Skip to main content
    QHSEtech
    All templatesAudits · 3 pages · Updated 19 October 2026

    KPI Definition Sheet Template

    Three-page KPI definition sheet for QHSE metrics: two metric sheets and a page of tools. Each sheet has twelve parts: metric name, the decision it supports and who owns the definition, numerator, denominator, period and cutoff, source and version, what is shown when data is missing, the rate base, how a past result is revised, baseline and target with who justified them, review frequency and a worked-example box. The first eight parts follow the eight-line sheet on the KPIs and metrics page, in the same order. Page 3 adds a change log, a table of rate bases and a check list to use before a metric goes on a dashboard. What counts as a case in an injury rate is set by the scheme or rule you report under; this form does not say what a case is, and a completed sheet does not show compliance with any rule.

    What's inside

    Pages 1 and 2: one definition sheet each, the same twelve parts, so one form defines two metrics

    1. Metric name: the name as shown on the dashboard, the unit shown, the definition version and date, and whether the metric is lagging (an outcome already seen), leading (an activity or a control) or other
    2. Decision and owner: the decision the metric supports, the result that triggers action, who reviews it and who maintains the definition
    3. Numerator: the case or completion rule, the unit counted, the exclusions and who settles a borderline case
    4. Denominator: what it measures (hours worked, units, assignments), how the amount is obtained, the people, sites or products covered and left out, and whether the coverage is the same as the numerator's
    5. Period and cutoff: event date or entry date, the calendar period, the reporting deadline and the rule for late records
    6. Source and version: the source records for the numerator and the denominator, the extract version and date, who approved the definition and the version it replaces
    7. Missing data: what is shown when the denominator is zero or unavailable (N/A and the reason, not zero) and when a zero is shown as a real zero
    8. Rate base: 200,000 hours, 1,000,000 hours, percent or another base, the formula as written, the decimals shown and whether the base is printed beside the number
    9. Revision of a past result: how a published result is changed and the earlier figure kept, who approves a revision and which reports it affects
    10. Baseline and target: the baseline and its period, the target and its date, the comparison used, and who justified them and why
    11. Review frequency: how often results are reviewed and how often the definition itself is
    12. Worked example: a box for the numerator, denominator, base, calculation and result on one past period, and whether it matches the published figure

    Page 3: tools for both sheets

    1. Change log for definitions: eight lines: date, metric, version, what changed, the effect on past results and who approved it
    2. Rate bases at a glance: 200,000 hours, 1,000,000 hours, 20,000,000 hours, per 100,000 workers and percent, with how to convert and what each is for, and a note that a group rate adds cases and hours instead of averaging site rates
    3. Before the metric goes on a dashboard: ten ticks and a place to say where the sheets are kept

    The section titles, the options and the example are ours, to change.

    What requires or shapes this record

    No source opened for this page says that a KPI needs a written definition sheet, and this form does not say that any rule requires one. The sources show the parts of a definition and where each part matters. In the US, the Bureau of Labor Statistics (BLS) explains how an employer computes an incidence rate from a count of cases, the hours actually worked and a base of 200,000 hours, and 29 CFR Part 1904 sets what a covered employer records, how quickly, what the year-end summary holds and how stored records are updated. In Great Britain, HSE's statistics prefer a rate per 100,000 workers when looking at change over time and publish recent figures as provisional before finalising them. What counts as a recordable or reportable case differs by country, rule and scheme, so it is for you to check in the text you report under; this form only has a place to write the rule down. This page is general information, not legal advice.

    • BLS, How To Compute Your Firm's Incidence Rate for Safety Management (US Bureau of Labor Statistics; page last modified 10 March 2025; opened 19 October 2026)

      BLS gives the formula number of injuries and illnesses × 200,000 ÷ employee hours worked, and says the 200,000 hours stand for 100 employees working 40 hours a week for 50 weeks and form the standard base. The count comes from the employer's injury and illness log or its summary, and the hours from the summary, from a BLS survey form or from payroll or other time records. Hours worked leave out non-work time such as vacation, sick leave and holidays even when paid, and where actual hours are not available for some employees, such as those paid on commission or salary, BLS allows an estimate from scheduled hours or 8 hours per workday. The same formula is used for subsets of cases, and for comparing illness rates by type BLS suggests 20,000,000 hours, a rate per 10,000 full-time employees. Its example is 7 cases in 400,000 hours, a rate of 3.5, and it says rates mean more when compared with employers doing similar work with workforces of similar size. The page is written for US employers, and BLS's server blocked scripts, so it was read in a browser. It does not say what a recordable case is. The form's parts 3, 4 and 8 and the table of rate bases follow this method in our own wording.

    • 29 CFR Part 1904, Recording and reporting occupational injuries and illnesses, sections 1904.4(a), 1904.29(b)(3), 1904.32 and 1904.33 (US; eCFR as of 15 October 2026, read on 19 October 2026)

      A US rule for the employers that Part 1904 covers (it has partial exemptions by company size and industry, and the full text is for you to read). Section 1904.4(a) says a covered employer records each fatality, injury and illness that is work-related, is a new case and meets the general recording criteria or the criteria for specific cases, and points to where each test is defined, so the case rule is a text to check. Section 1904.29(b)(3) says each recordable case is entered within seven calendar days of receiving information that it has occurred. Section 1904.32 asks for a review of the log at the end of each calendar year and a summary with column totals (zeros if there were no cases), the annual average number of employees covered and the total hours worked by all employees covered. Section 1904.33 asks the employer to keep the records for five years after the year they cover and to update the stored log for newly discovered cases and for changes in the classification of cases, removing or lining out the original entry; it says the annual summary need not be updated but may be. These are record-keeping rules, not KPI rules, and the form borrows only their lessons: define the case from the text, say how late information is handled, keep the hours and the cases on the same coverage and keep the earlier entry. The standard's own text is not reproduced here.

    • HSE, Work-related fatal injuries in Great Britain (HSE statistics page, updated 30 June 2026; opened 19 October 2026)

      HSE's page for Great Britain says that when looking at change over time it is preferable to use the rate of injury per 100,000 workers, because that allows for changes in the numbers in employment between years. It gives RIDDOR, the injury reporting regulations, as the data source and says that figures for 2025/26 are published as provisional and will be finalised in July 2027. That is a headcount base, not hours, and a figure that is shown first and settled later. The form's rate base part and its revision part ask you to choose the base on purpose and to say whether a published figure is provisional. The page does not define a KPI or tell an employer how to compute its own rate.

    Each source checked on 19 October 2026. Not legal advice; check the text in force for your site.

    Field by field

    Each sheet is one metric: fill it in before the metric goes on a dashboard, and give the same sheet to anyone who will calculate or display it. Parts 1 to 8 are the eight lines of the copyable definition sheet on the KPIs and metrics page, in the same order; parts 9 to 12 add what that sheet leaves out. The options, the section titles and the example are ours and can be changed. Keep the same parts between metrics so that sheets can be compared, and make a copy of page 1 for each further metric.

    Metric name
    Why: The name is what readers of the dashboard see, so it must stand for one calculation. The unit, the version and whether the metric is lagging (an outcome already seen) or leading (an activity or a control) tell the reader what kind of number it is.
    Common mistake: Using one name, such as "incident rate", for two calculations: one per 200,000 hours at one site and one per 1,000,000 hours at another.
    Decision and owner
    Why: BLS says rates help show problem areas and progress. A metric that supports no decision is decoration, so write what will be done differently if it moves, and the result that triggers action. The owner who maintains the definition is the person who can change it, not only who reads it.
    Common mistake: Naming a department as the owner, so nobody changes the definition when it turns out to be wrong.
    Numerator
    Why: The numerator is a rule, not a number: what makes one case, in what unit, with which exclusions, and who settles a borderline one. 29 CFR 1904.4 is an example of a US case rule with three tests, and the case rule of your own scheme is a text to check and copy from, not to recall.
    Common mistake: Copying a case rule from memory or from another site's report, or counting events when the scheme counts cases.
    Denominator
    Why: BLS uses hours actually worked, leaves out paid non-work time and allows an estimate in some cases. Say what the denominator measures, how the amount was obtained and whether it is an estimate, and who or what it covers. 29 CFR 1904.32 asks for the total hours of all employees covered by the log, so the denominator covers the same people as the numerator.
    Common mistake: Counting injuries to contractors in the numerator but only employees' hours in the denominator.
    Period and cutoff
    Why: Event date or entry date, the calendar period, the deadline and the rule for late records decide which period a case falls in. 29 CFR 1904.29 sets a time for entering a case in a US log (seven calendar days from receiving the information); that is about recording, and which period a case counts in for a KPI is a separate choice you write down.
    Common mistake: Leaving the date rule unwritten, so a case entered in October for a September event is in September on one report and in October on another.
    Source and version
    Why: BLS names the places the count and the hours come from (the log or summary, and payroll or other time records). Write the source records for both, the extract version and date, who approved the definition and which version it replaces, so another person can reproduce the figure.
    Common mistake: Pasting a figure into a slide with no extract date, so a later number that differs cannot be explained.
    Missing data
    Why: A rate cannot be calculated when the hours are zero or have not arrived. Show N/A and the reason, not zero. A zero is a real result only when the exposure is positive and complete: 29 CFR 1904.32 has employers enter zeros in the column totals when there were no cases, which is a count with the hours known.
    Common mistake: Showing 0.00 for a site whose hours were never received, which looks like a perfect record.
    Rate base
    Why: BLS's 200,000 hours equals 100 full-time workers for a year, and 1,000,000 hours is five times that base. The base turns a count into a comparable rate only if it is stated beside the number, along with the formula and the decimals.
    Common mistake: Comparing a rate per 1,000,000 hours with a rate per 200,000 hours as if they were on one scale, or printing the number with no base.
    Revision of a past result
    Why: A case found late or reclassified changes a result already published. 29 CFR 1904.33 has a US employer update its stored log and remove or line out the original entry, and HSE publishes recent figures as provisional and finalises them later. Write how your metric is revised, who approves it, that a revised figure is marked as revised and where the earlier figure is kept.
    Common mistake: Overwriting last quarter's figure in place, so the number that was reported and acted on cannot be found.
    Baseline and target
    Why: A baseline is the result of the same definition in an earlier period, and a target is a choice. BLS says rates mean more when compared with similar employers of similar size. Write who justified the target and what the reasoning was, so it can be reviewed when it is reached or missed.
    Common mistake: A target copied from another organisation's figure that uses a different case rule or rate base, with nobody named who can explain it.
    Review frequency
    Why: Two reviews are needed: of the results, as often as the decision is made, and of the definition itself, which can go out of date when processes, sites or the reporting scheme change. 29 CFR 1904.32 asks a US employer to check its log for completeness each year; the same idea applies to a metric's definition.
    Common mistake: Reviewing the figures every month but never the definition, so a change in the work leaves the metric measuring something else.
    Worked example
    Why: Calculating one past period by hand with the written rules, and comparing it with the figure already published, shows whether the sheet describes what the report actually does. BLS's own worked example, 7 cases in 400,000 hours giving 3.5, is the kind of check to make.
    Common mistake: Testing the sheet only with made-up numbers, so a mismatch with the live report never shows.
    Change log for definitions
    Why: A change to a numerator, denominator, base, cutoff or source can move a trend line without anything changing at work. Date each change, say whether earlier periods were recalculated and keep the old version.
    Common mistake: Changing a definition with no entry, so a break in the trend has no explanation a year later.
    Rate bases at a glance
    Why: Shows what each base stands for and how to convert between the hours bases (multiply a rate per 200,000 hours by 5 for 1,000,000 hours, by 100 for 20,000,000). A per-100,000-workers rate, as HSE uses for fatal injuries, is a headcount base and cannot be converted without hours. A group result adds cases and hours before applying the formula.
    Common mistake: Averaging the site rates: one case in 20,000 hours (10.00) and one in 180,000 hours (1.11) give 2.00 combined, not the average of 5.56.
    Before the metric goes on a dashboard
    Why: Ten ticks that gather the points above into one check, and a line for where the sheets are kept and who can change them. A sheet that cannot be found or is edited by anyone stops being the definition.
    Common mistake: Ticking every box with nobody having reproduced a published figure from the sheet.

    Worked example

    IllustrativeA definition sheet for an invented plant: four cases in 310,000 hours worked in one quarter, shown per 200,000 hours (invented company, people and references)

    SectionEntry
    1 Metric nameInjury case rate, Plant 1, definition version 2, shown as a rate per 200,000 hours worked. It is a lagging indicator: it counts outcomes already seen.
    2 Decision and ownerDecides whether the hand-injury actions started in the third quarter continue into the fourth. Reviewed by the plant manager and the EHS coordinator at the monthly meeting; the EHS coordinator maintains the definition. A rise of two or more cases in a quarter triggers a review of the actions.
    3 NumeratorCases that meet the written case rule of the reporting scheme the plant uses, copied from the scheme's text and checked by the EHS coordinator, one count per case. First-aid-only events are excluded, and a borderline case is settled by the EHS coordinator with the plant manager.
    4 DenominatorHours actually worked by all employees on the plant's log, from payroll and time records, with paid leave left out. No hours are estimated. Contractors' hours and injuries are both left out and reported separately, so the coverage is the same in the numerator and the denominator.
    5 Period and cutoffEvent date, calendar quarter, entries to be made by the seventh day after the quarter ends. A case entered later is counted in the quarter of its event date and the published result is marked as revised.
    6 Source and versionThe case register export of 9 October 2026 and payroll hours file version 3. Definition version 2, approved by the plant manager on 1 July 2026, replacing version 1.
    7 Missing dataIf hours for any month are missing, the quarter shows N/A with the reason (for example, September hours not received), not a rate and not 0.00. Plant 2's third-quarter result shows N/A for this reason and is left out of the group result until its hours arrive.
    8 Rate base200,000 hours, printed beside the number. Third quarter 2026: 4 cases × 200,000 ÷ 310,000 hours = 2.58 per 200,000 hours (the same result per 1,000,000 hours is 12.90).
    9 Revision of a past resultThe result published on 30 September had 3 cases: 3 × 200,000 ÷ 310,000 = 1.94. A fourth case, with an event date of 28 September, was entered on 9 October, so the quarter was revised to 2.58, marked as revised and approved by the plant manager, and the earlier 1.94 is kept in the change log.
    10 Baseline and targetBaseline: third quarter 2025, 3 cases in 285,000 hours = 2.11. Target: 2.00 or lower for the rolling 12 months to 30 September 2027, set by the plant manager and the EHS coordinator from the trend and the planned actions. It is an internal target, not a benchmark, and at this size one case moves the rate by about 0.65, so the review shows the case count and the hours beside the rate.

    Every figure, name and date is invented to show the method. The case rule is a placeholder: the real one comes from the scheme or rule you report under and is for you to check. The rate does not say whether Plant 1 is safe, and a result on so few cases should be read with the case count and the hours.

    When a spreadsheet stops being enough

    A word-processor file or spreadsheet is enough for a few metrics at one site where one person calculates and owns them. It stops being enough when the hours come from payroll or time systems at several sites and must be joined to the cases, when a late or reclassified case must change a published result while the earlier figure stays visible and flagged, when the same definition has to apply at every site and in every dashboard, and when you need to know which version of a definition produced a given number.

    A software record keeps each metric's definition as a versioned record with an owner and approval, links the numerator records and the denominator hours with the date of the extract, shows N/A and the reason when hours are missing, restates a period with a visible revision history and prints the rate base beside the number. Ask to see these with your own sheet: remove one site's hours, reclassify one case and recalculate the previous period. The list below applies the cross-site reporting and data-feed rule used by the QHSE software integrations guide. Nothing in it shows that a product keeps definitions, versions or restated results, so test those with your own metric.

    Same rule for every product: tagged “EHS Management”, “Safety Management”, “Quality Management”, “Incident Management”, “Incident Reporting”, “Audit Management”, “Audits & Inspections”, “Inspection Management”, “Inspections”, “Risk Assessment”, “CAPA Management”, “Document Management”, “Training Management”, “Contractor Management”, “Permit to Work”, “Safety Observation”, “Occupational Health”, “Environmental Management”, “ISO Management”, “Compliance Management”, “ESG Reporting”, “Carbon Accounting”, “Supplier Evaluation”, “Equipment Maintenance”, “Asset Management”, “CRM Integration”, “Workflow Automation” or “Access Control Integration”, with a sourced integration, API or connector capability in its record, or one the vendor stated to us, labelled as vendor-stated. Documented and conditional capabilities qualify; check the linked scope before treating a module as included.

    Tekmon pays for a sponsored placement (above); its place in this list follows the same rule as every entry. How lists are ordered

    All 98 matching profiles are shown. Documentation review, not hands-on testing or a claim of compliance.