Leading Indicators & Board-Level Safety Metrics
|
I have sat in enough board safety reviews across Western Europe and the Arabian Gulf to know the format without needing to refer to the agenda. Ten to fifteen minutes, HSEQ, one slide. The slide shows the same three numbers almost everywhere I go: total recordable injury rate, lost time injury frequency rate, and a fatality count, usually zero. The chairman thanks the safety director. The meeting moves to capital expenditure. Nobody in that room asked why the near-miss reports had dropped by a third over the quarter. Nobody asked when a critical control on the highest-risk process was last physically verified, rather than assumed present because a procedure said it should be. Nobody asked how long a corrective action sits open before it gets closed. The board reviewed a dashboard. It did not learn whether the next incident was already forming somewhere on the plant floor, because nothing on that slide was built to tell them that. SECTION 1: THE CASEThe slide is not wrong. TRIR (Total Recordable Injury Ratio), LTIFR (Lost Time Injurt Frequency Ratio), and a fatality count are real numbers that are correctly calculated, and every regulator in every market where Technique Works operates wants to see them. The problem is what a board concludes from them, which is usually, 'Nothing has gone badly wrong, so the operation is under control.' That conclusion does not follow. Every number on that slide is a lagging indicator: It measures something that has already happened. A fatality count of zero tells a board it has been fortunate for the period measured. It says nothing about whether that fortune is the product of genuine control or the product of gaps that simply have not lined up yet. Two organisations can show an identical zero on that line: one because its barriers, supervision, and verification are doing their job, and the other because the same gaps that will eventually produce an incident have not yet been tested by circumstance. The slide cannot tell the difference, and neither can the board reading it. None of these factors means your zero is fake. Some organisations earn a flat zero honestly through controls that genuinely work, and if that describes yours, nothing here is aimed specifically at you. The problem is structural, not personal: the number by itself cannot tell a board which kind of zero it is looking at, and most boards have no second number to check it against. There is a second problem, one that HSEQ professionals are often reluctant to acknowledge. When a board sets zero as the target and treats a flat zero as a positive result on its own, it creates exactly the incentive structure I have described elsewhere as the Watermelon Principle: green on the outside, red on the inside. A workforce under pressure to protect a zero learns quickly that the fastest way to do so is to stop reporting the near misses that would put it at risk. The metric does not improve. The organisation's visibility into its own risk collapses, and the board, watching the one number it knows how to read, has no way of noticing it. This scenario is not a hypothetical. It is the reason a genuinely proactive safety culture and a genuinely honest reporting culture rise and fall together, and it is the reason a flat, unchanging zero across several consecutive reporting periods should prompt a harder question than most boards ask: is this real, or is this the number we have trained people to produce for us? Recommended Reading: The Case
SECTION 2: THE EVIDENCEThe distinction between a lagging and a leading indicator is well-established and uncontroversial in the professional literature, but it often fails when faced with a board agenda. Many safety professionals invoke Heinrich's pyramid to justify investigating every minor injury and near miss with the same intensity as a fatality, on the theory that enough small fixes eventually prevent the big one. That is a misreading of the original argument. Heinrich's 1959 data described a ratio, roughly one major injury to twenty-nine minor injuries to three hundred incidents with no injury at all, drawn from events that all carried the potential for serious harm. The point was never that every paper cut matters equally. The point was that the near misses carrying real energy behind them are the earliest, cheapest, and most frequent signal an organisation gets that a serious event is building, if anyone is counting them and asking why the count moved. That is precisely what a leading indicator is built to do: it counts something before harm occurs, not after. Near-miss reporting trends, tracked separately for significant and general near misses, are one example. So is the critical control verification rate, which confirms that a control is actually functioning this quarter rather than merely being documented on file somewhere. Behavioural observation matters too, not as a box-ticking exercise but as a record of what supervisors are genuinely seeing and reinforcing on the floor. And corrective action closure velocity tells you something the other three cannot: whether a known gap gets fixed in days or quietly ages for a year because nobody owns it. None of these numbers require an injury to exist. That is what makes them predictive rather than historical. None of the four requires new software to start, and none of them requires Technique Works, or any consultant, to build. A critical control verification check is someone with the authority to stop work walking the highest-risk task this week and confirming, in person, that the specific control, the barrier, the interlock, the permit, is physically present and functioning, and then writing down what they found in a notebook if that is all there is. That is the entire method at its simplest. A system helps it scale across a dozen sites. It does not replace the walk; the walk is where all four of these measures actually start. For over a decade, the industry's own standard-setters have been formalising this distinction, and the argument is not going quiet. The American Petroleum Institute published the fourth edition of RP 754 in August of this year, fifteen years after the first, still organised around a four-tier structure that separates indicators suitable for public reporting from indicators meant for a facility's internal use, and leading measures that signal a developing weakness from lagging measures that confirm one already failed. The UK's Health and Safety Executive reached a similar conclusion earlier: HSG254, its guide to developing process safety indicators, lays out a six-stage methodology aimed explicitly at senior management, not the HSEQ department, because a board that only sees lagging data cannot exercise the oversight the role requires. ISO 45001:2018, Clause 9.1, requires an organisation to determine what needs to be monitored and measured to know whether its management system is actually working; a requirement that a slide with three lagging numbers on it does not satisfy on its own. None of this information is exotic. It has been sitting inside the standards Technique Works' clients are already certified against. Most boards have simply never seen the half of it that would tell them something useful before the fact. Recommended Reading: The Evidence
CASE STUDYPicture the same board and quarter, with a second slide added next to the first, not instead of it. The first slide stays exactly as it was: TRIR, LTIFR, and fatality count. The second slide shows the numbers the first one never captured. The near-miss reporting rate, up or down against the prior quarter, with a one-line note on why. The critical control verification rate for the site's highest-consequence operations is the percentage of controls physically confirmed functioning in the period, not assumed functioning because a procedure exists. The behavioural observation completion rate, showing whether supervisors are actually out on the floor doing what the programme says they do. The corrective action closure velocity is the median number of days between when a gap is raised and when it is closed. A board looking at both slides together can ask a different question than "Did anything go wrong?" It can ask, "Is our risk exposure moving in a direction we should be worried about?" months before that exposure produces an incident that the first slide will eventually report. The cost of not asking that question is concrete. The International Council on Mining and Metals, whose member companies run some of the most heavily audited critical-control programmes in heavy industry, reported a rise in member fatalities for two consecutive years, from 33 in 2022 to 36 in 2023 and 42 in 2024, reversing a long-term decline on which the sector had built its safety reputation. Nine of the twenty-four member companies reported zero fatalities that year. At board level, in nine separate boardrooms, the lagging number said the same thing the boardroom described at the start of this edition had heard: nothing to worry about. The trend across the same membership said otherwise, clearly enough that ICMM's own leadership named it publicly and committed to revising its Critical Control Management guidance, publishing the updated edition in 2026. Mining is different from petrochemicals, pharma, or logistics, and the hazards are distinct. What transfers is the mechanism, not the industry: the same twenty-four companies, reporting the same way, year after year, watching an aggregate number move in the wrong direction while most of their individual boardrooms had nothing but good news on their slide. It is also worth being honest about what the figure actually is. Thirty-three to forty-two is a raw count, not a rate. ICMM does not publish it against hours worked or headcount, so some part of the rise could reflect more exposure rather than worse control. That gap is not a flaw in the argument. It is the argument. Even one of the more data-mature safety bodies in heavy industry is working from a number one denominator that is short of being a genuine leading signal, which is the condition most boardroom dashboards are in, just with fewer zeros after the decimal point. Zero, at the level of a single site, in a single quarter, is not a finding. It is a data point, and a data point needs a trend and a leading signal around it before a board can tell whether it means anything. Recommended Reading: The Indicators
SECTION 3: THE PRINCIPLEJune's edition argued that the business continuity plan belongs to the business, not the department, and that the CEO carries the accountability once the threshold from operational to organisational is crossed. July's edition made the same argument about the investigation that follows a serious event: the finding an organisation produces is a leadership decision, not a technical one. This month's argument sits directly next to both. The dashboard a board reviews is also a leadership decision, made or defaulted into, about what the organisation is willing to know before something happens versus what it is willing to find out only after. A board that reviews three lagging numbers and calls the meeting complete chooses not to prioritise safety. It has chosen not to look and called the absence of bad news a result. There is a cost to that choice beyond the obvious one of a missed warning. The market in which Technique Works' clients operate is already pricing leading-indicator maturity, whether a board tracks it internally or not. Cefic's SQAS scheme, the assessment chemical companies use to select logistics providers, evaluates verification and system performance, not only incident history, because the companies commissioning those assessments learned some time ago that a clean incident record with no visibility into how it was produced is different from a controlled operation. A board that can only produce lagging numbers when a customer, an insurer, or an auditor asks for evidence of proactive management is not just carrying safety risk. It is carrying commercial risk; it may not know it has. An organisation that reviews only lagging indicators finds out about its risks the same way everyone else does: after the fact. An organisation that reviews leading indicators alongside them gets a chance to find out before, and that chance is the only thing any of these frameworks – ICAM, Tripod Beta, API's tiers, and HSE's six stages – have ever actually offered. None of them prevent an incident by themselves. They improve the chances that someone with the authority to act sees the warning in time to use it. Recommended Reading: The Principle
HSEQ MARKET INSIGHTS: AUGUST 2026Four data points are currently shaping the board-level safety metrics conversation. 1. API just revised the leading/lagging standard again this month. On 6 August 2026, the American Petroleum Institute published the fourth edition of RP 754, updating a four-tier framework first published in 2010 that separates leading indicators from lagging ones and public reporting from internal use. Fifteen years and four editions in, the industry that formalised modern process safety metrics is still refining how to measure what predicts an incident rather than just what records one. Source: American Petroleum Institute, RP 754 Fourth Edition announcement, August 2026. 2. A sector built on critical control verification still got the trend wrong for two years running. ICMM member companies, among the most heavily audited critical-control programmes in heavy industry, recorded 33 fatalities in 2022, 36 in 2023, and 42 in 2024, reversing a long-term decline. Nine of twenty-four member companies reported zero for the year. The aggregate trend moved in the wrong direction while a majority of individual boardrooms were looking at a number that told them otherwise. It is a raw count, not a rate; ICMM does not publish it against exposure hours, which is itself worth noting. Source: ICMM, 2024 Safety Performance of Members, icmm.com/en-gb/news/2025/2024-safety-performance; ICMM Critical Control Management Good Practice Guide, 2026 edition. 3. The guide written specifically for boards has existed since 2011 and remains largely unread outside major-hazard sites. HSE's HSG254 lays out a six-stage process for developing process safety indicators, addressed explicitly to senior management rather than the safety department. Adoption outside COMAH and major-hazard operations, the sectors it was written for, remains limited, which means most boards outside oil, gas, and chemicals have never been handed a structured method for building the dashboard this edition describes. Source: HSE, HSG254, Developing Process Safety Indicators. 4. The commercial market is already scoring leading-indicator maturity, whether your board tracks it or not. Cefic's SQAS scheme, used by chemical companies across Europe and increasingly the Gulf to evaluate logistics providers, assesses verification and system performance alongside incident history. A provider with a clean lagging record and no evidence of proactive management does not automatically win the tender. Source: Cefic, SQAS, Safety and Quality Assessment System. QUESTIONS FOR YOU TO CONSIDERFive questions grounded in this edition's argument. They are not rhetorical. They are diagnostic. 1. Pull the last HSEQ slide your board actually reviewed. List every number on it. How many of these measures reflect something that has already happened, and how many would have indicated that something was coming? 2. Do you know whether your near-miss reporting rate rose or fell last quarter, and could you explain why in one sentence, without guessing? 3. For your highest-consequence operation, when was a critical control last physically verified as functioning, rather than just assumed to be present because a procedure states it exists? 4. How long, on average, does a known corrective action sit open in your organisation before it is closed, and does anyone above HSEQ level know that number? 5. If your fatality count, or your most serious incident count, has been flat at zero for several reporting periods running, can you say with confidence that the result reflects genuine control rather than a workforce that has learned not to report? If the honest answer to any of these is "I would need to check," the dashboard question is already answered. PRACTICAL ACTIONFour steps, in sequence. The first one takes ten minutes. Step 1. Pull the HSEQ slide from your last board pack. Mark every metric on it with an L for lagging (something that has already happened) or a P for predictive (something that would have told you before it happened). Count each column. Step 2. If the slide is entirely, or almost completely, L, that is the finding. Do not add commentary. Just note the ratio. Step 3. Choose whichever of the four predictive measures from this edition is weakest, or entirely absent, in your organisation's current reporting, and confirm whether the underlying data already exists somewhere. In most cases it does. It has simply never been extracted and put in front of the people who could act on it. Step 4. Put that one number on the agenda for the next board cycle, alongside the lagging numbers already there, without replacing them. One new number is a change a board will actually engage with. A completely redesigned scorecard is a project that will not survive the next reorganisation. If you haven't yet established the standing to add something to the board agenda, consider starting with smaller initiatives. Run Steps 1 and 2 on your department's reporting and bring the ratio, not a request for a new system, to whoever you report to. A specific ratio is easier to act on than a general argument for better metrics. PERSONALISED RECOMMENDATIONSThe same argument lands differently depending on where you sit. For the CEO / General Manager You are not expected to design your organisation's safety metrics. You are expected to know what your current dashboard can and cannot tell you. A slide made up entirely of lagging numbers is not evidence of control. It is an absence of evidence, dressed as reassurance. Before your next board cycle, ask your HSEQ director one question: which of these numbers would have told us about a problem before it happened, and which ones only confirm one already did. If the honest answer is none of them, that is worth knowing before the number that has been flat for two years finally moves. For the COO / Operations Director The four predictive numbers this edition describes mostly already exist somewhere in your operation. They are collected at the site level and rarely aggregated or escalated, because nobody above the site level has asked for them in a form they can use. Identify where that data already lives before assuming it needs to be built from nothing. For the HSEQ Director This is the argument that moves your function out of the reporting business and into the risk-management business. A board that only sees lagging numbers will hold you accountable for outcomes you had limited ability to prevent in real time. A board that sees leading numbers alongside them can hold the organisation accountable for the conditions that produce outcomes, which is a fairer test of your work and a stronger position for you the next time a serious event forces the question of what the organisation knew and when. None of the above is a verdict on your current team, and forwarding it is not the same as reporting your department for a failure. Almost every board carries this exact gap, because almost none of them have ever seen what a second slide looks like. Whoever runs the finding owns it. Nobody has to have been wrong before now for it to be worth doing. NEXT MONTHSeptember's edition follows the arc directly. A leading indicator that flashes and gets no response is not an improvement on a lagging indicator. It is the same failure, arriving earlier, with a warning nobody acted on. Knowing your near-miss rate fell 15% this quarter is not the same as knowing why, nor is it the same as someone with the authority to change something, actually changing it. Most organisations that build a leading-indicator dashboard stop at the dashboard. The number moves, someone notes it, and the meeting moves on to capital expenditure, the same way it always did, just with an extra line on the slide. September's argument: a dashboard nobody is authorised to act on is not a leading indicator. It is a lagging indicator with better timing. CALL TO ACTIONEach edition of Technique Works HSEQ Insights is published monthly at insights.techniqueworks.com. If you find this edition useful, we encourage you to subscribe to receive it directly. The full archive, including July's edition on incident investigation, June's on business continuity, and the complete series from November 2025, is there now. If it belongs in your GM's inbox, your COO's, or your HSEQ director's, forward it as is. Subscribe to insights.techniqueworks.com Amador Brinkman · Technique Works |