Most reporting dashboards lead with the number of reports. It is easy to collect and hard to read, because a rise can come from more people reporting or from more things going wrong, and from the count alone you can’t tell which.
Below: the standard definitions of leading and lagging indicators, then the eight measures for frontline reporting we set out in The Frontline Reporting Index 2027, with what you can learn from each one and where the data usually sits.
What are leading and lagging indicators in safety?
Leading indicators are measures of the activities meant to prevent harm. Lagging indicators are counts of harm after it has happened.
In its guide, OSHA defines leading indicators as “proactive, preventive, and predictive measures that provide information about the effective performance of your safety and health activities.” Lagging indicators, in the same guide, “measure the occurrence and frequency of events that occurred in the past, such as the number or rate of injuries, illnesses, and fatalities.”
In HSG254, written for chemical and major hazard industries, the UK Health and Safety Executive draws the same line around control systems: “Leading indicators are a form of active monitoring focused on a few critical risk control systems to ensure their continued effectiveness.” Lagging indicators “are a form of reactive monitoring requiring the reporting and investigation of specific incidents and events to discover weaknesses in that system.”
Reporting sits on both sides of that line. The reports themselves feed lagging measures. How the reporting works, who can do it, how long it takes and whether anything comes back, is a leading measure, and it gets less attention than the count.
Why is the number of reports a weak KPI on its own?
Because the same number can mean opposite things. A site that doubles its near-miss reports may have made reporting easier, or it may have more near misses. A site with few reports may be safe, or it may have a reporting path nobody uses. We went through what a near-miss report should contain in our near-miss guide. In the report we wrote: “More records may reflect more participation, changed recording practice or more events. The count alone cannot distinguish them.”
From the count you learn nothing about the record itself. A report that reads “spill at loading bay” counts the same as one the next person can act on.
Eight measures for frontline reporting
In The Frontline Reporting Index 2027 we define eight measures, each with a definition and a unit. They are separate measures. We don’t combine them into a score, and “index” in the report’s title means this set of measures.
| Measure | Definition and unit, from the report | What you learn from it |
|---|---|---|
| Frontline Workflow Burden | Worker, supervisor, correction and duplicate-entry labour to create accepted records. Hours per 100 frontline workers per year. | Where the reporting effort sits, including the effort that moves to supervisors |
| Work-to-Record Lag | Elapsed observation-to-usable-confirmed-record time. Median minutes or hours. | How long information takes to become something others can use |
| Observer-Initiated Reporting Rate | Accepted records initiated by the observer. Percent of accepted records. | Whether records start first-hand or get written up by someone else |
| First-Pass Acceptance Rate | Submitted records accepted without clarification, correction or re-entry. Percent of submitted records. | How often a report has to go back to the person who made it |
| Reporting Distance | People, systems and handoffs between worker and committed record. Median handoffs and systems. | How many places the account can lose detail on the way |
| Assistance-Required Rate | Reporting attempts needing help due to language, access, complexity or uncertainty. Percent of attempts. | Who can report without help, and who can’t |
| Feedback Closure | Reporters told what happened after submission. Percent of reporters. | Whether reporting is a one-way channel |
| Modelled Capacity Opportunity | Baseline labour less an explicitly stated future-state range. Hours per 100 workers and local currency. | What a change could release, with the assumption written next to the number |
The third column is our plain-language reading of each measure. The definitions and units are quoted from page 16 of the report.
Which of these can you measure this month?
Most of them, for one workflow, from data you already hold or can collect in two weeks.
Observer-Initiated Reporting Rate. If your system records both who entered a report and who observed the event, divide one by the other. If it records only who entered it, that is the first finding.
Work-to-Record Lag. Compare the time of the event with the time the record was accepted. In our experience, many systems hold both timestamps without anyone setting them side by side.
First-Pass Acceptance Rate. Count the records sent back for clarification or edited after submission. If the workflow doesn’t log returns, ask reviewers to keep a tally for two weeks.
Feedback Closure. Ask a sample of last month’s reporters whether anyone told them what happened. It takes a week and a clipboard.
Reporting Distance. Map one workflow from the worker to the committed record on a whiteboard, and count the people and systems the account passes through.
Assistance-Required Rate. Ask supervisors to note each time they help someone report, and why, for two weeks.
Frontline Workflow Burden. Time a small sample of reports by role, including the supervisor’s time chasing detail and anyone’s time re-entering it.
Modelled Capacity Opportunity. Only once you have the baseline, and only with the future-state assumption written next to the result.
How do these relate to OSHA’s leading indicator examples?
They start earlier in the chain. OSHA gives examples in the same guide, such as the “Average time between worker report of a hazard or concern and management acknowledgement of the report” and the “Number of worker-reported hazards or concerns for which employers initiated corrective action within 48 hours (compared to the total number of such reports).”
Both of those begin once a report exists. The eight measures start before that, with how much work it took to produce the report and who produced it, and go on to whether it could be used first time and whether the person who made it heard anything back.
How do you measure whether the record is any good?
Separately from the eight. In the report we describe an audit of sampled records: “For each sampled workflow, define the relevant evidence before scoring records: context required for the task, fidelity to the worker’s account, uncertainty preserved, immediate actions distinguished from requested actions, and usability for the receiving team. Use a qualified independent reviewer and report component results.”
Keeping quality apart from process matters, because a process measure can improve while the records get worse. For incident reports, the five-question check in our incident report template article is a quick first pass before a formal audit. We put it this way in the report: “More words, completed fields and first-pass acceptance are not substitutes for factual accuracy.”
Where does Tandm fit?
We wrote the measures so they can be applied to any reporting process, including ours.
Tandm asks the follow-up questions at the job, in the worker’s language, and the worker checks the account before it goes into the system you already run. Those are the parts of the process that First-Pass Acceptance, Assistance-Required Rate and Work-to-Record Lag measure. Whether Tandm moves them in your operation is something to measure in one defined workflow, and in the report we set out how to run that comparison fairly.
Download The Frontline Reporting Index 2027. Free, 17 pages.
Sources
- Occupational Safety and Health Administration, Using Leading Indicators to Improve Safety and Health Outcomes, OSHA 3970, June 2019. osha.gov
- Health and Safety Executive, Developing process safety indicators: A step-by-step guide for chemical and major hazard industries, HSG254, 2006. hse.gov.uk
- Tandm, The Frontline Reporting Index 2027: The work-to-record gap, framework edition, October 2026. tandm.io