A security coverage metric measures how much of the environment a control actually reaches: the share of systems sending logs, the share of endpoints visible to the inventory tool, the share of eligible assets with application allowlisting installed. Coverage metrics exist because “we have a SIEM” and “every system logs to the SIEM” are different claims, and most incident investigations discover the gap between them too late.
This article explains why coverage belongs next to the familiar response metrics, how the Center for Internet Security’s Controls Assessment Specification defines it, and how to calculate the three coverage metrics that expose the most common blind spots.
Why coverage, not just speed
Mean time to detect and mean time to resolve are the security operations metrics everyone reports. Both have a blind spot: they can only count incidents you found. A team with 40 percent audit log coverage can post a fast MTTD on the incidents its logs happened to catch while missing the rest entirely. Coverage is the denominator that makes the speed metrics honest.
Coverage metrics are also the first thing an investigator needs. When a credential leaks, the question is “which systems did it touch,” and the answer depends on whether the inventory is complete and the logs exist. Low coverage does not just slow detection; it turns known questions into unanswerable ones.
How the CIS Controls Assessment Specification defines measurement
CIS publishes a Controls Assessment Specification alongside the CIS Controls. For each safeguard it defines inputs (the data you need), operations (what you do to the inputs), measures (the counts that result), and metrics (the formula that turns measures into a number). It focuses on whether a safeguard has been implemented across the environment, which is exactly what coverage means, and leaves “how well” to the separate CIS Controls Self Assessment Tool.
That structure matters because it forces you to name the denominator. Almost every coverage metric is a fraction, and the hard part is the bottom of it. You cannot calculate the share of systems sending logs without an authoritative list of systems, which is why asset inventory coverage comes first.
Three coverage metrics and how to calculate them
| Metric | CIS safeguard | Formula | What a low number means |
|---|---|---|---|
| Software inventory tool coverage | 2.1 Establish and maintain a software inventory (IG1) | Endpoints reporting to the inventory tool ÷ endpoints in the enterprise asset inventory | Devices you cannot see, patch, or scope during an incident |
| Audit log coverage | 8.2 Collect audit logs (IG1) | Systems and service providers configured to collect the required logs ÷ systems and service providers in scope | Investigations that end at “no data” and incidents that are never detected |
| Application allowlisting installation coverage | 2.5 Allowlist authorized software (IG2) | Capable assets with an allowlisting control installed ÷ capable assets | Endpoints where unauthorized software runs unchallenged |
Two companion metrics from the same family close the loop. Standard configuration coverage, from safeguard 4.1, measures the share of authorized software and systems with a documented, maintained secure configuration standard. Unauthenticated penetration testing coverage, from Control 18, measures the share of applications that have been tested from the outside. Both answer the same question as the first three: of everything we are responsible for, how much has the control actually touched?
Where the denominators come from
- Start with the enterprise asset inventory (CIS safeguard 1.1). If it does not exist, the first metric to build is inventory coverage against a discovery scan, and every other metric waits.
- Define “in scope” per metric. Audit log coverage should include cloud accounts and service providers, not just servers. Allowlisting coverage should divide by assets capable of running the control, so that a printer does not count against you.
- Reconcile sources. Inventory coverage is the inventory tool’s list divided by the asset inventory. If the two lists were built from each other, the metric is 100 percent by construction and measures nothing.
- Record the exclusions. An asset excluded from a denominator needs a documented reason and an owner, or the metric drifts upward as people exclude whatever is inconvenient.
Using coverage metrics with incidents
The case study linked below applied ten KPIs to incidents publicly reported in 2026. The coverage metrics were the ones that changed how each incident read. For a reported breach of a developer platform, audit log coverage was the metric that would have turned missing telemetry into a known investigation gap before the incident instead of during it. For a reported campaign against firewall appliances, software inventory coverage would not have stopped the attack, but complete visibility would have let the team find every affected device and its owner without hunting for forgotten assets.
The analysis also noted the limits. Allowlisting coverage shows whether the control is installed, not whether trusted software has been compromised upstream; for that, it has to be combined with integrity checks, monitoring, and vendor controls. A coverage metric is a precondition for other controls working, not a substitute for them.
What to report
- The number and the denominator, every time. “Audit log coverage: 212 of 260 in-scope systems, 82 percent.”
- The trend, by quarter. Coverage should only go up; a drop usually means new assets were added without the control.
- The gap list. Coverage below 100 percent is a list of named systems. Attach it, with owners.
- Coverage next to speed. Report MTTD alongside audit log coverage so that a fast detection number cannot be read as complete detection.
See it applied
- Security KPI and Metrics AnalysisMy case study: ten KPIs, including the five coverage metrics, each applied to a publicly reported 2026 incident.
- MTTD vs MTTRThe speed metrics that coverage makes honest.
- CIS Controls v8 Gap AnalysisAn incident in which the missing asset inventory was the first link in the chain.
Frequently asked questions
- What is a security coverage metric?
- A measure of how much of the environment a security control actually reaches, expressed as a fraction: assets where the control is in place divided by assets in scope. Examples include audit log coverage, inventory coverage, and allowlisting coverage.
- What is audit log coverage?
- The share of in-scope systems and service providers configured to collect the required audit logs. It corresponds to CIS Controls v8 safeguard 8.2 and determines whether incidents can be detected and investigated.
- Why do coverage metrics matter more than MTTD?
- They do not replace it; they qualify it. MTTD only counts incidents that were detected. Coverage tells you what share of the environment detection could have reached, so the two should be reported together.
- Where are coverage metrics defined?
- The CIS Controls Assessment Specification defines inputs, operations, measures, and metrics for each CIS safeguard. Coverage-style metrics are those that divide assets with a safeguard implemented by assets in scope.