Probability scale
Administrators only: This page describes a setting reserved for organisation administrators.
The probability scale defines the four likelihood levels used to qualify the risks identified during the analysis.
Role of the probability scale
Each risk identified in a project is assessed along two axes: its impact and its probability of occurrence. The probability scale provides the reading grid that frames this second dimension.
The descriptions configured on this page constitute the likelihood reference set of the organisation. They are used by the AI when generating risks and by analysts when positioning a risk in the matrix.
Precise descriptions, drawing on numerical frequencies or concrete time markers, significantly improve the consistency of likelihoods generated by the AI from one project to another.
Scale structure
The scale comprises four levels, each associated with a textual description:
| Level | Expected scope |
|---|---|
| Low | Rare event |
| Medium | Infrequent event |
| High | Regular event |
| Very high | Very frequent or almost certain event |
These default labels can be modified. You can adjust the wording to reflect your organisation's risk culture.
Example description
For the Low level:
Less than one incident every five years.
This format, which expresses probability as an observable frequency, gives the AI a concrete reference point to qualify a risk scenario.
Modify the scale
From the page, you can:
- Modify the description of an existing level to suit your organisation's context
- Align the wording with a reference methodology (EBIOS RM, ISO 27005, etc.)
The changes are taken into account immediately on subsequent AI generations.
Use by the AI
During the risk analysis, the AI consults the probability scale to:
- Assign a likelihood to each generated risk
- Position the risk in the impact × probability matrix
- Prioritise risks and suggest remediation measures consistent with the exposure level
Express each level using a quantitative bound (frequency, annual probability, etc.). Numeric descriptions allow the AI to qualify likelihood much more stably than purely qualitative wording alone.