Leading Indicators & the Authority to Act
|
Reported, Rewarded, Never Closed Last week I posted a question on LinkedIn that I ask in almost every boardroom I sit in: zero overdue actions on the dashboard, and the room relaxes, so what does "closed" actually mean here? The replies came from HSEQ managers, operations people and an HR professional, in different industries and different countries, and they described the same thing. One wrote that a dashboard with zero overdue actions "makes leaders smile while the risk remains." Another made the point more precisely than most audit reports do: a tracker can show 90 per cent closure because ninety low-priority actions were closed, while the ten high-priority ones were never touched. Then, this week, I saw the pattern on the ground. The site belongs to a large multinational group, with a reporting programme that most organisations would envy. People report near misses and good saves. They are evaluated on how much they report, and they are rewarded for it. The front end works, and it works because somebody designed it to. Follow a report past the moment it is submitted, and the design ends. Items are closed in the system and come back a few months later, worded almost the same way, because the condition that produced them never changed. The person who reported it hears nothing. And the volume is now so large that the organisation has quietly accepted it as too much to act on. Nobody said so. Nobody had to. SECTION 1: THE CASEThis is the half of the leading-indicator argument that August's edition left open. August argued that most boards review a dashboard built from lagging numbers and that leading indicators, near-miss trends among them, would warn them before an incident instead of after. That argument stands, and it ended with a specific recommendation: put one leading number on the board agenda next to the lagging ones. That recommendation also stands. This month is the part that makes it worth doing, because a number on an agenda with nobody attached to it changes nothing at all. But a near-miss report is only a leading indicator if something happens because of it. A report that is counted, rewarded and never closed tells a board that people are reporting. It says nothing about whether the risk they reported is still there. A lot of organisations treat reporting volume as proof that their safety culture is healthy. As a matter of fact, reporting volume measures the front end of the system only. It tells you people are willing to write things down. It does not tell you whether writing things down changes anything, and after a year of nothing changing, the willingness goes too. Look at how most of these systems are built, and the imbalance is obvious. Reporting has a target, a form, an app, a monthly count and, often, a reward. Closure has a status field. Somebody, usually a supervisor, changes "open" to "closed", and the dashboard turns green. Nobody checks whether the condition changed, nobody checks whether the control works, and nobody goes back to the person who raised it. Rewarding the count makes it worse. When people are evaluated on the number of reports, they learn to produce reports: easy ones, safe ones, and the ones that will not embarrass a supervisor. The difficult report, the one about the shortcut everyone takes because the schedule demands it, gets written less often because nobody has seen that kind of report go anywhere. The count goes up. The information in it goes down. It is the watermelon again, green outside and red inside, only this time the green is the reporting rate. Then there is the boardroom, which is where this becomes uncomfortable. Start with the structure rather than the people. HSEQ gets ten minutes and one slide four times a year. The person preparing that slide remembers what happened to the last colleague who presented a red number. Everything the board sees has passed through people who know what the board rewards. By the time it reaches the table, it has been filtered three times, and nobody in that chain had to do anything dishonest to filter it. Boards like green. That is not a criticism; it is how the incentive works. Green means the operation is under control and nothing needs doing. Red means there is a problem; someone has to go and solve it, and the person asking about it may be asked to own it. Many board members are not close enough to the operation, not often enough on the floor, and not given enough detail to know what sits behind the colour. So they get sold green, and they buy it, because green is the only colour that does not come with homework. None of that requires anyone to lie. It only requires that nobody in the chain is obliged to look underneath. Recommended Reading: The Case
SECTION 2: THE EVIDENCEThe clearest public record of this failure is twenty years old, and it still reads like a description of a site visit this month. At three minutes past three on a Sunday morning in December 2005, the level gauge on Tank 912 at the Buncefield oil storage depot, north of London, stopped moving while petrol kept flowing in. The independent high-level switch behind it was inoperable too. At 05:37 petrol began pouring from the vents; by 06:01 around 250,000 litres had escaped, and the vapour cloud ignited. The fire engulfed more than twenty tanks and injured over forty people. Nobody died, largely because it was early on a Sunday. The gauge had stuck fourteen times since the end of August. Supervisors had a fix for it, raising the gauge and letting it settle, a practice the site called "stowing". Sometimes the fault went into the log; sometimes it did not. The Competent Authority's investigation found that the Operations Manager was unaware of how often the gauge was failing because the fault log was not used consistently, "especially where the symptoms of a defect were apparently remedied quickly". That is the back-end problem in one sentence. Each fault was closed on shift. Nobody saw fourteen of them together. Its conclusion is worth reading slowly: "Had such a framework been in place, the measurement of a number of relatively simple indicators would have alerted management to the underlying problems that led to the incident." Relatively simple indicators. Collecting was never the hard part. HSE wrote the fix down in 2006, in HSG254, and it is three sentences long. "A tolerance should be set for each leading indicator." "The management team should set the tolerance, not the person responsible for the activity." And the one most dashboards skip: "Deviations from tolerances must be followed up; otherwise, there is little point in collecting the information." The standards most of these sites are certified against say the same thing. ISO 45001:2018, clause 10.2, does not stop at taking corrective action; it requires the organisation to review the effectiveness of any action taken. A status field changed from open to closed is not a review of effectiveness. Clause 5.1(j) makes it a leadership duty to protect workers from reprisals when they report incidents, hazards, risks and opportunities, and clause 5.3 requires top management to assign responsibilities and authorities. Seveso III, Annex III, which sits behind the Dutch BRZO and every major-hazard regime in the EU, requires the safety management system to cover near misses, "particularly those involving failure of protective measures, and their investigation and follow-up". Follow-up, effectiveness, authority. The words are already in the documents your auditors check. What is missing, in most organisations, is the machinery behind them. Recommended Reading: The Evidence
CASE STUDYIn July, at least five men were killed when a gas cylinder exploded in a residential district of a Gulf capital. At the time of writing, the cause has not been published, and this edition makes no assumptions about it. I mention it because anyone who has run HSEQ for an operator with a large workforce in the region will recognise the category and will know that the near-miss reports on that category rarely stop at one. Consider a composite, drawn from a pattern I have seen at more than one site in the Gulf: a chemical logistics operator, a few hundred drivers and warehouse staff, most of them employed through contractors, many of them far from home, housed in company-arranged accommodation. The operator runs a near-miss card programme and runs it well on paper. Cards are counted weekly. There is a reporter of the month. Reporting rates are high, and the regional board sees them rise quarter after quarter. Among the cards, the same subject kept returning: gas cylinders stored where they should not be, in the loading bay and in the accommodation kitchens. It was reported and discussed with the supervisor and marked closed. Reported again four months later. Closed again. Reported a third time. Each closure was genuine in the sense that someone moved the cylinders that week. The reason the cylinders kept ending up there remained unchanged: lack of designated storage, no ownership, and no follow-up a month later. In this region there is a second layer to the back-end problem. A contractor's worker, a long way from home, whose contract depends on not causing trouble, does not escalate past his supervisor. Stop-work authority exists in writing. In practice, the report card is the most he will do, and when the same card has come back three times with no visible result, he stops writing that one and writes an easier one. The count stays up. The useful information disappears. The changes that worked were not expensive. Cards were triaged by potential severity, not counted. The operator adopted what one of last week's commenters described: rank actions by potential consequence so that ninety closed low-priority items can no longer hide ten open high-priority ones. Triage invites its game, and it is worth naming before someone else does. If a high rating starts a clock whereas a low rating does not, ratings drift downwards on their own. This operator handled it in the cheapest way available: the rating is set by two people rather than one, and every month the site manager reads ten cards rated low. That costs him twenty minutes, and it is the only reason the ratings hold. Every high-potential item received a named owner, the manager of the area concerned, with a fixed number of working days to return one of three decisions: fixed and verified in the field; accepted in writing with a compensating measure and a date; or activity stopped. Any item closed and reopened was reported to site leadership automatically, because a second report on the same subject is proof the first closure was incomplete. The reward shifted from the number of cards to cards that led to a verified change. And every reporter received an answer within a set period: what was found, what was done, and, where nothing could be done yet, why. The reporting count did not disappear from the group report, because group HSE sets a target for it, and a site does not have the option to opt out. What changed was what the site said about it. The card count is still there. Next to it sits the number of cards that produced a change someone verified in the field and two lines from the site manager explaining why the first number fell while the second one rose. The group accepted this because the second number had never existed before. The load is worth stating plainly, because most managers assume it is unaffordable and stop there. Only high-potential cards get the full treatment, which at this operator meant twelve to fifteen a quarter across the site. Each one costs the area manager roughly twenty minutes: reading it, deciding, and confirming the control on a walk he was already doing. The monthly meeting got shorter rather than longer because the conversation moved from what the number is to what the owner decided. That last step matters more in the Gulf than anywhere I have worked. Being answered, by name, in front of peers is recognition that costs nothing, and in a workforce that rarely expects to be heard, it does more for the next report than any reward scheme. The first effect was fewer cards. The site leadership had been told in advance to expect it, which is the main reason the change survived. The cards that did come in were the ones worth reading. Recommended Reading: The Response
THE PILE YOU ALREADY HAVEEverything above describes what happens to the next report. It does not touch the several thousand already sitting in the system, and that pile is where most people stop reading. They are not arguing with the logic. They are looking at a backlog nobody could work through in a year and concluding, quietly, that the whole subject is unaffordable. That conclusion is what is actually protecting the status quo, and it deserves a direct answer. A backlog is not a work problem. It is a decision problem, and it can be settled in an afternoon. Sort every open item by potential severity, using the same two-person rating. The top tier, which in most organisations is a small fraction of the total, goes into the system described above: named owner, response time, and three permitted decisions. Everything else is assigned one documented decision, taken by the management team rather than by whoever is nearest: these items are closed as a batch, on this date, on the basis that their potential consequence is low and the conditions behind them are accepted as they stand. Write down what would bring one of them back, which is usually a repeat report or a change in the operation. Then tell the people who raised them. One message that states what was decided, on what basis, and what the rule is from now on is more effective for the next twelve months of reporting than any campaign. Two objections come back, and both have answers. The first is that closing items in bulk looks worse in an audit. It looks considerably better than five hundred items sitting open for three years with nobody's name against them, which is what an auditor finds today. A decision that is documented and taken by the right people is defensible. Silence is not. The second is the fear of closing the wrong one. That is what the two-person rating and the return rule are for, and it is why the top tier is handled individually. The risk of batching low-potential items is real and small. The risk of a backlog that nobody reads is real and large, because the serious item is in there too, invisible among the rest. It is the Buncefield gauge, filed fourteen times and never considered a pattern. A backlog is a decision nobody has made yet. Make it once, in the open, and the system in front of you starts working. SECTION 3: THE PRINCIPLEJune's edition argued that the business continuity plan belongs to the business, not the department. July argued that an investigation's finding is a leadership decision. August argued that the dashboard board review is a leadership decision too. September closes the sequence. Designing what happens after the report is also a leadership decision, and it is the one that decides whether the other three mean anything. A reporting system has two halves. The front end collects the card, the app, the target, and the reward. The back end closes: a line agreed in advance at which a report becomes an instruction, a named owner who receives it, a clock on the response, a check in the field that the condition really changed, and an answer to the person who raised it. Most organisations have spent years and serious money on the first half of this process. The second half usually consists of a status field and good intentions. This matters to a board for a reason beyond the next incident. Picture the data room. A buyer's HSEQ adviser, or your own, exports the action tracker and sorts it by subject. The same finding closed three times in two years. High-potential near misses marked closed with no field verification. Reporters who never received an answer. A closure rate of 96% adds to this. That adviser will not read your tracker as evidence of control. He will read it as a record of what the organisation knew, when it knew it, and what it chose not to do. That rarely moves the headline price, and anyone who says it does hasn't experienced one of these. What it does is slower and more irritating: the process stalls while the buyer asks for evidence you cannot produce, money goes into escrow, a specific indemnity appears in the agreement, or a remediation line lands in the capex plan the buyer is already modelling against your first year. All of it arrives at the worst possible moment to be discussing it. An open item on a tracker is uncomfortable. An item closed three times and still open underneath is a liability with a date stamp on it. None of this needs a new system. It needs the people at the top to stop rewarding green and start asking what is underneath it. Most boards have never seen the difference because subordinates had no reason to show it. Recommended Reading: The Principle
HSEQ MARKET INSIGHTS: SEPTEMBER 2026Four reference points for anyone whose reporting programme is better at collecting than at closing. 1. The practitioners are describing the same gap from different directions. Responses to a September post on what "closed" really means came from HSEQ, operations and HR backgrounds. The recurring points are closure without verification, actions ranked by count rather than consequence, and findings that reopen because the first closure was administrative. One commenter's framing is worth repeating at board level: 90% closure can mean the one action that mattered is still open. Source: comment thread on "Zero overdue actions on the dashboard", Amador Brinkman, LinkedIn, September 2026. 2. HSE's guidance says who sets the threshold, and it is not the person being measured. HSG254 requires a tolerance for each leading indicator, set by management, "not the person responsible for the activity," and states that deviations "must be followed up; otherwise, there is little point in collecting the information." Source: HSE, HSG254, Developing Process Safety Indicators, 2006. 3. ISO 45001 asks for effectiveness, not closure. Clause 10.2 requires organisations to review the effectiveness of all actions taken, including corrective actions. A tracker that records only open and closed cannot show that a review took place. Source: ISO 45001:2018, clause 10.2. 4. EU major-hazard law expects near misses on protective measures to be followed up. Seveso III, Annex III, requires the safety management system to cover near misses, "particularly those involving failure of protective measures, and their investigation and follow-up." The UK's COMAH Regulations 2015 carry the same wording in Schedule 2, and the Dutch BRZO transposes the same directive. Source: Directive 2012/18/EU, Annex III; COMAH Regulations 2015, Schedule 2. QUESTIONS FOR YOU TO CONSIDERSix questions. They are diagnostic, not rhetorical. 1. How many near-miss reports did your organisation receive last quarter, and how many of those led to a change someone verified in the field? If you know the first number and not the second, that is the finding. 2. Which findings on your tracker have been closed more than once? Who in your organisation is informed when that happens? 3. For your highest-potential reports, is there a named owner and a response time, or does the item belong to "the monthly review"? 4. Ask three people who reported something in the past six months what happened to their report. Do their answers match your dashboard? 5. When your board last saw a red number on the HSEQ slide, what happened to the person who presented it? 6. How many items are open on your tracker right now, how old is the oldest, and who decided they should stay open? If the honest answer to the last part is nobody, you have no backlog. You have a decision that has been put off for several years. PRACTICAL ACTIONFive steps, in sequence, with what each one actually costs. Step 1. Count the reopened items. One afternoon. Export your action tracker and sort it by subject or location. Count the items that were closed and then reported again. Do not explain them yet. The count is the finding, and it is the only slide you need to open the conversation. Step 2. Settle the backlog. Half a day, the management team in one room. Rate every open item for potential severity, with two people per rating. The top tier is moved to Step 4. The rest is closed as one documented batch decision, with the basis written down and the conditions that would bring an item back. Then tell the people who reported them what was decided. Step 3. Test ten closures in the field. Half a day. Take the 10 highest-potential items closed last quarter and walk them with the area manager, rather than doing them yourself. A photo does not count as verification; only a test of the control does. Go as two people checking a system, because going alone turns it into an audit of a colleague, and you will only ever do it once. Step 4. Assign every high-potential item a line, a name and a clock. Twenty minutes per item. The management team sets the tolerance, assigns one named owner to the line, establishes a fixed response time, and allows three permitted outcomes: fixed and verified; accepted in writing with a compensating measure and a date; or activity stopped. On a site with a working reporting programme, the average workload is usually 12 to 20 items a quarter, so the whole thing costs four to 6 hours a quarter spread across several managers. Agree on the escalation in the same meeting: when the clock runs out, the item moves to the owner's manager automatically, with nobody having to chase anyone. Step 5. Close the loop and change the reward. Minutes a week, forever. Every reporter gets an answer within a set period: what was found, what was done, and if nothing yet, why. If your scheme rewards the number of reports, keep the count for the group, and beside it, report the number that produced a verified change. Tell the leadership team in advance that the first number will fall. That warning is what keeps the change alive past its second month. If you cannot run all five, run Step 1 on its own. A count of reopened items, on one page, is very hard for anyone to set aside. PERSONALISED RECOMMENDATIONSFor the CEO / General Manager At your next review, when the closure rate is green, ask one question: how many of these were verified in the field, and how many were closed more than once? Then watch who answers and how quickly. Three things separate a real answer from a confident one. A real answer names the person who made the decision, not the department. It carries a date. And it describes something checked on the floor rather than something updated in the system. Get all three, and your system is working. Get none, and you have your finding without having read a single procedure. You are not expected to design the system. You are expected to stop the organisation from rewarding you with green. The honest answer may be red for a while, and a red number that someone is working on is worth more to you than a green one nobody can defend. For the COO / Operations Director Most of the items on that tracker describe your operation, and most of the owners should be your managers, not HSEQ. The items that reopen are telling you where a fix has been made on shift and never made in the system: the storage that was never designated, the part that was never ordered, and the procedure everyone works around. Start there. Your first objection will be capacity, and it is a fair one, so here is the bill. Only high-potential items carry the full treatment: twelve to twenty a quarter on a typical site, about twenty minutes each, most of it inside a walk your managers already do. Four to six hours a quarter across your management team, and the monthly meeting gets shorter because the discussion stops being about what the number is. Your second objection will be the group reporting target, because a falling card count looks like a failing site. Keep reporting the count. Put the number of reports that produced a verified change beside it, with two lines explaining the relationship. Group functions tend to accept that quickly, because nobody was measuring the second number, and everyone knows the first one on its own is soft. For the HSEQ Director You built the front end, and it probably works. The board and the workforce decide your credibility in the back end. Propose the tolerances, insist the owners are named in the line, measure verified closure rather than status changes, and ensure that every reporter hears back. An HSEQ function that collects thousands of reports it cannot close is carrying risk; it has no authority to remove. The objection I hear most is that you cannot assign work to a production manager and that when nothing happens, you are still the one standing in front of the board. That is precisely why the escalation has to be agreed upon before you need it and why it has to run on its own: when an owner misses the response time, the item moves to his manager automatically, with no message from you attached. Then you are not chasing colleagues, which is the posture that costs an HSEQ function its influence faster than anything else. You built a route, and the route did its job. None of this evidence is a verdict on your team. Nearly every organisation with a mature reporting programme has the same gap, because they designed the programmes to collect. Finishing the design is the missing part. NEXT MONTHOctober moves from the tracker to the equipment. Buncefield had one more barrier behind the gauge: an independent high-level switch designed to stop the flow if everything else failed. It did not work that night because a padlock that held its test lever in position had never been fitted. The supplier had not explained its purpose. The site thought it was there to stop tampering. October's edition is about the Hierarchy of Controls in practice: what happens when an engineered control is quietly defeated or quietly replaced by a procedure someone has to remember. The controls at the top of the hierarchy are the ones that work while nobody is watching. Most sites have fewer of them than their risk assessments claim. CALL TO ACTIONIf this edition described your organisation, or one you sit on the board of, you are not alone in it, and it is a conversation worth having before it becomes an investigation. We are always open to talking it through, with no agenda beyond understanding your situation. Send me a direct message on LinkedIn, write to info@techniqueworks.com, or pick up the phone: +31 (0) 85 212 95 44. Each edition of Technique Works HSEQ Insights is published monthly at insights.techniqueworks.com. Subscribe to receive it directly; the archive, including August's edition on board-level safety metrics, July's on incident investigation and June's on business continuity, 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 · info@techniqueworks.com Technique Works · HSEQ Insights Newsletter · Edition 13 · September 2026 |