Blog · Board and management reporting · Manufacturing
Safety dashboards compare aggregate workplace incidents with hours worked and corrective-action status. This guide works through site rates, company totals, and an action backlog with checks that keep the reporting consistent.
Safety dashboards compare aggregate workplace incidents, exposure hours, and open corrective actions over a defined period. The useful comparison shows counts beside rates and the action backlog beside its due dates, so managers can distinguish a change in recorded outcomes from a change in the reporting population.
Start with the establishment, review period, and workforce included in the calculation. Match case counts and hours to that same boundary. A plant's employee cases should not be divided by hours that also include unrelated locations or a contractor population absent from the numerator.
Have the responsible safety team confirm case classifications before the aggregate export is prepared. Keep names, medical details, and individual investigation narratives outside the management dashboard. The reporting file can use establishment, period, case category, and aggregate counts without reproducing personal records.
Separate a metric from a decision. The KPI glossary explains why a measure needs a purpose and owner. Here the purpose might be to review a site change, investigate an unexplained denominator shift, or assign overdue follow-up work. A lower displayed rate alone does not explain what changed.
Use hours actually worked for the included population. Oregon OSHA's recording guide gives the total case incidence rate, or TCIR, and the days-away, restricted-work, or transfer rate, or DART, using a base of 200,000 hours. That base represents 100 employees working 40 hours a week for 50 weeks: 100 × 40 × 50 = 200,000.
TCIR = Recordable injury and illness cases / Hours actually worked × 200,000
DART rate = Cases involving days away, restricted work, or transfer / Hours actually worked × 200,000
DART is a subset of recordable cases, not an additional category to add to the recordable total. The numerator is cases, not days lost. Preserve the original counts and calculate the rates from them rather than importing rounded rates from separate reports.
Label quarterly and annual views explicitly. Applying the standard base to a quarter normalizes for hours; it does not create an observed full-year result. Small case counts can produce large movements, so the dashboard needs the count and denominator next to each rate.
An illustrative US manufacturer reviews July through September 2026 across two plants. Its safety team supplies approved aggregate case counts, and payroll supplies the matching hours-worked totals. Rates are rounded to two decimal places only after calculation.
| Plant | Hours actually worked | Recordable cases | DART cases | TCIR | DART rate |
|---|---|---|---|---|---|
| North | 120,000 | 2 | 1 | 3.33 | 1.67 |
| South | 80,000 | 1 | 0 | 2.50 | 0.00 |
| Company | 200,000 | 3 | 1 | 3.00 | 1.00 |
North's TCIR is 2 / 120,000 × 200,000 = 3.33. South's is 1 / 80,000 × 200,000 = 2.50. The company rate is 3 / 200,000 × 200,000 = 3.00.
Do not average the displayed plant rates. The unrounded simple average is 2.92, which gives each plant equal weight despite different hours. Dividing the total cases by total hours returns the correct company result. Weighting the unrounded site rates by their hours gives the same answer, a useful reconciliation check.
Now change only North's exposure hours from 120,000 to 100,000, keeping its two cases. Its TCIR becomes 2 / 100,000 × 200,000 = 4.00. The case count did not change; fewer hours explain the rate movement. Do not describe this variation as evidence that a specific practice caused more incidents.
Incident rates look backward. The action register shows what follow-up remains open at the reporting cutoff. Oregon OSHA's accident-investigation guidance includes acting on recommendations and following up to confirm appropriate corrective action was taken. A management report should therefore retain evidence of completion, not just a status label.
For this example, an action leaves the open balance only after the designated reviewer accepts its closure evidence. No actions were canceled, reopened, or transferred during the quarter. Opening and closing balances use the same scope.
| Plant | Open at start | New actions | Verified closures | Open at end | Overdue at end |
|---|---|---|---|---|---|
| North | 4 | 2 | 3 | 3 | 1 |
| South | 3 | 3 | 2 | 4 | 1 |
| Company | 7 | 5 | 5 | 7 | 2 |
The check is opening actions plus new actions less verified closures equals closing actions: 7 + 5 − 5 = 7. Overdue actions are a subset of that closing balance: 2 / 7 = 28.6%, rounded to one decimal place.
Define overdue as an open action with its due date before the cutoff. An action due on the cutoff date is due that day, rather than already overdue. If an owner extends a deadline, retain both the original date and approved revised date so rescheduling does not silently remove the exception.
An action list should also show owner, date raised, original due date, current due date, status, and a reference to closure evidence. Show high-priority unresolved work individually within the appropriate internal review process. A backlog count treats unlike actions equally and should not replace that review.
Arrange the dashboard in three sections: incidents with exposure, trends using a stable scope, and the open-action register. The dashboard software guide covers the broader checks for reliable source measures; the immediate task here is to make each aggregate traceable to its supplied register.
The safety lead owns the case definitions and explanations. Payroll or operations confirms worked hours. Action owners supply completion evidence, and the designated reviewer confirms closure. Keep these responsibilities visible when a number is missing or disputed.
Agree which measures deserve routine escalation. As KPI versus metric versus measure explains, a long list of measures is less useful than a small set tied to action. In this example, the review can ask why North's reported hours changed and who will resolve the overdue actions; the rates cannot answer those questions by themselves.
Bring aggregate case counts, matching hours, and a nonpersonal action-status export to Covirage, which maps the columns and uses its tools to compute and check the figures. The external AI model explains the comparisons with every figure traced to its rows. Create a free Covirage account, which takes about a minute, then set up your business and bring a sample of your own data.
See Covirage for industrial manufacturers for the manufacturing reporting context.
Include aggregate incident counts, hours actually worked, clearly defined incidence rates, and open corrective actions by owner and due date. Show the period, population, and source cutoff. Add a clear distinction between actions marked complete and actions whose completion has been verified.
TCIR uses all recordable injury and illness cases. DART uses cases involving days away from work, restricted work, or transfer. Both divide their case count by hours worked and multiply by the same standard base. Count cases for DART, rather than adding the days associated with them.
The denominator may have fallen. If the same number of cases occurs over fewer hours worked, the calculated rate rises. Review hours, case counts, work mix, and the reporting period together before attributing the change to a particular operational cause.
A zero rate means no cases in the selected category were recorded for the chosen period and population. It does not describe every hazard, reporting gap, or unresolved action. Review the underlying records and follow-up work rather than treating an outcome measure as assurance.