Definition
Compliance Exposure Duration (CED) measures the average time, in hours, that non-compliant configurations persist before being detected: specifically, the gap between when a misconfiguration is introduced and when the compliance scanner catches it.
The Measurement Gap
The compliance dashboard showed green across the board. Every system passed. Every control was in compliance. The quarterly report went to the auditors with confidence.
Three days later, a PCI QSA doing a spot check found a misconfigured firewall rule on a cardholder data environment server: one that had been non-compliant for 11 days. The daily compliance scan had never caught it because the scan ran at 2 AM, the configuration change happened at 3 AM, was auto-remediated the next morning before the 2 AM scan, got re-introduced by a different engineer who didn't know the first change had been reversed, and sat there for 10 days before the QSA arrived.
Every scan result was clean. The configuration was non-compliant for 11 days. Both things were true simultaneously.
This is the failure mode that Compliance Exposure Duration (CED) exists to surface: the temporal gap between when a non-compliant configuration is introduced and when it's detected. Configuration compliance tools run on schedules. Regulations care about the actual state of your systems, not just their state at scan time. Those two facts are in constant tension, and your current metrics don't track that tension at all.
The legacy metrics you have (control testing pass rate, remediation SLA adherence, audit finding count) were built for a world where compliance was a periodic event. You prepare for an audit. You test controls. You document findings. You remediate. Everything is framed around the audit cycle. That model worked when configuration changes happened slowly and auditors visited annually.
Cloud environments make hundreds of configuration changes per day. Automated deployments, autoscaling events, network policy updates, security group changes: the environment is in constant motion. A compliance scanner running once a day produces 24 hours of exposure between scans by construction. A scanner running once a week produces up to 168 hours. And those numbers assume the scanner catches the non-compliance the first time it sees it.
Control testing pass rate tells you what percentage of controls passed during the test. It says nothing about the time between tests. Remediation SLA adherence tells you how quickly teams fix findings after they're reported. It says nothing about how long the finding existed before it was detected. Audit finding count tells you how many issues the auditor found. It doesn't tell you how long each of those issues was sitting there before the audit.
The measurement gap is precise: there is no metric that tells you the average duration of non-compliant states between compliance scans. Your compliance dashboard reports point-in-time health. The actual security and regulatory exposure accumulates in the white space between scan events. An organization with a "95% compliance rate" might be non-compliant for 18 hours on average before any non-compliance is detected. Another with the same 95% rate might detect issues within 90 minutes. These are very different risk profiles. Your current metrics cannot tell them apart.
Formula
Compliance Exposure Duration (CED) measures the average time, in hours, that non-compliant configurations persist before being detected: specifically, the gap between when a misconfiguration is introduced and when the compliance scanner catches it.
Formula: Plain Language
For each non-compliance event detected in a measurement period, calculate the time between when the non-compliance was introduced (the configuration change that caused it) and when it was detected (the scan that caught it). Average that time across all events in the period.
Formula: Notation
CED = (1/N) × Σ (T_detect,i − T_introduce,i)
Where:
- N = total non-compliance events detected in the measurement period
- T_detect,i = timestamp when non-compliance event i was detected by the compliance scanner
- T_introduce,i = timestamp of the configuration change that introduced non-compliance i
Units: Hours. Lower is better. A CED of zero would mean real-time detection with no latency: theoretical, not achievable with scheduled scanning.
Worked Example
Your PCI environment processes 15 non-compliance events in October. The time between introduction and detection for each:
- 8 events detected within the same scan cycle: 1, 2, 0.5, 1.5, 2, 3, 1, 2 hours
- 4 events that spanned a scan miss: 26, 27, 25, 48 hours (daily scan; non-compliance introduced shortly after a scan ran)
- 3 events where the configuration change timestamp was unavailable, so the midpoint estimate was used: 12, 15, 11 hours
CED = (1 + 2 + 0.5 + 1.5 + 2 + 3 + 1 + 2 + 26 + 27 + 25 + 48 + 12 + 15 + 11) / 15 = 177 / 15 = 11.8 hoursThat 11.8-hour average is almost certainly longer than your compliance team expected. The monthly report said "15 non-compliances detected and remediated." CED says the average exposure window before detection was nearly 12 hours. For a PCI cardholder data environment, 12 hours is a significant exposure window.
Sign Convention
Lower CED is better. Higher CED means non-compliant states persist longer before detection. There is no upper bound in theory, a configuration that is introduced between scans and then not caught by the next scan due to a scan failure could persist for days.
Edge Cases
Configuration change timestamp unavailable: Use the midpoint between the last clean scan and the detecting scan as the introduction time. This overestimates exposure by approximately 50% on average, which is acceptable: it produces conservative numbers that err toward surfacing risk rather than hiding it. Document which events used estimated timestamps. If estimated timestamps exceed 30% of events, the configuration change logging needs attention before CED numbers are trustworthy. Auto-remediated non-compliances: Include in CED. The exposure existed even if auto-remediation reduced the time-to-fix after detection. CED measures the gap before detection, not the time to fix after detection. Non-compliances that recur: If a non-compliance is detected, remediated, and then the same issue is re-introduced later, count each occurrence as a separate event. Recurring non-compliances are a process problem that deserves separate tracking, but they belong in CED each time they occur. N equals zero: If no non-compliances are detected in the measurement period, CED is undefined. This is worth verifying: it could mean the environment is genuinely clean, or it could mean the scanner isn't running or isn't reaching everything in scope. Check scanner completion logs before celebrating.
Cross-References
Upstream: Change Risk Score Accuracy (CRSA, GMI-IT-012). Changes scored low-risk by the change model receive less scrutiny. A low-scored change that introduces a non-compliance may sit in the exposure window longer because nobody is watching it closely. Strong CRSA supports shorter CED for changes that represent the highest actual risk. Downstream: Baseline Currency Index (BCI, GMI-IT-035). CED is only meaningful if the compliance standard being enforced is current. If the baseline is 18 months stale, CED measures exposure against outdated requirements, a low CED against a stale baseline is false assurance. BCI and CED must be read together. Legacy: Audit finding count. High CED means non-compliant states exist longer before detection, which means more opportunity for an auditor to observe a non-compliant state during a site visit or evidence review. CED improvement doesn't eliminate audit findings, but sustained improvement reduces the probability that findings represent prolonged exposure rather than point-in-time gaps.
Proposed formula, pending validation pilots.