining programs are altered). You keep the heaviest consequence and note the others.
  • Count indirect consequences. Penalties, notification to an authority, loss of a customer: they often weigh more than the cost of restoration.
  • Take timing into account. A three-day stoppage does not weigh the same in a quiet period and the day before a delivery.
  • Have the business do the rating. The risk owner is best placed to say what an interruption costs.

Assess the likelihood

Likelihood is the chance of the scenario happening. For an accidental event, history helps: past failures, frequency of errors. For a deliberate act, you reason instead about the motivation and means of the source, the exposure of the targeted asset and the obstacles to get through.

Diagram of the analysis of risk R1 in the example, ransomware entering through remote maintenance. On the left, the consequences, assessed with the production director: workshop stopped for more than ten days, penalties, level 4 out of 4. On the right, the likelihood, assessed with the CIO: permanent remote access, shared account without multi-factor authentication, network not separated, level 3 out of 4. The two ratings meet in a four-by-four grid: the cell for consequences 4, likelihood 3 is red. Level of risk: 12, high.
Analysing a risk: two justified estimates, then a reading in the grid set with the context.
  • Exposure: can the asset be reached from the internet, by a supplier, by every employee?
  • Known vulnerabilities: configuration flaws, missing patches, shared accounts, lack of separation.
  • Controls already in place and their real effectiveness, checked rather than assumed.
  • What is happening elsewhere: incidents already suffered by the organisation, by its peers, by its sector. Public reports from cybersecurity agencies are useful here.

Determine the level of risk

The level of risk is derived from the two estimates, using the rule set in the criteria: a reading in a grid, or a calculation. There is nothing to decide at this stage, only a rule to apply. If a result surprises everyone, it is a sign that a rating or a scale definition needs revisiting, not that the level should be corrected by hand.

The analysis covers the situation as it is, with existing controls. Some organisations also estimate a "gross" risk, with no controls at all; this is not an obligation, and the exercise is often artificial. Our article Inherent risk and residual risk reviews these notions.

Running example: analysing Mécanique Vauclair's four risks

Analysis of the risks kept (fictitious example, scales from the "Context" chapter)
Ref.ConsequencesLikelihoodLevel
R1 Ransomware through remote maintenance4: workshop stopped for more than ten days, late-delivery penalties3: permanent remote access, shared account without multi-factor authentication, network not separated12, high
R2 Theft of a customer's drawings4: breach of confidentiality, contract at risk2: email is filtered, but drawings are accessible to the whole design office8, medium
R3 Administration error on the ERP2: two days of re-entry, limited delays3: frequent interventions, no procedure or prior test6, medium
R4 Theft of a laptop1: loss of the hardware, encrypted disk3: frequent travel3, low

These levels do not yet say what to do. That is the purpose of evaluation, which compares them with the acceptance criteria.

Frequently asked questions

How do you calculate a level of risk under ISO 27005?

The standard gives no formula. The level of risk is determined from the assessment of consequences and the assessment of likelihood, using the rule the organisation has set in its criteria: most often a grid that crosses two scales. Annex A of the standard gives examples, without imposing them.

Do you need a quantitative analysis?

No. ISO/IEC 27005 allows qualitative, semi-quantitative and quantitative analysis. A quantitative analysis is only worthwhile if you have reliable data on losses and frequencies. Most organisations start with a qualitative analysis with well-defined levels.

What is the difference between consequence, impact and severity?

They refer to the same idea: what the organisation loses if the risk materialises. ISO/IEC 27005:2022 uses "consequence", the ISO 31000 term. EBIOS Risk Manager speaks of "severity", and common usage of "impact". What matters is to use a single term in your documents.

Should existing controls be taken into account in the analysis?

Yes. The analysis describes the real situation: likelihood and consequences are assessed with the controls already in place, provided you have checked that they work. The level obtained then serves as the reference for measuring the effect of treatment.

Sources

This chapter summarises and rephrases the official publications below. If in doubt, the official text prevails.

A platform and service that adapt to you

Our platform is designed for fine-tuned configuration and broad adaptability to your needs.