There is an old management adage: "What gets measured gets managed."
In the world of industrial safety, however, we often measure the wrong things. For decades, organizations have relied on Lagging Indicators—metrics that tell us what went wrong after someone has already been hurt.
To build truly resilient systems, engineering leaders must shift their focus to Leading Indicators and Functional Safety Metrics. These are the numbers that predict performance and prevent incidents before they occur.
Here is a guide to the safety metrics that actually matter.
1. The Trap of Lagging Indicators (TRIR & DART)
Most executive dashboards feature the Total Recordable Incident Rate (TRIR) or Days Away, Restricted or Transferred (DART).
While these are required by regulators (like OSHA), relying on them to manage safety is like driving a car by only looking in the rearview mirror.
The Problem: A low TRIR can create a false sense of security. You might have zero injuries for a year simply due to luck, while critical risks (like a bypassed interlock) sit silently waiting to cause a catastrophic event.
The Takeaway: Keep tracking them for compliance, but do not use them to judge the health of your safety program.
2. Leading Indicators: Measuring Prevention
Leading indicators measure the activities that prevent accidents. They answer the question: "Are we doing the work required to stay safe?"
Near Miss Reporting Rate: A high number here is actually good. It means your culture is healthy enough that people report close calls without fear of blame.
Audit Closure Velocity: When a safety audit finds a gap, how many days does it take to fix it? This measures your organization's responsiveness.
Preventive Maintenance Compliance: Are safety-critical components (e.g., light curtains, brake pads) being swapped out on schedule, or are maintenance tickets piling up?
3. Functional Safety Metrics (The Engineering View)
For teams building robots, autonomous vehicles, or machinery, "safety" isn't just about PPE; it's about system reliability. This requires specific engineering metrics:
Traceability Coverage (%)
In the V-Model, every safety requirement must have a corresponding validation test.
The Metric: Percentage of requirements linked to a passed test case.
The Goal: 100%. Anything less means you have "orphan" requirements that might not work when needed.
Risk Burndown
During the design phase, you identify hundreds of risks.
The Metric: The rate at which these risks are mitigated (moved from "High" to "Low" via design controls).
The Insight: If your risk burndown line is flatlining, your product launch is in jeopardy.
Probability of Failure per Hour (PFH)
This is the "Gold Standard" for hardware reliability.
The Metric: The statistical probability that a safety function (like an E-Stop) will fail dangerously.
The Goal: To stay within the "Budget" for your target SIL level (e.g., $10^{-7}$ for SIL 3).
4. The "Culture" Metric
How do you measure a feeling? While difficult, you can quantify safety culture through Engagement.
Participation Rate: What percentage of the engineering team actively contributes to the Hazard Analysis (HARA)? If it's only the Safety Manager, you have a silo problem.
Stop-Work Authority Usage: How often do engineers or operators hit the "Red Button" to stop a process because they feel unsafe? A non-zero number is healthy—it means they feel empowered.
Summary
Great safety performance isn't about achieving "Zero Injuries" on a scoreboard; it's about achieving 100% Confidence in your system design and processes.
By shifting your focus from counting bandages (TRIR) to verifying logic (Traceability and Reliability), you move from a reactive posture to a proactive one—building systems that are safe by design, not just by luck.