ISACA CISM: Measuring Security Program Performance
Information security programs can generate thousands of metrics without proving that they reduce business risk. Counting blocked attacks, completed training sessions or devices running an endpoint agent describes activity; it does not necessarily show whether critical services are becoming safer. An ISACA CISM security manager needs measures that connect program objectives, control performance, residual risk and executive decisions.
On this page
As of October 25, 2026, CISM’s existing outline includes information security program management as a major domain. ISACA has announced a new outline effective November 3, 2026. The basic management discipline is stable across this transition, but exam candidates should verify the version applicable on their sitting date instead of treating future domain weighting as already effective.
Start with outcomes and risk hypotheses
Metrics become useful when they answer a decision. If the goal is to limit account takeover, useful questions include how quickly suspicious privileged sessions are detected, how consistently privileged access is reviewed, how many unmanaged high-impact credentials remain and what business systems those credentials could reach. A metric should point toward action when it changes.
Link each measure to a risk hypothesis. For example: “If high-impact internet-facing vulnerabilities remain unremediated beyond our agreed response window, customer service disruption becomes more likely.” From that hypothesis, select exposure-aware measures, such as the number of exploitable vulnerabilities on critical systems exceeding their risk-accepted remediation windows.
Differentiate coverage, effectiveness and outcomes
Control coverage answers where a safeguard is deployed. Effectiveness answers whether it works under realistic conditions. Outcomes describe what happened to the risk or business process. Ninety-nine percent multifactor-authentication enrollment may look excellent while one unprotected emergency administrator account retains access to all production data.
Test the control from the perspective of failure. Can a phishing-resistant method resist the simulated attack? Can an endpoint alert be triaged before an attacker changes course? Does a backup restoration actually meet the system’s recovery time and integrity requirements? Coverage percentages become more meaningful when paired with criticality and test evidence.
Use leading indicators and lagging indicators together
Lagging indicators, such as confirmed security incidents, financial loss and recovery duration, show realized consequences. Leading indicators, such as overdue privileged-access reviews, unresolved critical exposure and failing recovery tests, may reveal deteriorating conditions before an incident occurs. Neither category is sufficient alone.
A low incident count can mean strong controls, weak detection or simple luck. A high number of detected phishing attempts can mean improved detection, greater targeting or a changed logging rule. Always document the denominator, collection method and interpretation limits. Metrics should not reward a team for suppressing reports or downgrading severity to improve a dashboard.
Build a decision-oriented executive scorecard
A practical scorecard can organize risk by critical business service, control objective, current residual exposure, trend, accountable owner and required decision. Use a small number of coherent indicators rather than presenting every alert produced by security tooling. Executives need to know whether the risk exceeds approved appetite, whether planned mitigation is working and what resources or risk acceptance are requested.
For example, a payments platform might report the percentage of privileged identities protected by a strong authentication control, backup restore test results, high-risk supplier dependencies and unresolved attack paths into sensitive systems. Present the impact of an exception: an unprotected support account tied to payment production deserves more attention than dozens of harmless configuration variations on a retired test system.
Measure supplier and process risk without false precision
Security programs increasingly depend on external cloud, identity, software and managed-service providers. A vendor’s certification or questionnaire score does not eliminate integration and concentration risk. Metrics should track the importance of the outsourced function, agreed controls, evidenced testing, open remediation items and whether exit or contingency plans can work.
Risk scoring itself has limitations. A single red-amber-green label can obscure the size of a loss or the uncertainty of an estimate. When quantitative measures are available, state underlying assumptions; when only qualitative judgments are defensible, disclose why. Security managers should not manufacture precise percentages of “risk eliminated” from arbitrary checklist scores.
Make measurement change the program
Review trends with the people who can act on them. Repeated recovery-test failures may justify redesigning backups or scheduling a cross-functional exercise. Rising late patch rates may reflect resource shortages or difficult legacy dependencies rather than employee negligence. The correct response could be compensating controls, architecture change or an explicit risk acceptance with an expiration date.
Validate metrics when tools, assets or reporting thresholds change. A new scanner can double the number of findings without any change in exposure. Establish versioned definitions and an audit trail of metric transformations so management can compare periods meaningfully.
For CISM reasoning, a strong metric is one that makes risk ownership visible and enables a justified choice. The program should know what it is protecting, how its controls fail, what evidence supports the conclusion and which decision follows.