Three numbers
Track how long exposure lasts, how long decisions take, and whether verified repair keeps pace. Each answers a different question. They're never added into one score.
Days at Risk
How long does known exposure stay open?
Count how long each known exposure remains unmitigated, including exposures still open, from its earliest supported awareness record.
Where the next hour of fixing goes, and which fixes are stuck between teams.
One list of the services or exposures you already track.
For internet-facing systems with a vulnerability in CISA's Known Exploited Vulnerabilities catalog: 3 days if the attack could be automated or gives total control, otherwise 14 days, in calendar days. That federal deadline clock has its own start rule (see the reference page); it isn't the same as Days at Risk's awareness clock, and timing isn't the only required action. Source: CISA Binding Operational Directive 26-04 (2026), binding on U.S. federal civilian executive branch systems; elsewhere a screen, not a compliance check.
Shown as a distribution, never one average. Unknown starts stay visible; accepting a risk does not stop its clock.
Time to Good Decision
How long do decisions take?
Time from the first anomalous signal to a recorded decision that holds through its validation window.
Whether the response process is fast enough, and where it stalls.
One incident class, a target time to decide, and a separate window the decision must hold through.
No applicable benchmark is established here. Compare the 90th-percentile time to decide with your response target, and check separately that decisions held through their validation window.
Reversed and still-open cases are shown beside the figure. A decision that held isn't proof it was right.
Ambient Breach
Is verified repair keeping pace?
A warning when new uncertainty (gaps, contradictions, unverified actions) arrives faster than the team can verifiably repair it.
Whether the team is keeping up, and whether to add staff, automation or escalation.
One type of work, counted the same way on both sides over the same interval.
No applicable benchmark is established here. Compare what arrives with what is verifiably repaired, week to week, for the same unit type and interval.
Not a statement that a breach has happened. A flat backlog doesn't prove there is reserve.
Read the numbers beside the maturity profile for the same outcome. The numbers show whether things are improving; the profile shows where to look.
