The three numbers: definitions
Full definitions of Days at Risk, Time to Good Decision and Ambient Breach: how each is read, what it isn't, where it sits in the shared library, and the reference points that apply.
1Days at Risk
Definition. Days at Risk counts how long each known exposure stays unmitigated, starting at its earliest supported awareness record.
How it's read
- A plain count of days per known risk, shown as a distribution, never as one average.
- The clock starts on the earliest day you can show the risk applied to a system someone is accountable for. Not the day someone got around to acknowledging it, and not just the day an advisory was published.
- If the start can't be shown, it's marked unknown, and the item is counted separately rather than dropped.
What it isn't
Not a risk score, and not likelihood × impact. A slow acknowledgment can't make the number smaller, and accepting a risk doesn't stop its clock.
Where to investigate
Read beside the maturity profile for the same outcome. Check Bounded work (may anyone fix it?) and Loop (is the fix verified?).
In the library
Related library reference: Security Posture family · KPI-SP-07 Related search term: vulnerabilities past due (a reading of Days at Risk, not a separate measure).
2Time to Good Decision
Definition. Time to Good Decision measures the time from the first anomalous signal to a recorded decision that holds through its validation window.
How it's read
- Per incident class (DNS, identity, phishing, outage and so on): choose a target time to decide and a separate validation window. Compare the elapsed-time P90 with the response target; show reversed and open cases alongside it.
- "Good" is a test: the decision held through its validation window. It is not proof the decision was substantively right.
- Reversed and still-open cases are shown beside the P90, with both the response target and the validation window stated, so leaving them out can't make the number look fast.
What it isn't
Not acknowledgment time and not recovery time. A reversed decision fails the test, but the incident doesn't disappear from the record.
Where to investigate
Read beside the maturity profile for the same outcome. Check Detection (is it noticed, and has that been tested?) and Coverage (is there anything to decide on?).
In the library
KPI-DR-01 · F15 Related search terms: MTTD, MTTI, mean time to respond. Related, not equivalent: a mean is not a P90, and those clocks stop at different events.
3Ambient Breach
Definition. Ambient Breach warns when uncertainty affecting decisions arrives faster than the service's demonstrated repair capacity can safely absorb.
How it's read
- The simple reading is a net rate of unresolved work: new gaps, contradictions, unexplained changes and unverified actions, minus repairs that are verified, per unit of time and per unit type.
- Count one type of work unit over the same reporting interval on both sides.
- The fuller form, from the Kinetic Trust Protocol, compares arrival against the demonstrated maximum durable repair capacity, with a reserve for how critical the area is. A flat backlog doesn't prove that reserve is safe.
What it isn't
Not a statement that a breach has happened. It measures whether ambiguity is being absorbed.
Where to investigate
Read beside the maturity profile for the same outcome. Check Coverage (blind areas) and Loop (repairs that are never verified).
In the library
F24 · KPI-AB-01 … KPI-AB-10 · KPI-AB-03 (net rate)
4Reference points
Use an applicable, comparable, sourced benchmark where one exists. Where none exists, say so and set a local target with the evidence it needs.
| Measure | Reference point | Applies when | Source |
|---|---|---|---|
| Days at Risk | 3 calendar days when attackers could automate the attack or it gives them total control (with a forensic triage of the asset where it gives total control); 14 calendar days otherwise. This deadline clock starts at the earlier of two events: CISA adds the vulnerability to the KEV catalog, or the agency enumerates or identifies it on an asset (under BOD 23-01) and updates the CDM Program Agency and Federal Dashboard (Appendix A, meaning (f)). It is distinct from Days at Risk's awareness clock, and timing is not the only required action. | The system is publicly exposed and the vulnerability is in CISA's Known Exploited Vulnerabilities (KEV) catalog. Binding on federal civilian executive branch systems; a useful screen elsewhere, not a compliance check. Full requirements: the directive and its implementation guidance. | CISA Binding Operational Directive 26-04, Appendix A, Table 1 (June 10, 2026) |
| Time to Good Decision | No applicable benchmark is established here. Set a response target and a separate validation window per incident class; compare the P90 time to decide with the target. | For context only, a different clock: in incidents Mandiant investigated in 2024, the global median dwell time was 11 days. Dwell time ends at detection, not at a decision that holds. | Mandiant M-Trends 2025 |
| Ambient Breach | No applicable benchmark is established here. Compare what arrives with what is verifiably repaired, week to week. | Where arrivals and verified repairs are measured for the same unit type and reporting interval; otherwise the reading is unknown. A flat backlog does not prove reserve. | None |
BOD 26-04 revoked BOD 19-02 and BOD 22-01 (directive, Background). Its timelines are in calendar days (Appendix A, meaning (a)). The journey on the home page uses its first four Table 1 rows; if it's unclear whether the 3-day case applies, the journey screens at 3 days, which is conservative, not a legal classification.
5Rules for all three
- Each answers a different question. None sums up the others, and they're never combined into one score.
- Related measures elsewhere are related, not equivalent. Being related doesn't license comparing or ranking them.
- Show the distribution and the cases left out (unknown start, reversed, still open), not only the headline figure.
- The numbers show whether things are improving; the maturity profile shows where to look.
6Sources
- Protecting Your Organization by Measuring Days at Risk (2025). The origin; its composite score is historical, not today's definition.
- Days at Risk: How Complacency Fuels Catastrophic Outcomes (2025). Background.
- Time to Good Decision (2026)
- Ambient Breach (Kinetic Trust Protocol research)
- CISA, Binding Operational Directive 26-04: Prioritizing Security Updates Based on Risk (June 10, 2026)
- Mandiant, M-Trends 2025