QHSE KPIs & metrics: formulas and worked examples
Choose a measure because it supports a decision: where to investigate, which action needs attention or whether a control is working. Define its population, period and calculation before putting it on a dashboard.
TekmonAd
Explore its published safety, quality and maintenance workflows. Confirm modules and pricing in a scoped proposal.
Give every metric a written definition
- Decision and owner: who reviews the result, what triggers action and who maintains the definition.
- Numerator and denominator: case or completion rules, units, exclusions and the people, sites or products covered.
- Period and cutoff: event date versus entry date, calendar period, reporting deadline and handling of late records.
- Source and version: source records, hours or output data, approval history and the rule for revising prior results.
- Missing data: show N/A and its reason when the denominator is zero or unavailable. Zero events with positive, complete exposure can produce a genuine zero rate.
For a completion percentage, define the cohort before calculating: actions due in a month, training assignments due by a date or units inspected in a batch. Dropping overdue items from the denominator makes an incomplete process look complete.
Copy a definition sheet for each metric
The five points above, as a form with the rate base added. Fill one in for every metric before it goes on a dashboard, and give the same sheet to every vendor in a demo. It is an editorial template: change the fields to suit your reporting scheme.
- Metric name: The name as it will appear on the dashboard.
- Decision and owner: Who reviews the result, what triggers action and who maintains the definition.
- Numerator: The case or completion rule, the unit and the exclusions.
- Denominator: The hours, units or assignments, and the people, sites or products covered.
- Period and cutoff: Event date or entry date, calendar period, reporting deadline and the handling of late records.
- Source and version: The source records, the approval history and the rule for revising a prior result.
- Missing data: What is shown when the denominator is zero or unavailable: N/A and the reason, not zero.
- Rate base: For a rate: 200,000 hours, 1,000,000 hours or another base, stated beside the number.
More on indicators, reporting and calculators is in the QHSE library.
Six worked examples
All numbers below are hypothetical teaching examples, not industry benchmarks. The operational measures illustrate definitions you can adapt and agree with your team.
Total Recordable Incident Rate (TRIR)
Recordable cases × 200,000 ÷ hours worked
3 × 200,000 ÷ 600,000 = 1.00
Three recordable cases and 600,000 employee hours in the same reporting period. The base represents 100 full-time equivalent workers.
Check: Use the applicable recording criteria. Match the population behind cases and hours; do not add contractor cases without their corresponding hours.
BLS incidence-rate method · source checked 20 September 2026
Days Away, Restricted or Transferred (DART) rate
DART cases × 200,000 ÷ hours worked
2 × 200,000 ÷ 600,000 = 0.67
Two cases involving days away, restricted work or transfer, using the same hours as the TRIR example. Round the final result.
Check: Count cases, not days. A case involving both absence and restricted work is counted once in the combined numerator.
BLS incidence-rate method · source checked 20 September 2026
Corrective actions completed on time
Actions completed by their due date ÷ actions due in the period × 100
18 ÷ 24 × 100 = 75%
Of 24 actions due in September, 18 were completed on time. The other six remain in the denominator whether late or still open.
Check: Freeze the due-date cohort at the reporting cutoff and retain changes to due dates. Track effectiveness checks and reopened actions separately.
Training assignments completed on time
Required assignments completed by deadline ÷ required assignments due × 100
108 ÷ 120 × 100 = 90%
A September cohort contains 120 required course assignments; 108 were completed by deadline. One employee may have several assignments.
Check: Do not relabel this as the percentage of trained employees. Document joiners, leavers, exemptions and competence checks.
First-pass acceptance
Units accepted without rework ÷ units inspected on first pass × 100
460 ÷ 500 × 100 = 92%
Of 500 units inspected in one production cohort, 460 met the agreed acceptance criteria on the first attempt.
Check: Keep product mix and inspection criteria visible. Do not count re-inspected units again or combine defect counts with unit counts.
Energy intensity
Energy consumed in kWh ÷ units produced
12,000 ÷ 5,000 = 2.4 kWh per unit
The metered boundary consumed 12,000 kWh while producing 5,000 comparable units in the same period.
Check: Report total consumption alongside intensity. Changes to product mix, meter boundaries or output can change the ratio without an efficiency improvement.
LTIR and severity: write down the lost-time case definition, rate base and day-counting convention used by your reporting scheme. A rate per million hours is five times the same rate per 200,000 hours; neither is comparable until the definitions and base match.
Combine counts and hours before calculating a group rate
These hypothetical sites use the same reporting period, case definition and 200,000-hour base. Their hours differ, so the sites must not receive equal weight.
| Site | Cases | Hours | Rate |
|---|---|---|---|
| A | 1 | 20,000 | 10.00 |
| B | 1 | 180,000 | 1.11 |
| Combined | 2 | 200,000 | 2.00 |
Combined rate: 2 × 200,000 ÷ 200,000 = 2.00. The simple average of the site rates is about 5.56 and does not represent the combined hours. Keep individual sites visible too: a group result can hide a local problem.
For external benchmarking, match the industry, size, reporting rules and period. Show the case count and exposure alongside a rate, especially for small teams. There is no universal TRIR threshold that proves a workplace is safe. See the dated statistics and comparison limits.
Balance outcomes with evidence that controls work
Leading indicators
Examples include critical controls checked, corrective actions completed and required training completed. Pair activity counts with evidence of quality: a completed inspection is useful when its findings are addressed and the control is effective.
Lagging indicators
Examples include recorded injuries, days lost, defects and spills. They describe outcomes already observed. Review severity, recurring causes and exposure alongside the frequency.
More near-miss reports can reflect increased reporting, changed exposure or more events. Investigate the context; do not reward a falling count in isolation or use a fixed pyramid ratio to predict serious injuries.
Health: track completion of assessments or monitoring against the population for whom they are required. Restrict access to personal health information; a management dashboard should use suitable aggregate measures.
Environment: retain absolute totals as well as intensity measures. For emissions, disclose inventory boundaries, factor versions and scope coverage. The GHG Protocol Scope 3 Standard supports tracking a company's emissions over time; it does not by itself make different companies' inventories comparable. Source checked 20 September 2026.
Ask a vendor to reproduce these results
Use a small sample you control. Ask for the underlying records and exported calculation, not just a dashboard screenshot.
- Load the two-site example. Confirm site rates and the combined 2.00 result; change the reporting period and inspect which cases and hours remain included.
- Remove a site's hours. Confirm the rate is marked unavailable and the missing exposure is visible; restore the hours and trace the revision.
- Load the 24-action due-date cohort. Complete an overdue action and change another action's due date. Inspect how the original report and revision history are preserved.
- Reclassify an incident and recalculate the prior period. Confirm who approved the change and which exported reports are affected.
- Open the same dashboard as a site manager and as a group reviewer. Verify access boundaries and export the definitions, filters and source records needed to reproduce each total.
What vendors document, task by task
These are counts from our 212 researched profiles, checked through 2026-10-14, of what each profile states about the capability a demo task relies on. Where a profile has no statement, we found none; that does not mean the product lacks it. Every statement is the vendor's own and was not tested, and a statement about a dashboard says nothing about whether it calculates a rate correctly. Documented means the vendor's pages describe the capability; conditional means they tie it to a plan, module, edition, region or set-up. See the capability map and the integrations study for the rules. No product is named and nothing is ranked.
| Demo task | The record states | Profiles | So in the demo |
|---|---|---|---|
| Tasks 1 and 2 | Has a documented or conditional reporting-and-dashboards statement84 documented; 46 conditional only. | 130 of 212 | Use your own sample: load the two sites, change the period and remove one site's hours, then inspect what the dashboard still shows. |
| Task 3 | Has a documented or conditional workflow-and-approvals statement60 documented; 40 conditional only. | 100 of 212 | Complete an overdue action and change a due date, then check that the original cohort is still reproducible. |
| Tasks 3 and 4 | Has a documented or conditional audit-trail statement17 documented; 11 conditional only. | 28 of 212 | Press hardest here: ask to see who changed a record, when, and which exported reports the change affects. |
| Task 5 | Has a documented or conditional security-and-sign-in statement31 documented; 53 conditional only. | 84 of 212 | Open the same dashboard as a site manager and as a group reviewer and check what each can see. |
| Task 5 | Mentions a CSV interface125 mention an API. | 33 of 212 | Export the definitions, filters and source records, and check that you can reproduce each total outside the product. |
A dashboard statement is the most common of these, and a record of who changed what is among the rarest. The tasks about revisions and approvals are therefore the ones where you are least likely to be able to check a claim beforehand.
Continue with a worked requirements brief, the QHSE glossary or the source reference.