top of page

Probability of Action (PoA) Rubric

  • Writer: FAIR INTEL
    FAIR INTEL
  • 6 days ago
  • 4 min read

Overview

Probability of Action (PoA) measures the likelihood that a threat actor will act once contact with a target occurs. In FAIR methodology, PoA captures the threat actor's motivation and commitment to follow through on an attack opportunity.

PoA is expressed as a decimal between 0.00 and 1.00 and is combined with Contact Frequency (CF) to calculate Threat Event Frequency (TEF):

TEF = CF × PoA

PoA Estimation Context

Inferred vs. Observed Probability of Action

This rubric estimates PoA based on threat intelligence reporting, not direct observation. PoA is inferred from:

  • Actor's historical behavior and follow-through patterns

  • Stated or implied motivations

  • Resource investment in campaign infrastructure

  • Target alignment with actor's known objectives


Disclaimer

The PoA values in this analysis are inferred from threat intelligence reporting and represent analyst judgment based on observable indicators. Actual probability of action may vary based on factors not visible in reporting.


Probability of Action Scale

Scale Design: Equal Intervals

The PoA scale uses equal 0.20 intervals for consistency and simplicity:

Tier

Range

Meaning

Very High

0.80-1.00

Near certainty of action; actor is highly motivated and committed

High

0.60-0.80

Strong likelihood of action; actor has clear motivation and demonstrated commitment

Moderate

0.40-0.60

Uncertain; actor may or may not act depending on circumstances

Low

0.20-0.40

Action unlikely unless conditions are favorable; actor is opportunistic or distracted

Very Low

0.00-0.20

Action rare; actor has minimal motivation or competing priorities

Why Equal Intervals?

Unlike CF (which uses calendar-anchored cutoffs), PoA represents a probability. Equal intervals:

  • Are intuitive to interpret (each tier spans 20 percentage points)

  • Avoid arbitrary cutoffs

  • Align with common probability language (e.g., "60-80% likely" = High)


Probability of Action Rubric

Tier

Range

Observable Criteria

Very High

0.80-1.00

State-sponsored actor with strategic mandate. Clear financial monetization path with proven success. Target exactly matches known victimology. Significant sunk cost in campaign (custom tooling, infrastructure). Actor has history of persistent follow-through. No indication of competing priorities.

High

0.60-0.80

Established actor with consistent operational history. Clear motivation (financial, espionage, hacktivism). Target aligns with known targeting patterns. Dedicated campaign infrastructure observed. Actor demonstrates adaptability when initial attempts fail.

Moderate

0.40-0.60

Actor motivation is mixed or unclear. Target partially aligns with known patterns. Campaign infrastructure is shared or minimal. Actor may pursue easier targets if resistance encountered. Some history of abandoned campaigns.

Low

0.20-0.40

Opportunistic actor without clear strategic objectives. Target is peripheral to main victimology. Minimal campaign investment observed. Actor known to shift focus frequently. Action depends heavily on target vulnerability.

Very Low

0.00-0.20

Actor has no apparent motivation for this target type. Target does not match any known patterns. No dedicated infrastructure. Actor is dormant, disrupted, or focused elsewhere. Action would only occur through error or accident.

Evidence Mapping Guide

Use the following to map article evidence to the appropriate tier.

Motivation

Observable Evidence

Typical Tier

State directive or strategic espionage mandate

Very High

Proven financial model (ransomware, fraud, data sales)

Very High

Clear ideological or hacktivist objective

High

Financial motivation without proven monetization

Moderate

Unclear or mixed motivations

Moderate

Opportunistic, no clear objective

Low

No apparent motivation for target type

Very Low

Operational Commitment

Observable Evidence

Typical Tier

Multi-year campaigns against same target set

Very High

Persistent retargeting after initial failure

Very High

Sustained campaigns with regular activity

High

Adapts TTPs in response to defenses

High

Single campaign or limited duration

Moderate

Abandons campaigns when resistance encountered

Low

Sporadic activity, easily deterred

Low

No sustained operational history

Very Low

Resource Investment

Observable Evidence

Typical Tier

Custom malware, zero-days, dedicated infrastructure

Very High

Multiple custom tools, sustained infrastructure

High

Mix of custom and public tools

Moderate

Primarily public tools, minimal infrastructure

Low

Off-the-shelf tools only, free infrastructure

Very Low

Target Alignment

Observable Evidence

Typical Tier

Target exactly matches known victimology

Very High

Target aligns with sector and geographic focus

High

Target partially matches (sector OR geography)

Moderate

Target is adjacent to known targeting

Low

Target does not match any known patterns

Very Low

Determining the Tier

  1. Map article evidence to each of the four categories above

  2. Identify the tier that appears most frequently across categories

  3. If evidence is mixed (e.g., two High, two Moderate), select the tier that best reflects overall threat posture and document the uncertainty

  4. Select the full range for that tier—do not pick a point estimate


Endpoint Justification

Each PoA estimate must justify what conditions push toward the low versus high end of the selected tier range.

Factors That Push Toward Low End of Range

  • Target is lower-value within the profile

  • Actor has competing campaign priorities

  • Limited sensitive data exposure at target

  • Actor has shown inconsistent follow-through

  • Target has demonstrated strong defenses

Factors That Push Toward High End of Range

  • Target is high-value within the profile

  • Target aligns with strategic objectives

  • Target holds sensitive data of interest

  • Actor has persistent operational history

  • Target has weak or unknown defenses


Handling Insufficient Evidence

When threat intelligence does not provide sufficient evidence to map to rubric criteria:

Step

Action

1

Identify which rubric criteria cannot be assessed

2

Document the specific evidence gap

3

Default to Moderate tier as baseline

4

Mark estimate as [LOW CONFIDENCE]

5

Note impact on analysis reliability

6

Recommend update when additional intelligence becomes available

Why Default to Moderate?

  • Moderate represents the mathematical center of the scale

  • Avoids overstating (High/Very High) or understating (Low/Very Low) risk

  • Provides a consistent, repeatable baseline across analyses

  • Allows calculations to proceed while flagging uncertainty

When NOT to Default

Do not default to Moderate if:

  • Evidence explicitly indicates a different tier (even if incomplete)

  • Partial evidence strongly suggests High or Low

  • The analysis would be misleading with a Moderate estimate

In these cases, use the best available evidence and document the uncertainty.


Scale Limitations

1. PoA is Not Directly Observable

Unlike technical indicators, probability of action cannot be measured directly. It is inferred from:

  • Historical behavior patterns

  • Observed resource investment

  • Target alignment analysis

  • Attribution confidence

Different analysts may reach different conclusions from the same evidence.

2. Motivation is Often Unknown

Threat intelligence rarely provides definitive insight into actor motivation. Analysts must infer motivation from:

  • Targeting patterns

  • Data exfiltration behavior

  • Monetization attempts

  • Attribution to known groups

Low-confidence attribution weakens PoA estimates.

3. PoA Can Change Rapidly

Unlike CF (which reflects operational tempo), PoA can shift based on:

  • Geopolitical events

  • Law enforcement actions

  • Target defensive changes

  • Actor leadership changes

PoA represents a point-in-time estimate.

4. Ranges Reflect Uncertainty

The ranges (e.g., 0.60-0.80) represent analyst uncertainty, not variability in actor behavior. Use the full range in calculations.


Version History

Version

Date

Changes

3.0

January 2026

Revision


Comments


bottom of page