Expertise · Safety
Three moments, three kinds of leadership
A person is on the floor without the protection the task requires. Whether the leader negotiates, instructs or asks “why” depends on which moment this is: designing the rule, the minute of danger, or the learning afterwards.
A technician steps over the line, and the shift leader has one second
On a fictional continuous-process site a shift leader sees a maintenance technician walk into an isolated area without completing the lock-out. The leader has two reflexes. One is to stop everything and give a clear order. The other is to ask what is going on and hear the person out. Both are right, at different moments, and wrong at the others. A leader who debates the lock-out in the moment teaches that the standard is negotiable. A leader who has never asked why technicians skip it, and only ever issues orders, learns about the problem after an injury.
The difference is not toughness against kindness. It is a classification: how urgent is the danger, how severe and how reversible is the harm, is the standard non-negotiable, and is there time to deliberate? That classification decides the form that participation takes: collaborative when the rule is designed, directive in the minute of danger, evaluative afterwards.
Five questions that classify a safety decision
- UrgencyIs danger present now, or is this a design question for later?
- Harm potentialHow serious could the harm be, and to whom, including people outside the team?
- ReversibilityCan the harm be undone? A bruised schedule and a lost limb are not the same kind of cost.
- Non-negotiable normsIs there a legal, contractual or site standard that is not up for discussion, whatever anyone prefers?
- Deliberation timeIs there room to talk before acting? If not, the instruction comes first and the conversation after.
Illustration · four things a leader has to hold apart
The same leader, four different jobs
Why the safety rule was not followed
After a near miss, “who did this?” is the least informative question. Four different explanations can produce the same skipped step, and each has a different remedy.
1 · Work as imagined, work as done
Mechanism: the written procedure assumes conditions the job rarely has, and the skip is an adaptation that makes the job possible.Cue: skips cluster at the same step and the same equipment whoever is on shift, and experienced people describe the same workaround.Move: walk the job with the people who do it, then change the procedure, the equipment or the planning, not the person.2 · What the leader asks first
Mechanism: when a line stops, the first question is “when are we running again?”. People infer which priority wins.Cue: ask what the shift leader asks first after a stoppage, and ask three operators what they think the leader would say about delay versus safety.Move: fix the first question as “what happened, is everyone safe?” and ask it every time, including when it is inconvenient.3 · Voice is not safe
Mechanism: people do not report near misses or doubts because the last person who did was blamed or ignored.Cue: low reporting is ambiguous, since it can mean few events or silence. Compare reports with what operators say in confidence, and check how specific the reports are.Move: answer every report within a set time, say what changed, and protect the reporter’s standing.4 · Knowing disregard
Mechanism: a deliberate breach of a clear standard by someone who understood the risk, with no obstacle that explains it.Cue: the first three have been ruled out, and the standard was clear, workable and trained.Move: a proportionate, documented response under the existing disciplinary and legal rules. This is the exception, and it is judged last.The order matters. Starting at number four can overlook causes in one, two or three.
The work
What we do on a safety engagement
- Classify the decisionsWe sort the recurring decisions on your site (stop-work, permit, escalation, refusal) with the five questions. You get: a table showing where people are consulted, where the answer is fixed, and who decides.
- Walk the workWe follow selected tasks as they are actually done, with the people doing them, and compare with the written procedure. You get: a short list of gaps between procedure and practice, each with a candidate cause.
- Design participation before the eventSessions in which the people affected test the reasons for a rule and challenge it. Challenge is welcomed and recorded; non-negotiable norms stay fixed, and we say so at the start. You get: revised procedures, or a clear explanation of why not.
- Write the acute-risk instructionA short, plain script for the minute of danger: who speaks, what is said, what happens on refusal. You get: a one-page protocol and a rehearsal with the leaders.
- Set up the learning reviewA way of asking after an event that separates honest error, at-risk habit and reckless disregard, and gives each a different response. You get: a review template and a briefing for those who lead reviews.
How we know it is working, and how we avoid fooling ourselves
Agreed before the work starts
- Leading indicatorsNear-miss reports: how many, how specific, how fast answered, and what share led to a visible change.
- Behaviour at the standardAdherence at the agreed non-negotiable steps, sampled without warning and reported without names.
- Lagging indicators, with careInjury counts on one site are small numbers and move by chance. Three quiet months prove very little, so we read them over long periods and next to the others.
- A pre-agreed triggerIf reports fall while observed adherence does not improve, we treat it as silence, not safety, and look at voice first.
A falling number of reports is good news only after you have found out why it is falling. That is the single most common misreading we expect to correct.
How certain are we?
Research and practice in safety
- EstablishedPerceived psychological safety, and leaders’ visible priority on safety, are associated with more reporting and safer outcomes.
The association is replicated in many settings. The causal direction and the size of the effect are less certain, so we test locally.
- Plausible applicationSeparating honest error, at-risk habit and reckless disregard when responding to events supports learning (the logic usually called just culture).
Widely used and sensible. Controlled evidence on injury rates is limited, so we measure reporting quality as well.
- Our hypothesisClassifying a decision by urgency, harm, reversibility, fixed norms and time makes the right form of participation obvious, and removes the false choice between a tough and a soft leader.
We will report it if site leaders find the classification unhelpful or wrong.
What safety advice cannot replace
- A legal or regulatory safety audit. Statutory duties, prevention advisers and occupational physicians keep their roles, and we work alongside them.
- Technical risk analysis of plant and processes. That is engineering work; we deal with the human and organisational side.
- A promise of fewer incidents. We promise an honest look at what is going on and a measured test of what we change.
- A reason to negotiate a non-negotiable standard, or to hand out blame in the name of “learning”.