Measuring outcomes
Leading indicators for EHS dashboards: which ones are worth tracking
Which leading indicators belong on an EHS dashboard, which to drop, how to pair them with lagging measures, and how to use AI and sensor data without fooling yourself.
By LIPAI WANG · Updated · 10 min read · 12 sources
Reviewed by LIPAI WANG, Editor, on against the cited sources. How we review
A leading indicator earns its place on an EHS dashboard when it ties to a specific hazard or critical control, measures against a stated denominator, and triggers a decision when the number moves. In practice, this usually means tracking measures of follow-through (are hazards being fixed, and how fast?) and control health (are critical controls checked, and do they work?). These belong next to the usual lagging injury rates. Activity counts, raw alert totals, and "days without injury" boards rarely justify their spot.
What is a leading indicator, and why do dashboards need them?
OSHA describes leading indicators as proactive and preventive measures that show how well safety and health activities are working and reveal potential problems in a program, and lagging indicators as measures of events that have already happened, such as the number or rate of injuries, illnesses and fatalities [1]. Its short version is that a good program "uses leading indicators to drive change and lagging indicators to measure effectiveness" [1].
Dashboards need both because lagging indicators arrive late and in small numbers. US private industry recorded 2.3 recordable cases per 100 full-time equivalent workers in 2024 [10]. A 200-person site at that rate would expect around four or five recordables a year, too few to show a monthly trend and easily moved by chance. Sites that pick their worst period as a starting point will also tend to improve anyway, a statistical effect called regression to the mean [11].
But there is a harder lesson. The US Chemical Safety Board found that at BP's Texas City refinery, where an explosion killed 15 workers in 2005, a very low personal injury rate gave a misleading picture of process safety performance [4]. A dashboard built entirely on injury rates can look perfectly green while the controls that prevent serious incidents decay in the background.
ISO 45001 requires organizations to monitor, measure, analyze and evaluate health and safety performance [12], but it leaves the choice of measures up to you. That selection process is where most dashboards fail.
What makes a leading indicator worth tracking?
OSHA's 2019 guide says good leading indicators follow SMART principles, which it defines as specific, measurable, accountable, reasonable and timely [2]. That provides a helpful baseline. For a dashboard, put any proposed metric through four more tests:
- Is it tied to a hazard or control? OSHA's guide describes three ways to build indicators: from data you already collect, for controlling an identified hazard, and for improving an element of your safety program [2]. The second route produces the most useful dashboard measures, because you know what risk the number is meant to reflect.
- Does it have a denominator? "42 hazard reports" means nothing without hours worked, headcount, vehicle movements or camera-hours.
- Does someone act when it moves? Name the owner and the threshold that triggers a response. If the honest answer is "we would note it," leave it off.
- Can it be checked independently? If the only source is the system being evaluated, plan a periodic manual check.
The published research supports taking a cautious approach. A 2023 synthesis of 93 publications found that the definition, application and function of leading indicators are mostly ambiguous and inconsistent across the safety literature [6]. And a 2025 scoping review of 48 studies found that most reported a positive effect of leading indicators on lagging outcomes, but rated the evidence base as weak: designs were not suited to showing causation, internal validity was moderate to low, and no two studies answered the same question [5]. Those reviewers concluded that the choice of indicators should rest on evidence that they improve safety [5]. Until that evidence exists for your specific context, treat every indicator as a working hypothesis.
The dual assurance idea: one leading and one lagging measure per control
HSE's guide for major hazard sites, HSG254, introduced what it calls "dual assurance": set a leading and a lagging indicator for each critical risk control system [3]. A leading indicator acts as a routine check to ensure key activities happen as intended, catching holes in the control before anything goes wrong. The lagging indicator then records failures of that control when they occur, which includes near misses and precursor events rather than just injuries [3]. HSG254 also warns that audits can be too infrequent to detect rapid change and may confirm that a system exists rather than that it delivers the intended outcome [3].
The approach was written for process safety, but it transfers well to any site where a few hazards dominate the risk. For a warehouse, a dual assurance pair for traffic management might look like this:
| Critical control | Leading indicator (is the control working?) | Lagging indicator (has it failed?) |
|---|---|---|
| Pedestrian segregation in vehicle aisles | Share of scheduled barrier and walkway checks completed and passed | Pedestrian entries into vehicle-only zones per 1,000 vehicle movements; vehicle and pedestrian near misses |
| Forklift pre-use inspection | Share of shifts with a completed pre-use check for every truck in use | Defects found after an incident that a pre-use check should have caught |
| Racking inspection | Inspections completed on schedule; damage findings closed within the set time | Racking strikes reported; damage found that was not previously reported |
| Lockout for maintenance | Share of sampled jobs where an observer confirmed correct isolation | Isolation failures, including near misses |
This setup keeps the dashboard short. Every line simply answers two direct questions about a control that actually matters.
Which leading indicators are usually worth tracking?
The list below is editorial guidance based on the sources cited, not a standard. Adapt it to your hazards.
Follow-through measures
- Median time to respond to a hazard report. OSHA's guide gives response time to hazard reports as an example of a good leading indicator, noting that slower responses could signal a lack of management concern and that workers may stop reporting if management is not responsive [2].
- Overdue high-risk corrective actions, as a count and as the age of the oldest open item. This shows whether findings turn into fixes.
- Share of hazard and near-miss reports that received feedback to the reporter. It tracks whether the reporting system is worth using from a worker's point of view.
Control health measures
- Critical control verifications completed and passed, as in the dual assurance table above [3].
- Repeat findings: the share of inspection findings that appeared in a previous inspection of the same area. A high rate suggests fixes are not sticking.
Reporting culture measures
- Hazard and near-miss reports per 100 workers or per 200,000 hours. Report volume is ambiguous: it rises when hazards increase and when trust improves. Read it alongside feedback and closure measures, never as a target to drive down.
Technology-generated measures
AI video analytics, proximity sensors, and telematics can feed a dashboard with continuous exposure metrics that manual observation simply cannot match:
- Exposure rates tied to a physical denominator, such as close-proximity events per 1,000 vehicle movements or pedestrian zone entries per 100 camera-hours.
- Time elapsed from a high-severity alert to acknowledgment by a supervisor.
- Data availability: the share of planned camera-hours or device-hours actually monitored. Without tracking uptime, a sudden drop in detections might just mean the cameras went offline.
These measures change when cameras move, zones are redrawn, thresholds change or models update. Keep a change log, mark changes on the charts and run a structured human observation sample on a regular schedule to check that detections still match reality.
Which indicators should come off the dashboard?
| Indicator | Why it misleads | Better alternative |
|---|---|---|
| Training hours or courses completed | Measures effort, not competence or risk | Completion for named high-risk tasks before first assignment |
| Number of observations or inspections completed | Rewards volume; easy to inflate | Findings per inspection and repeat-finding rate |
| Total AI alerts or detections | Moves with coverage, thresholds and model changes | Detections per denominator, with uptime shown alongside |
| Days since last injury | Can create pressure not to report | Recordable and near-miss trends shown with their denominators |
| Number of near-miss reports, with a target to reduce | Rewards silence | Report rate plus feedback and closure measures |
OSHA's 2019 guide describes leading indicators built from training attendance, and those can serve as decent starting points for an employer totally new to the concept [2]. The problem comes when companies leave them on a management dashboard long after they stop providing useful information. A 2015 study of 60 companies found that firms rated above the median for adopting safety management rules used leading indicators far more often and monitored their indicators more regularly, while compliance-related indicators were the most common across all firms [7]. That suggests a clear path forward: use fewer, better-chosen measures, and review them frequently.
How should targets and incentives work?
Tying a target to the wrong measure shifts behavior in the wrong direction. US employers must have a reasonable procedure for employees to report injuries and illnesses promptly and accurately, and must not discriminate against an employee for reporting [8]. OSHA's 2018 clarification says programs that reward workers for reporting near misses or hazards are always permitted, and that rate-based incentive programs are permitted only if they are not implemented in a way that discourages reporting [9]. The same document suggests counterbalancing measures, such as rewarding employees for identifying unsafe conditions and checking whether workers feel free to report [9].
So a good practical rule is this: attach your targets to follow-through measures like actions closed on time or controls verified. Keep injury and incident counts for review, not for reward.
How should the dashboard be laid out?
- Group by hazard, not by data source. A reader should see everything about vehicle and pedestrian risk in one place, whether the numbers come from the EHS system, cameras or inspections.
- Show trends, not just traffic lights. Run charts with a median line and at least 12 months of history make it easier to tell a real shift from normal variation.
- Show denominators and coverage next to every rate.
- Separate site, region and board views. Boards need a few control-health and follow-through measures for the highest-risk hazards, plus lagging rates with multi-year context. Supervisors need the detail.
- Review the set every year. OSHA's guide recommends periodically reassessing goals and indicators and changing them based on what you learn [2]. Drop indicators that never move or never drive a decision.
Summary
A useful EHS dashboard is short. Choose a small number of leading indicators linked to your highest-risk hazards and critical controls, pair each with a lagging measure in the dual assurance pattern HSE described in HSG254 [3], and give every rate a denominator and an owner. Favor follow-through and control-health measures over activity counts. Use technology-generated data for exposure and response times, but show coverage alongside it and check it against human observation. Keep incentives away from injury counts. Because the evidence linking particular leading indicators to injury reduction is still weak [5], review the set every year and drop measures that do not change what anyone does.
Frequently asked questions
+How many leading indicators should an EHS dashboard have?
Fewer than most organizations use. A site dashboard typically works with a handful of indicators tied to its highest-risk hazards, plus the standard lagging rates. OSHA's guide notes that some employers find it helpful to limit how many leading indicators they use at one time. If nobody can say what decision an indicator drives, remove it.
+Is near-miss reporting a leading indicator, and should we set a target for it?
Yes, report volume is commonly used as a leading indicator of reporting culture, but it is ambiguous: a rise can mean more hazards or more willingness to report. Avoid targets to reduce near-miss reports. Pair the count with measures of what happened next, such as the share of reports that received feedback and the time to close the resulting actions.
+Can AI video detections replace manual safety observations on the dashboard?
They can supplement them, not replace them. Detections are continuous and cover hours that people cannot, but they only see what the cameras see and they change when the system is reconfigured. Keep a smaller program of structured human observation running so you have an independent check on what the technology reports.
+Should leading indicators be tied to bonuses?
Be careful. Rewarding activity counts invites box-ticking, and rewarding low injury or incident counts can discourage reporting. In the US, OSHA has said rate-based incentive programs are permissible only if they are not implemented in a way that discourages reporting. Rewarding follow-through, such as closing high-risk actions on time, is generally safer.
Related reading
Sources
- [1]OSHA, Leading indicators
- [2]OSHA, Using Leading Indicators to Improve Safety and Health Outcomes (OSHA 3970, June 2019)
- [3]HSE, Developing process safety indicators: a step-by-step guide for chemical and major hazard industries (HSG254, 2006)
- [4]US Chemical Safety Board, BP America Texas City refinery explosion, final investigation report (2007)
- [5]Watkins D, et al. A scoping review of the evidence base for the performance of leading indicators for improving safety outcomes. Journal of Safety Research, 2025
- [6]Bayramova A, Edwards DJ, Roberts C, Rillie I. Constructs of leading indicators: a synthesis of safety literature. Journal of Safety Research, 2023
- [7]Pawłowska Z. Using lagging and leading indicators for the evaluation of occupational safety and health performance in industry. International Journal of Occupational Safety and Ergonomics, 2015
- [8]OSHA, 29 CFR 1904.35 Employee involvement
- [9]OSHA, Clarification of OSHA's position on workplace safety incentive programs and post-incident drug testing (October 11, 2018)
- [10]US Bureau of Labor Statistics, Employer-reported workplace injuries and illnesses, 2024 (released January 2026)
- [11]Barnett AG, van der Pols JC, Dobson AJ. Regression to the mean: what it is and how to deal with it. International Journal of Epidemiology, 2005
- [12]ISO 45001:2018 Occupational health and safety management systems
Get the free Safety Tech Buyer's Toolkit
A 15-page PDF: the public evidence checklist, a weighted vendor scorecard, 40 RFP questions and pilot acceptance criteria. Subscribe and download it straight away; then one email a month when we publish or update guidance. No vendor promotions. Unsubscribe any time.