SHARE Knowledge Hub

Improving Health and Safety Culture

Improving health and safety culture is about changing the everyday decisions, conversations and behaviours that determine whether risks are controlled in practice. Its mechanism is visible when pressures, defects or bad news meet real work: leaders either resource the controls and listen to workers or allow risky shortcuts to become normal. Weak culture can leave hazards unreported and safeguards degraded until someone is injured or made ill.

Updated 2026-10-19Great Britain and Northern Ireland

Understanding Improving Health and Safety Culture

Health and safety culture is the pattern of shared expectations, decisions, behaviours and working conditions that shapes how people identify hazards, use controls, respond to uncertainty and learn. It is visible in what leaders resource, what supervisors tolerate, whether workers can report problems, how contractors are treated and whether physical and organisational controls work under normal and abnormal pressures.

Why this matters at work

Choose a balanced set of measures. Leading evidence may include proportion of critical controls tested on time, high-risk defect age, safe-stop response, quality of pre-work planning, worker consultation response, competence validation, overdue actions and exposure controls. Lagging evidence includes injury, illness, dangerous occurrence and near miss data, interpreted with reporting quality and exposure. Set a clear definition and response threshold for each measure; a number with no decision attached is decoration.

Harm mechanisms include entanglement, crushing, stored-energy release, fall, dropped load, burn, explosion, toxic exposure, noise and vibration. Potential outcomes include fatal trauma, amputation, fractures, burns, lung disease, hearing loss, dermatitis and musculoskeletal disorders. Prevent through design, isolation, guarding, containment and maintenance before training and PPE; ensure rescue and emergency response are realistic.

How organisations manage this in practice

Use several kinds of evidence: verified repair of important controls, quality and timeliness of responses, worker confidence in raising problems, and the recurrence of similar failures. An increase in near-miss reports may reflect improved trust, increased exposure or both. Examine report content and operating conditions before judging the trend. A fall may reflect prevention or suppression. Management review should ask what remains unsafe, which groups were missed and whether actions survived staff or production changes. Retain the survey method and response rates so later comparisons are meaningful; aggregate sensitive comments to avoid identifying individuals unnecessarily. Set local review points proportionate to risk and bring them forward after an incident, reorganisation or evidence of fear. No culture programme removes the employer's continuing duty to assess and control actual workplace hazards.

Implementation example: the employer resources the control and names its owner; an engineering or technical owner checks design and operating performance; supervisors verify safe conditions and stop or escalate failed barriers; workers use arrangements and report defects; managers verify action closure.

A culture-improvement programme should be a controlled change to the conditions identified in the baseline, not a campaign of slogans. Prioritise immediate uncontrolled risks first, then select a small set of systemic issues such as maintenance delay, weak supervision, under-reporting, poor consultation or unsafe workload. Define the intended change in behaviour and barrier performance, the owner, resources, milestones, worker involvement and evidence of success before launching training or communications.

What are the key facts about improving culture?

Culture improves when daily decisions match the stated values.

Leadership must be visible

People judge commitment by resources, time, decisions and response to bad news.

Consultation is a control

Workers understand work-as-done and often know where controls fail.

Fair discipline protects trust

A just approach separates error, at-risk behaviour and reckless conduct.

Training must change work

Attendance records alone do not prove competence or culture.

Evidence must be acted on

Surveys, inspections and committees only help when actions are closed and verified.

What does improving health and safety culture mean?

It means deliberately improving how people think, decide, communicate and act around risk.

A poor culture often shows itself before an accident: shortcuts become normal, near misses are hidden, supervisors walk past defects, production always wins, training is rushed and worker concerns disappear into silence. A good culture makes risk visible and manageable.

Improvement needs more than a campaign. It needs leadership, competence, worker involvement, communication, monitoring, fair accountability and learning built into the management system.

Leadership commitment shows up when a maintenance backlog is funded instead of deferred again. Worker involvement shows up when a toolbox discussion changes the method statement. These decisions matter because people learn from what is resourced and changed, not only from what a policy says.

Fair accountability investigates the reason for an unsafe choice before assigning blame. Competence is demonstrated when supervisors coach and check real work, while useful communication is task-specific and tested through questions. Together these practices make reporting and correction normal.

Choose the relevant scope

For construction workface decisions, use the separate construction culture and risk-assessment guide. This guide covers general organisational culture improvement.

Construction culture and risk assessment

How can poor culture lead to injury or ill health?

Culture does not injure a person by itself. It changes whether a physical or psychosocial hazard is identified, controlled and challenged before exposure occurs.

For example, production pressure may lead a supervisor to accept a bypassed guard. The source is moving machinery, the route is contact with a hand, and the possible outcome is immediate crush or amputation injury. In a different team, repeated excessive workload and ignored fatigue reports can contribute to longer-developing stress-related ill health and more errors. Near misses, unclosed defects, silence in meetings and repeated workarounds are early warnings. A worker should stop the affected task when a critical control is absent and escalate to the named supervisor; the manager must secure the hazard, investigate the system cause and authorise restart only after verification. See HSE HSG65 Managing for health and safety.

The same pattern can affect contractors and bystanders. Incident counts alone may look low while reporting is suppressed, so combine worker accounts, field checks, control performance and action closure.

Why is culture not posters and slogans?

Posters can support a message, but culture is proved by decisions, resources and response to bad news.

A slogan can remind people of a value, but it does not fix a poor permit system, a weak supervisor, a tired team, a broken guard, an unrealistic programme or a backlog of safety-critical maintenance. If the practical system does not support safe work, people will learn that the slogan is only decoration.

This matters for senior managers and learners because culture improvement must be tied to real controls: leadership routines, consultation, competence, reporting, action close-out, procurement, contractor standards and review.

A poster campaign can raise awareness, but it needs a linked action, an owner and a supervisor check. A stand-down is useful when it removes a barrier and tracks the change, rather than ending as a one-off speech. Behaviour observations should ask how the system made a shortcut attractive instead of making the worker the sole cause.

Leaders make their words credible by aligning budgets, deadlines and accountability with control needs. Reward schemes should recognise useful challenge and verified improvement; rewarding only low incident counts can discourage reporting.

How can a control hierarchy change unsafe routine work?

For repeated bypassing of a machine guard, remove the need to reach the danger zone or redesign the access before relying on a rule or reminder. Workers and supervisors should test the new method on a real shift and stop if protection fails.

Take the repeatedly bypassed machine guard. Eliminate the need to enter the danger zone by redesigning the process where feasible. Use genuine substitution where a less hazardous process replaces the task. Engineer improved feed reliability, fixed guarding and effective interlocking; lower intervention frequency is not itself substitution. Control residual access through isolation, maintenance, permits and reliable checks. PPE can reduce some residual injury but cannot make a moving nip point safe. Discipline means clear authorisation, fair coaching, reporting and verified corrective action—not blaming the operator while the production pressure or defect remains. A manager should stop the line if the guard is defeated, assign a repair owner, record the test and authorise restart only when the protection works.

Control priorityTask-specific option and feasibilityOwner, verification, failure and stop point
EliminationRedesign the task to remove access to the moving danger zone during normal production.Engineering validates output and safe maintenance access; stop if bypass remains necessary.
SubstitutionNo distinct substitution is established in this guard-bypass example. Compare a genuinely less hazardous replacement process where feasible; improving feed reliability and material presentation is an engineering redesign that reduces interventions.Production lead trials cycle and jam rate; reject a change that creates a new manual reach.
EngineeringFit fixed guarding and effective interlocks so people cannot reach moving parts.Engineer commissions and function-tests interlocks; stop on a failed guard or defeated switch.
AdministrativeProvide isolation procedures and authorisation for maintenance, worker consultation, defect reporting and supervisor checks. Physical disconnection, securing against reconnection and proving safe are the isolation barriers; paperwork alone cannot isolate machinery.Manager samples shifts and closes reports; hold machine for repeat bypass or overdue defect.
Personal protectionTask PPE addresses residual cut/splash risk only; it cannot make access to moving parts safe.Supervisor checks selection and condition; do not restart while the guarding barrier is absent.

These are illustrative controls for the stated task, to be selected and verified by a competent local dutyholder. L22 - Safe use of work equipment supports the topic-specific risk route; the exact engineering design and stop decision require site approval.

How are culture and climate different?

This distinction helps learners avoid vague answers.

Safety climate is the current perception of safety in the organisation. It can be measured through surveys, conversations and worker feedback. Safety culture is deeper and more stable: the patterns of assumptions and behaviour that shape what people do when no one is watching.

A climate survey can be useful, but only if the organisation acts on the findings. Repeated surveys without visible action can damage trust.

A climate survey captures how people currently perceive safety. Read it alongside conversations and field observations, then return to workers with actions; a survey without response can weaken trust. Culture is the more persistent pattern behind those answers, changed through repeated leadership decisions and work design.

Look for different subcultures across shifts, departments and contractors. A workaround, a quiet near miss or a repeated audit finding may be an early signal that one group’s controls have drifted. Trust grows when concerns receive a decision and feedback, not just a place in a database.

What does weak culture and strong culture look like?

The simplest test of culture is how people behave when risk, time, cost and bad news collide.

Weak cultures hide problems, normalise shortcuts and make people believe speaking up is pointless or risky. Strong cultures make it normal to pause, challenge, report, learn and improve controls. The difference shows up in ordinary moments: a supervisor walking past a defect, a manager reacting to a delayed job, or a worker deciding whether to report a near miss.

This comparison is valuable because it turns the abstract word 'culture' into behaviours that can be observed, coached and audited.

When bad news reaches a manager, a weak response searches for someone to blame or a way to protect the target. A stronger response asks what risk is present, supports a pause and learns why the system failed. If a supervisor sees unsafe work, walking past it normalises the defect; stopping, coaching and removing the cause makes the standard real.

The same test applies to near misses, production pressure, contractor concerns and audits. Reporting should lead to fair investigation and shared learning; changed plans need review before work resumes; contractors should be heard at interfaces. Repeated audit failures need a verified repair at the workface, not administrative closure.

How does leadership improve culture?

Leadership is tested when safety conflicts with time, cost, production or reputation.

Workers watch what leaders do with difficult choices. If a leader praises a team for finishing early after known shortcuts, the culture message is clear. If a leader stops work, resources a control and thanks a worker for raising a concern, the message is different.

Leadership involvement should include site visits, risk conversations, action ownership, performance review, incident learning and checking that managers are not rewarding unsafe output.

Under production pressure, a leader should pause and replan if a critical control is missing. Safety-critical maintenance needs priority, an owner and evidence of close-out. Thanking a worker who stops a job and checking the risk teaches a different lesson from criticising the delay.

Contractor concerns need listening and verification across the site interface. Incident review should ask why the system allowed the failure, then test the resulting control. These are decisions that employees can see and repeat.

How do cooperation and consultation improve culture?

Consultation helps controls reflect work as done, not just work as imagined.

Workers often know which controls are awkward, which tools fail, which documents are ignored and which tasks drift under pressure. Involving them creates better decisions and stronger ownership.

Consultation can be direct, through representatives, through committees, through walkabouts, toolbox talks, surveys or focused work groups. The method should fit the workforce, language, literacy, shift pattern and risk.

Use one-to-one discussion for a small team or sensitive issue, with a short action note and feedback to the person who spoke. A toolbox talk can test task controls when questions and resulting actions are captured, not merely attendance. Walkabouts reveal work as done and should end with owners and verified close-out.

Representatives and committees give formal routes for recurring or cross-team concerns. Put their inspection findings on an agenda that leads to a management response; bring decisions back to people on every affected shift. The relevant GB or NI consultation law still determines who must be consulted and when.

What communication methods support culture?

Communication should be two-way, understandable and matched to the audience.

Culture improves when communication helps people make better decisions. A long email may be useful for policy record, but poor for a high-risk task. A face-to-face briefing may be better where workers need to ask questions, challenge assumptions or show understanding.

Use diagrams, spoken explanation, written prompts and hands-on demonstration as the task and workforce require. Check understanding through questions and observed practice, including for people with different language or accessibility needs.

  • Face-to-face briefing — This allows questions, tone and body language. However, it can be inconsistent if not planned.
  • Written procedure — This creates a stable reference. However, it may not be read or understood at the workface.
  • Visual prompt — This helps quick recognition. However, it can become wallpaper if overused.
  • Digital update — This is fast and traceable. However, it can overload people or miss workers without access.
  • Demonstration — This shows practical skill. However, it needs a competent demonstrator and time.

How does competence support a better culture?

Competence is more than attending training; it is the ability to do the work safely in context.

A strong culture makes competence visible. People are trained, supervised, assessed and refreshed. Supervisors know when someone is new, rusty, under pressure or doing a rare task. Competence records are linked to risk, not kept as a separate administrative exercise.

Training should include practical scenarios, local examples, what can go wrong and when to stop work.

  • Training matrix — This can show that people have received required learning. Ask: Is it linked to risk and role?
  • Observed assessment — This can show that people can apply the control. Ask: Was it assessed in real conditions?
  • Supervisor coaching — This can show that standards are reinforced at workface. Ask: Are coaching findings recorded and acted on?
  • Refresher plan — This can show that knowledge is maintained. Ask: Is refresher triggered by change or incident?
  • Contractor competence — This can show that external workers meet site requirements. Ask: Is competence checked before work starts?

How should fair discipline support culture?

Fair discipline means clear standards and fair investigation, not either blame or excuse.

A culture that never holds anyone accountable can drift into unsafe norms. A culture that blames honest error suppresses reporting. The mature position is a just culture: understand the context, fix system weaknesses and apply accountability where conduct is reckless or dishonest.

This is especially important for safety learners because disciplinary systems are a specific culture-improvement topic.

For an honest error in a poor system, improve the design, procedure, supervision and training; blame would hide the next error. If a whole team routinely takes a shortcut, investigate the workload and method that made it normal before judging one person. Thank a worker who reports their own mistake, assess the immediate risk and use the information to improve control.

A deliberate reckless breach may require formal accountability after a fair investigation. A manager who pressures unsafe work also needs escalation; a just culture cannot exist if senior decisions are exempt from scrutiny. Keep the response proportionate and explain the reasoning without exposing private details.

Keep learning and formal employment decisions distinct

A learning investigation does not mechanically decide disciplinary action. Check evidence, system pressures and the person’s opportunity to respond before any formal HR decision. Use the applicable GB or Northern Ireland employment procedure and seek appropriate HR advice; apply fair accountability to managers as well as workers.

Acas: disciplinary procedure (Great Britain)

How should safety committees and representatives add value?

Committees should turn worker voice and evidence into decisions, not become meeting rituals.

A useful committee has the right people, clear agenda, access to evidence, action owners and senior management response. It should review trends, inspections, incidents, worker concerns, training, health risks and outstanding actions.

Worker representatives need time, training and a route to escalate concerns. Consultation should include contractors and shift workers where their work is affected.

Start each meeting with the highest-risk unresolved issue, fresh worker concerns and the evidence needed for a decision. Representation should cover high-risk teams, shifts and affected contractors rather than just office staff. Use inspection, incident, audit and worker evidence together; opinion alone is weak assurance.

Give actions an owner, due date and field verification. If the same action returns unchanged, escalate the resource or authority gap to senior leaders. A committee is effective when workers can see a response and a safer method, not merely minutes.

What does a 90-day culture improvement plan look like?

A 90-day plan is a first cycle, not a promise that culture will change in 90 days. Begin with the highest-risk control gaps and evidence from people doing the work.

The owner should listen across shifts and contractors, compare feedback with incident and maintenance records, then fund a few changes people can see. Supervisors check the revised work in use, and leaders tell staff what was fixed, what remains open and who owns the next decision.

  • Days 1-15: listen — Hold worker discussions, supervisor conversations and site walkabouts; ask where controls are hardest to follow. Evidence to retain: Listening notes, themes and quick-win list.
  • Days 16-30: diagnose — Compare feedback with incidents, audits, actions, maintenance backlog, training and leadership decisions. Evidence to retain: Baseline report and priority risks.
  • Days 31-60: act — Fix a small number of visible barriers, improve consultation, resource overdue high-risk actions and coach supervisors. Evidence to retain: Action tracker, owner list and communication.
  • Days 61-80: verify — Check work as done, speak to workers and sample whether changes actually helped. Evidence to retain: Field verification and worker feedback.
  • Days 81-90: reinforce — Recognise good challenge, update standards, close the loop publicly and choose the next improvement cycle. Evidence to retain: Management review and next 90-day plan.

Day zero and one complete improvement loop

Make unsafe work safe immediately. The planned listening and improvement cycle never postpones a failed guard or other urgent control. Example: repeated jams → operators explain the awkward feed → an engineering owner resolves the feed design and production constraint → the change is tested on the next difficult shift → workers receive the findings and a route to reopen the defect.

What evidence shows culture is improving?

No single number proves culture is improving. A rise in near-miss reports may reflect growing trust, while a fall may reflect silence; compare each trend with worker accounts and verified control performance.

Sample work on different shifts and with contractors. Match reports to action owners, resources, field observations and worker feedback; reopen an action if the defect recurs. Escalate deferred safety-critical decisions rather than treating a closed spreadsheet cell as a repaired control.

  • Near-miss reporting — This may show that workers are willing to raise concerns. Interpretation: High numbers may mean trust, not failure.
  • Action close-out quality — This may show that management acts on issues. Interpretation: Closing actions without verification is weak.
  • Observation trends — This may show that work as done is being checked. Interpretation: Do not use observations only to blame workers.
  • Survey themes — This can reveal perceptions of the current safety climate. Interpretation: Act on findings or trust declines.
  • Committee actions — This may show that consultation leads to decisions. Interpretation: Minutes alone do not prove involvement.
  • Leadership decisions — This may show that values are shown under pressure. Interpretation: Track deferred safety-critical decisions.

Record a usable baseline

State who was sampled, which shifts and contractors were included, the denominator for trend rates, and any reporting-system change. Protect anonymous comments by restricting raw access and avoiding small-group attribution. Compare reports with observed safeguard condition, action verification and worker feedback; one survey score does not prove cultural change.

How does culture improvement look in a real scenario?

A scenario makes the topic practical rather than motivational.

A warehouse has a near miss where a pedestrian steps into the path of a fork-lift truck. In a weak culture, the event is quietly blamed on the pedestrian, the driver is warned informally and no one wants to report it because it will create paperwork and attention. The same layout, pressure and shared route remain.

In a stronger culture, the near miss is reported and investigated fairly. The review looks at route design, supervision, lighting, pedestrian behaviour, driver workload, production pressure, previous concerns and whether leaders have delayed segregation spending. The learning is then turned into physical segregation, revised routes, supervisor checks and worker feedback.

The warehouse response should move from warning people to redesigning the shared route. Leaders fund segregation and review the delivery target; drivers and pedestrians walk the route together before the change is final. Supervisors then assess route understanding during real shifts rather than repeating a generic briefing.

A fair investigation separates an individual mistake from a system that placed people and vehicles together. Hand-back requires physical route checks, observed use and a review of later near misses. If the old route reappears under pressure, reopen the action.

How does culture improvement transfer between sectors?

The cultural signal is whether people can surface a real control failure and see it fixed. Each sector should test its own task, exposure route and decision owner rather than copy a slogan.

For each example, the worker first reports the failed control, the supervisor holds the unsafe work, and the manager removes the pressure or resource gap that made the shortcut normal. The responsible person then verifies the control during real use and gives feedback to the workforce. If the same defect recurs, reopen the investigation rather than repeating the briefing.

Task and peopleHarm route and strongest improvementCheck and stop point
Manufacturing maintenance: operators and fitters report a repeatedly bypassed interlock.Machinery contact could crush a hand; redesign the access and resource repair before restart.Supervisor checks the interlock in use and stops the line if defeated again.
Construction: subcontractors report rushed changes to work at height.A fall could cause fatal injury; replan the sequence and provide suitable collective protection.Principal contractor checks the changed method and holds work until protection is verified.
Hospitality: cleaners report repeated slips on a wet floor.Workers and guests may fall; fix drainage and cleaning sequence.Manager inspects the surface and stops access when the area cannot be made safe.

What should each role learn and demonstrate?

Training should change decisions in the field. Match learning and assessment to the person’s authority, then revisit competence when work or evidence changes.

Before work, workers and contractors need to know the controls and how to raise a concern. During the job, supervisors must observe actual decisions, hold work when a safeguard fails and escalate resource gaps. After a change or incident, leaders and advisers should test whether learning reached every shift, the control works in practice and the record supports the restart decision.

Role and learningDemonstration and assessorRecord and reassessment trigger
Workers and contractors: report hazards, challenge unsafe work, use controls.Supervisor asks for a real stop-and-report example and observes the task.Induction and observation; revisit after method change or unreported near miss.
Worker representatives: gather views, consult early and follow action closure.Committee chair or competent adviser observes a consultation and checks feedback.Meeting/action record; revisit when representation or workforce changes.
Supervisors: detect drift, hold unsafe work, coach fairly and escalate resources.Manager tests a defect scenario and samples field decisions.Field check and intervention log; reassess after repeated shortcuts or incidents.
Senior leaders: resource controls, respond to bad news and test assurance.Board or independent competent reviewer samples decisions against risk evidence.Minutes, budget and action verification; revisit after major change or weak signals.
Safety advisers and auditors: triangulate surveys, incidents and observed control performance.Technical peer reviews sampling and conclusions.Audit trail and corrective-action verification; update after new evidence or method.

A briefing proves communication, not competence. Practical observation and a documented decision are stronger than attendance or a fixed refresher date.

A useful competence check is to ask a supervisor to handle a repeated guard bypass at the actual machine: pause the task, hear why the bypass occurs, arrange an engineering correction and explain who may release the equipment. The manager then samples another shift to see whether production pressure has recreated the shortcut. HSE safety-culture guidance supports looking at the system and observed behaviour together.

How should management check and improve these controls?

HSE’s Plan, Do, Check, Act approach helps an organisation plan controls, put them into use, check performance and correct failures. It is management guidance, not a substitute for the GB or Northern Ireland legal duties that apply to the task.

Check whether concerns reach decision makers, receive resources and lead to verified physical changes. A repeated guard defect should appear in the risk record, maintenance action and management review. HSE management guidance

How should learners use this page for workplace safety learning?

A good GNC answer applies a culture-improvement decision to the facts of the scenario and names evidence that the decision worked.

In a scenario answer, describe the leader’s actual decision, resource and follow-up instead of saying managers should care more. Explain how consulting workers reveals work-as-done and improves controls; identify the legal route where relevant. For fair discipline, separate honest error, normalised shortcut and reckless conduct rather than prescribing punishment for every unsafe act.

A useful answer on committees names representation, trend review, action ownership and escalation. Evidence of improvement needs verified control changes, worker feedback and field observation alongside a trend; a completed survey alone is not enough.

What warning signs show culture is weak?

Weak signals should trigger a decision before someone is harmed.

Near misses that disappear from reports may indicate low trust or an awkward reporting route. Ask workers directly, make reporting easier and show what changed. Repeated overdue high-risk actions suggest a resource or leadership decision problem; escalate the priority rather than quietly extending the due date.

If supervisors pass defects without challenge, coach and check their field decisions. Committees with no decisions need a sharper agenda and action owners. Attendance-only training needs observed assessment and a trigger for relearning after change or failure. Each warning should lead to a named owner and a test that the control now works.

How can the hierarchy of controls reduce repeated guard bypass through safer conveyor intervention?

What do the acronyms and key terms mean?

These terms help distinguish a measurable perception from the deeper management practices that shape work.

TermFull formMeaning here
OH&SOccupational health and safetyPrevention of work-related injury and ill health.
PDCAPlan, Do, Check, ActCycle for planning controls, using them, testing them and improving them.
HSG65Managing for health and safetyHSE guidance on practical management arrangements.
Safety climateManagement termPeople’s current perceptions of safety, often sampled by survey.
Safety cultureManagement termShared assumptions and normal decisions that shape risk control.
Weak signalAssurance termEarly evidence of drift, such as an unreported near miss or workaround.