A control room can report that every officer completed training, every incident was logged and every inspection was signed off. None of that tells you whether the team will recognise a developing threat, make sound decisions or manage the first critical minutes of an incident. Security performance metrics should expose that difference.
The problem is not a lack of data. Most organisations collect plenty of it. The problem is that much of it measures administrative activity rather than operational capability. A monthly dashboard may look reassuring while the underlying security posture remains weak.
That matters more as organisations prepare for risks that demand judgement, coordination and confident action. Martyn’s Law has sharpened attention on protective security in the UK, but compliance alone will not make a venue, site or organisation ready. Readiness is demonstrated through competent people, workable arrangements and evidence that both perform when conditions are not ideal.
What security performance metrics should measure
A useful metric helps someone make a better decision. It identifies a weakness early enough to correct it, confirms that an improvement has worked, or shows that a risk is increasing. If it does none of those things, it may be a reportable figure, but it is not a useful performance measure.
Start with the security outcomes that matter in your environment. A transport operator may need to measure the quality and speed of escalation from frontline staff. A retailer may need to know whether managers can identify suspicious behaviour and preserve business continuity during disruption. A corporate site may need assurance that access control failures are identified, owned and resolved before they become normalised.
Those outcomes are different, but the question is consistent: can people detect, decide, communicate and act effectively?
Many organisations default to easy measures because they are available. Training completion rates, patrol numbers, visitor volumes and incident totals all have value in context. They become misleading when treated as proof of capability. A 100 per cent training completion rate may mean only that people watched a module. It does not show what they retained, whether they can apply it, or whether a supervisor would trust their judgement under pressure.
A better approach combines measures of activity, quality and effect. For example, it is reasonable to track whether security staff have completed relevant learning. Pair that with a capability assessment, scenario based observation and a review of decision quality in exercises or real incidents. The combined picture is harder to manipulate and much closer to reality.
Separate assurance from capability
Assurance has a clear place. You need to know that critical checks happen, systems are maintained, policies are current and incidents are recorded. These are basic controls. Ignoring them creates avoidable exposure.
But assurance is not capability. A documented evacuation plan cannot direct people. A risk assessment cannot spot abnormal behaviour. A well written procedure cannot resolve confusion between security, operations and senior management when the situation is moving quickly.
This distinction should shape the scorecard. Measures of assurance should answer whether essential arrangements exist and are being maintained. Measures of capability should answer whether the organisation can use those arrangements effectively.
Consider an access control issue at a busy site. An assurance measure may record the percentage of faulty doors repaired within the agreed time. That is useful. A capability measure goes further: how often did staff identify the fault before it created an unmanaged entry point, apply appropriate interim controls, report it clearly and verify that the repair resolved the problem? The first metric measures supplier or maintenance performance. The second measures security judgement and ownership.
Both are needed. Treating either one as sufficient creates a blind spot.
Use leading indicators before failure teaches the lesson
Incident data is often given too much weight because it is concrete. It tells you what happened, where and when. It can reveal patterns, but it is largely backward looking. Low incident numbers can mean good security, low reporting confidence, poor recording discipline or simply good fortune.
Leading indicators provide earlier warning. They show whether the conditions for effective performance are present before an incident forces the issue. Examples include the quality of supervisor briefings, completion of corrective actions after exercises, the time taken to communicate a credible concern, and the proportion of staff who demonstrate the required standard in practical assessments.
The word quality matters. Counting briefings tells you that they occurred. Sampling them can show whether staff understand the relevant threat picture, local vulnerabilities and their individual responsibilities. Counting corrective actions can create another paper chase. Testing whether the action changed behaviour is the point.
One useful test is to ask whether an experienced manager could explain why a figure changed and what they will do next. If the answer is no, the metric is probably too detached from operational reality.
Measure decisions, not just response times
Response time is a familiar metric, and it has a role. Delayed escalation or a slow emergency response can increase harm. Yet speed without judgement is not performance.
A security officer who rapidly reports every uncertain observation may flood the system with poor quality information. A manager who closes an incident quickly may appear efficient while missing the need for further review. Equally, a team that takes slightly longer to assess a concern may make better decisions because roles, thresholds and communication routes are clear.
Measure timeliness alongside decision quality. In exercises, assess whether people recognised the issue, selected proportionate actions, communicated useful information and maintained command of their normal responsibilities. In real incidents, conduct a structured review that examines what was known at the time, not just what became obvious afterwards.
This is particularly important for behavioural risk. Staff need confidence to report concerns, but they also need the ability to describe what they observed without making assumptions or labelling people carelessly. The measure should reflect the quality of observation, reporting and escalation, not a crude target for the number of reports submitted.
Build a scorecard people cannot game
Every metric changes behaviour. If a target rewards volume, people will produce volume. If it rewards rapid closure, difficult issues may be closed too early. If it rewards zero incidents, reporting may quietly decline.
This is why a balanced scorecard is safer than a single headline figure. It should contain a small number of connected measures across four areas:
- control reliability, such as critical maintenance, checks and corrective action completion
- capability, shown through assessments, exercises and observed performance
- decision and response quality, including escalation, communication and command
- learning, demonstrated by recurring issue reduction and improvements that remain in place
The measures should be reviewed together. A rise in reported concerns may be positive if staff confidence and report quality have improved. A reduction in incidents may be positive, or it may require scrutiny if reporting, inspections or staff engagement have also fallen.
Set baselines before declaring success. Without a clear starting point, targets become arbitrary and progress is difficult to defend. Baselines should include more than numbers. Record recurring issues, common decision failures, unclear responsibilities and friction between teams. These are often where performance breaks down.
Give metrics an owner and a consequence
A dashboard without ownership is decoration. Each measure needs someone who understands the operational process behind it, can investigate adverse movement and has the authority to drive corrective action. That may be a security manager, but not always. Facilities may own system faults, operations may own staff availability, and senior leaders may own the decisions that determine whether known vulnerabilities are funded and fixed.
Review cycles should match the risk. A critical access issue cannot wait for a quarterly meeting. Broader capability trends may need monthly review and deeper assessment after an exercise, significant incident or organisational change.
The final discipline is to close the loop. When a metric identifies a weakness, define the action, test the change and measure again. If an exercise reveals poor escalation, do not simply issue a reminder. Clarify thresholds, practise the communication route, assess the people involved and check whether performance improves next time.
Mildot Group’s capability evaluations are designed around this principle: immediate feedback has value when it directs meaningful development, not when it merely produces a score. The aim is not to make the dashboard greener. It is to make decisions better when it matters.
The uncomfortable question for any security leader is simple: if the figures improved next month, would the organisation genuinely be safer, or would it only be better at reporting itself? The answer should determine what you measure next.
