/ Articles

Safety KPIs: Which Metrics Show Your Workforce Is Ready

November 6, 2026

← Back to Articles

Two workers in hard hats and orange coveralls record readings on a clipboard at a plant doorway

Most safety scorecards are built around events that already happened. By the time a recordable shows up on the log, any knowledge gap behind it may have been there for weeks.

A good set of safety KPIs does two jobs. It tells you how the program performed, and it warns you where the next failure is forming. This guide covers both kinds of metric, how OSHA and the Bureau of Labor Statistics define the common injury rates, and where workforce knowledge fits on an EHS scorecard.

Table of Contents

Leading vs Lagging Safety KPIs

OSHA's guide Using Leading Indicators to Improve Safety and Health Outcomes (OSHA 3970, 2019) draws the line. For the purposes of that document, leading indicators are "proactive, preventive, and predictive measures" that show how well your safety and health activities are performing. Lagging indicators "measure the occurrence and frequency of events that occurred in the past, such as the number or rate of injuries, illnesses, and fatalities."

OSHA does not ask you to pick one. The guide says: "A good program uses leading indicators to drive change and lagging indicators to measure effectiveness." It is also explicit about its status: "This document is not a standard or regulation, and it creates no new legal obligations."

Lagging indicatorsLeading indicators
What they countInjuries, illnesses, fatalities, lost daysActivities and conditions before an incident
When you see themAfter the eventBefore the event
Typical useMeasure program effectivenessDrive change and find weak spots
WeaknessSmall numbers swing widelyOnly useful if they track something that matters

The last row is the hard part: an activity count (meetings held, forms filed) can look healthy while the risk underneath grows.

How OSHA and BLS Calculate TRIR and DART

The injury rates on most scorecards start with OSHA's recordkeeping rule. Under 29 CFR 1904.4, an employer required to keep records must record each fatality, injury and illness that meets three tests. Each case must be work-related, be a new case, and meet one or more of the general recording criteria of 1904.7 or the specific-case rules in 1904.8 through 1904.12.

The general criteria in 1904.7 are death, days away from work, restricted work or transfer to another job, medical treatment beyond first aid, or loss of consciousness. A significant injury or illness diagnosed by a physician or other licensed health care professional also meets the general criteria, even without those outcomes.

Part 1904 has partial exemptions. Under 1904.1, for example, a company with 10 or fewer employees at all times during the last calendar year does not need to keep these records. That exception ends if OSHA or BLS informs it in writing that it must keep records under 1904.41 or 1904.42. It must still report to OSHA the work-related fatalities, in-patient hospitalizations, amputations and losses of an eye covered by 1904.39.

The rate itself is a fixed formula. Answering a company that asked whether the formula is biased against smaller employers, OSHA's 2016 interpretation letter gives it as "(Number of injuries and illnesses X 200,000) / Employee hours worked = Incidence rate." The letter explains that "The 200,000 figure in the formula represents the number of hours 100 employees working 40 hours per week, 50 weeks per year would work, and provides the standard base for calculating incidence rate for an entire year."

BLS gives the same formula on its page "How To Compute Your Firm's Incidence Rate for Safety Management," with these details:

  • Total recordable case rate (often called TRIR): count all recordable cases for the year. From the OSHA Form 300A summary, that is Column H (days away) + Column I (job transfer or restriction) + Column J (other recordable cases).
  • DART rate: count only cases with days away from work, job transfer or restriction, or both, which is Column H + Column I.
  • Hours: the hours all employees actually worked, excluding nonwork time even when paid, such as vacation, sick leave and holidays.

OSHA's 2016 letter adds a caution for smaller sites: one case moves a small establishment's rate far more than a large one's. It suggests comparing against BLS rates for employers doing similar work with a similar workforce size and combining several years of data.

Safety KPI Examples for Each Side

The lagging list is short and standardized; the leading list is each site's choice. The examples below are this article's own suggestions, except where OSHA 3970 is named.

Lagging safety KPI examples

  • Total recordable case rate (TRIR)
  • DART rate
  • Lost workdays and restricted days
  • Fatalities and the severe injuries reported under 1904.39

Leading safety KPI examples

  • Hazard and near-miss reports submitted, and the share closed out
  • Corrective actions closed on time
  • Planned inspections and audits completed
  • Pre-task and job hazard analyses reviewed in the field
  • Worker participation in safety committees and walk-throughs
  • Completion and knowledge measures (next two sections)

Keep the leading list short. OSHA 3970 says good leading indicators are based on SMART principles: "Specific, Measurable, Accountable, Reasonable, and Timely". Three to six indicators that each trigger a decision beat twenty that nobody reads.

Where Completion Rates Fit

OSHA 3970 has a section on education and training. Its sample indicators include "Percentage of workers receiving mandatory training on schedule" and "Percentage of improvement on post-training assessment scores over pre-training assessment scores." Another compares the number of workers trained to recognize and report a hazard or near miss with "the number of workers that report understanding the training they have received."

The guide's own example of acting on that data makes the point. In it, a leading indicator may reveal that you offer regular training on recognizing and reporting hazards, but "workers still do not clearly understand how to recognize and report certain hazards after completing the training." Training on schedule and understanding are two different numbers.

That gap is the subject of our article on tracking employee training, which compares what most teams track with what predicts readiness. For the safety scorecard, the point is narrower: a completion rate is a valid leading indicator of activity, and it says little about what a worker could recall on the floor six months later.

Knowledge Level as a Leading Indicator

If the hazard is a skipped isolation step or a misread label, what the crew knows sits upstream of the incident. Measuring it gives you a leading indicator tied to the risk itself.

Two numbers matter: where knowledge stands now, and which way it is moving. Blank Slate's True Knowledge Level™ is a real-time readiness metric that shows what a person or team knows now. Blank Slate delivers AI-optimized spaced retrieval: spaced repetition (reviewing material over time) combined with retrieval practice (answering from memory instead of rereading). Sessions are short (about 60 seconds), three or four times a week, adding up to a few minutes a week, on top of the training content you already have.

A baseline reading is often lower than leaders expect. In a 16-month Air Force study, described in a conference paper prepared for I/ITSEC 2026, fully qualified F-16 and C-17 pilots averaged 66.91% accuracy the first time they answered operational knowledge questions. That is a snapshot of what qualified professionals knew, not a measure of forgetting. With brief Blank Slate sessions, final average accuracy for the 37 pilots analyzed rose to 81.36%, a 21.6% relative improvement (Smith & Randolph, I/ITSEC 2026).

Tracked over time, the same measure shows whether a gap is closing. In a year-long case study of 70 data-center security guards presented at MODSIM World 2025, the half who started below the team's median went from about 56% to over 90% accuracy. They closed the gap with the high performers after about 180 days, using three sessions a week, six questions each (Smith, MODSIM World 2025).

Knowledge measures can also line up with assessment results. In the Air Force Pararescue study of Blank Slate presented at I/ITSEC 2023, app users had an 84% first-time pass rate on an applied knowledge test, versus a 46% first-time pass rate for non-users. That is an association, not proof of cause.

None of these studies measured recordable injury rates, and a knowledge metric does not replace your lagging indicators. It is an early read on one input to them. How supervisors act on that signal is covered in frontline worker training.

Safety KPIs in Manufacturing

In manufacturing, TRIR and DART describe a whole plant. Leading indicators work better hazard by hazard. A few pairings to consider:

  • Hazardous energy: lockout/tagout procedures audited and knowledge of the isolation steps for each machine group (see how to write a lockout/tagout program crews follow).
  • Chemical hazards: labels and safety data sheets spot-checked, and whether workers can read a label's pictograms and signal word.
  • Material handling: forklift and pedestrian near-miss reports, and pre-shift inspection completion.
  • Rare tasks: knowledge checks before infrequent jobs such as annual shutdowns or confined space entries, when recall is most likely to have faded.

The pattern is the same in each pairing: one activity measure and one knowledge measure for the same hazard. If the activity number is green and the knowledge number is falling, you have found your next conversation before the log shows it.

What This Means for EHS Leaders

Keep TRIR and DART, calculated the way OSHA and BLS define them, so your numbers compare fairly. Then choose a few leading indicators tied to your top hazards, and include at least one that measures what the workforce knows, not only what it attended. Evidence that workers understood their training is also what an audit looks for (see EHS compliance).

A Knowledge Gap Audit shows what the team knows today, where gaps may exist and where to focus first. See how Blank Slate supports industrial safety teams, and how it coexists with the LMS that holds your records.

Frequently Asked Questions

What is a safety KPI?

A safety KPI is a metric an organization uses to judge its safety and health performance. Lagging KPIs, such as recordable and DART rates, count injuries and illnesses after they happen. Leading KPIs, such as hazard reports closed or knowledge levels, measure conditions and activities before an incident.

What is the difference between TRIR and DART?

Both use the formula BLS and OSHA publish: cases times 200,000, divided by employee hours worked. TRIR counts all recordable cases (OSHA 300A Columns H, I and J). DART counts only cases with days away from work, job transfer or restriction (Columns H and I).

Does a completion rate count as a leading indicator?

Yes. OSHA's leading indicators guide lists the percentage of workers receiving mandatory training on schedule as one example. It also suggests measuring improvement on post-training assessment scores and how many workers report understanding their training. Together these show more than completion alone.

Injury rates tell you how last year went. A Knowledge Gap Audit gives you a one-page readiness snapshot within days: what's retained, what's slipping, and where to reinforce first.

Request a Knowledge Gap Audit

More Articles