How to Improve Safety Performance: A Practical Plan

If your injury rates haven't moved for a while, but your team is still flat out closing actions, running toolbox talks, and updating SWMS, you don't have a paperwork problem. You've got a system problem. Most Australian construction and manufacturing businesses already meet the basics of WHS administration. What's missing is the connection between what the site sees, what supervisors report, what managers review, and what changes in training, planning, and work design.

That gap matters most when the event you're trying to prevent is rare but severe. In Australia, the national fatality rate has hovered around 1.2 deaths per 100,000 workers, with 195 worker deaths in 2023, which is why mature organisations are pushing past minor injury counting and putting more effort into proactive controls and earlier signals of failure, as outlined in this summary of Safe Work Australia data. The practical question isn't whether you care about safety. It's whether your audits, inspections, incidents, and pre-starts are feeding one improvement loop or sitting in separate folders.

Table of Contents

• Moving Beyond the Plateau in Safety Performance
  • What the plateau actually means
  • Reactive systems hit a ceiling
• Establish Your Baseline with Leading Indicators
  • Start with a mixed baseline
  • The leading indicators that actually matter
  • Don't benchmark a broken process
• Demonstrate Visible Leadership Commitment
  • What visible commitment looks like on site
  • What doesn't work
  • The trade-off leaders need to accept
• Implement Proactive Hazard and Risk Controls
  • Use the hierarchy properly
  • Build controls at task level
  • Treat SWMS as operating documents
• Systematise Your Audits and Investigations
  • Stop treating inspections as rituals
  • Build one learning workflow
  • Focus on recurrence, not just severity
  • What to change when patterns emerge
• Manage Contractor and Subcontractor Safety
  • Control the contractor lifecycle
  • Watch the interfaces
  • Don't confuse submission with compliance
• Close the Loop with Continuous Improvement
  • Review what matters and ignore the noise
  • Run a deliberate review cycle
  • Prioritise severity, not volume

Moving Beyond the Plateau in Safety Performance

A plateau usually looks the same. Injury metrics look stable. Audit scores look acceptable. Supervisors say they're doing the right things. Yet the same access issues, plant interactions, manual handling shortcuts, isolation failures, and contractor coordination gaps keep turning up.

That's where most businesses get stuck. They keep adding activities instead of fixing the links between activities.

What the plateau actually means

When safety performance stalls, it rarely means people have stopped trying. It usually means the organisation is still managing WHS as separate compliance tasks. Inspections sit in one system. Training records sit somewhere else. Incident investigations get closed, but the lessons don't reshape SWMS, pre-start topics, or supervisor checks.

A mature WHS system isn't one with more forms. It's one where one finding changes the next job.

If you're working out how to improve safety performance, start by treating safety as an operating system, not a set of obligations. That means every audit, field observation, near miss, and corrective action has to push into a decision. Retrain. redesign the task. change the sequence. upgrade the plant. tighten contractor controls. remove the hazard where possible.

The same principle shows up in driver training. The useful part of A-1 Driving School's safe driving insights isn't the topic itself. It's the focus on practical habits, real-world feedback, and repeated correction instead of one-off instruction. WHS improvement works the same way. People improve when feedback is timely and tied to the task.

Reactive systems hit a ceiling

If your current process mostly starts after an incident, you're already behind. A reactive model can satisfy parts of compliance, but it won't reliably catch the conditions that lead to serious harm. That's why the shift from lagging response to earlier intervention matters so much in higher-risk operations.

A useful way to frame the difference is this comparison of proactive vs reactive safety management. The businesses that improve fastest don't just investigate what went wrong. They build routines that surface weak signals before the event.

Establish Your Baseline with Leading Indicators

You can't lift safety performance if you only measure outcomes after the fact. Most lagging indicators are too slow, too blunt, or too dependent on luck. They tell you what has already happened. They don't tell you whether your controls are healthy today.

Safe Work Australia reported 139,000 serious workers' compensation claims in 2023–24, with a median time lost of 5.5 weeks per claim, which is a strong reminder that serious events still have real operational consequences, not just compliance ones, as summarised in this overview of safety performance metrics.

Start with a mixed baseline

A proper baseline uses both lagging and leading indicators, but it gives more management attention to the leading side. Keep the lagging measures. You still need to know claim trends, incidents with potential for serious harm, and where work is being lost. But don't let them dominate the conversation.

Build your baseline around the health of the system itself.

The leading indicators that actually matter

Don't create a dashboard full of vanity measures. Pick indicators that show whether risk controls are being identified, assigned, completed, and checked.

A practical baseline usually includes:

• Inspection completion: Were scheduled inspections done when they were meant to be done?
• Corrective action close-out: How long does it take to move from finding to fix?
• Verification of controls: Was the action merely closed in software, or was the control checked in the field?
• Near-miss quality: Are reports specific enough to support a decision, or are they just noise?
• Training currency: Are workers current for the tasks and plant they use?
• Repeat issues: Are the same hazards turning up across crews, shifts, or sites?

Practical rule: If a KPI doesn't help a supervisor decide what to do next, it's probably the wrong KPI.

Don't benchmark a broken process

Before setting targets, test the integrity of the data. Many businesses say they have poor reporting when the underlying issue is poor definition. One supervisor logs housekeeping as a hazard. Another only logs high-potential exposures. One manager closes actions when parts are ordered. Another closes them after the guard is installed and checked.

That's why your baseline needs operating definitions. What counts as complete. What counts as verified. What counts as overdue. Without that, trend lines are mostly theatre.

If you need a clear framework for building that scorecard, this guide to leading and lagging indicators in WHS is a useful reference point. The value is in choosing fewer measures, defining them properly, and reviewing them consistently.

Demonstrate Visible Leadership Commitment

Most workers can tell inside a week whether leadership is serious about safety or just fluent in safety language. They don't judge commitment by a policy signed by the PCBU. They judge it by what leaders ask about, what they attend, what they challenge, and what they fix.

That's why visible leadership has to happen in the field, not only in meetings.

What visible commitment looks like on site

Formal training has its place, but safety performance improves faster when organisations pair it with field coaching and pre-job briefings, because the primary gain comes from a fast feedback loop between frontline observations and management action, as discussed in this safety improvement paper.

That means leaders need routines workers can see:

• Join walk-arounds: Not as silent visitors. Ask what is making the task harder, what control is unreliable, and what has changed since last week.
• Lead part of the pre-start: Keep it short. Focus on task change, interfaces, and conditions that could defeat the planned controls.
• Review serious near misses personally: Not every event needs executive involvement, but high-potential exposures should get visible follow-up.
• Close the feedback loop: If a worker raises an issue, tell them what was done, by whom, and by when.
• Challenge weak controls: If the answer is always “remind people”, leadership is accepting administrative drift instead of pushing for stronger controls.

What doesn't work

Leaders often make two mistakes. The first is outsourcing safety to the H&S team. The second is appearing only after something goes wrong. Both send the same message. Safety is secondary until the event forces attention.

A better test is simple. If you removed your policy manual tomorrow, would workers still recognise leadership commitment from daily behaviour?

If managers only show up for incident reviews, workers learn that reporting creates scrutiny, not support.

The trade-off leaders need to accept

Visible commitment takes time away from production discussions, procurement issues, and client pressure. That's real. But if site leaders never spend time on field verification, task coaching, and unresolved hazards, they're not saving time. They're borrowing it from the next disruption.

In construction and manufacturing, strong leadership isn't performative. It's operational. It means supervisors are backed when they stop a job, managers follow through on plant fixes, and directors ask about recurring hazards with the same seriousness they bring to output and quality.

Implement Proactive Hazard and Risk Controls

If you want to know how to improve safety performance, look at the quality of your controls, not the quality of your slogans. Most organisations already know the hierarchy of controls. The problem is they still default to the bottom of it. More signage. More reminders. More PPE enforcement. More toolbox talks after the fact.

That approach can support a control set, but it won't compensate for poor work design.

Use the hierarchy properly

The right question is not “What PPE do we need?” It's “How far up the hierarchy can we go before the job starts?”

In practice:

• Elimination: Remove the need for the task or the exposure. For example, redesign access so workers don't need to enter a danger zone.
• Substitution: Replace a hazardous substance, tool, or method with a safer option where it fits the job.
• Engineering controls: Install guarding, interlocks, extraction, isolation points, barriers, mechanical handling aids, or physical separation.
• Administrative controls: Use permits, sequencing, supervision, traffic plans, exclusion rules, and clear work instructions.
• PPE: Use it, but don't pretend it's the main control when better options are available.

A lot of SWMS fail because they describe hazards well but settle too quickly for administrative wording. “Worker to take care.” “Follow procedure.” “Use caution.” None of that changes the risk profile.

For a practical explanation of applying stronger controls, this guide on the hierarchy of controls in WHS is worth keeping handy.

Build controls at task level

Job Safety Analysis, or an equivalent task-based process, works because it breaks a job into steps, maps hazards at each step, and assigns controls tied to the work sequence, as outlined in this overview of safety performance at work. That's much more reliable than generic training because it forces the team to think about where the exposure sits.

Here's a simple comparison:

Treat SWMS as operating documents

For high-risk construction work, SWMS should be used the way the law intended. As working documents. Not just induction attachments.

That means reviewing them when:

• The sequence changes
• Different plant or materials are introduced
• A subcontractor interface creates new risk
• A near miss exposes a weak control
• Supervisors notice workers are bypassing the method because it doesn't match reality

Good task analysis doesn't make the paperwork longer. It makes the work method sharper.

In manufacturing, the same principle applies even where you're not relying on SWMS. Break the job down. Identify where hands, bodies, plant, energy, and process conditions interact. Then choose controls that survive production pressure, fatigue, weather, variation in skill, and handover between shifts.

Systematise Your Audits and Investigations

Most sites audit one thing, inspect another, investigate a third, and report all three in different formats. Then they wonder why the same failures return. The problem isn't that they lack activity. The problem is that the activity doesn't learn.

A stronger model treats inspections, audits, hazard reports, and investigations as one closed-loop system. Every input creates a finding. Every finding gets analysed. Every analysis leads to a corrective action. Every action gets verified. Then the results shape training, procedures, supervision, and design.

Stop treating inspections as rituals

A routine inspection is not just a compliance check. It's a data source. The same goes for a hazard card, a plant defect, a permit breach, or a subcontractor non-conformance.

The most effective improvement model is a closed-loop system that combines hazard reporting, root-cause analysis, and corrective-action tracking, with recurring findings used as a signal for changes to training or procedures, as explained in this practical guide to better safety.

If your inspection sheet only proves the inspection happened, it isn't doing enough.

Build one learning workflow

A workable structure looks like this:

1. Capture the issue

   Use one consistent method for inspections, audits, hazards, and incidents. Different categories are fine. Different logic is not.

2. Classify the finding

   Separate housekeeping noise from control failure, and separate control failure from exposure with serious harm potential.

3. Assign action

   Every finding needs an owner, due date, and expected control outcome. “Monitor” is not an action.

4. Verify effectiveness

   Don't close an item because someone sent an email. Close it when the guard is fitted, the sequence changed, the training delivered, or the condition physically removed, and someone checks it worked.

5. Review trends

   This is the step many organisations skip. If the same issue appears across multiple areas, you're looking at a systemic problem.

Focus on recurrence, not just severity

A single event can be serious. But repeated low-level findings often point to a bigger management weakness. I see this all the time with traffic separation, isolation discipline, temporary access, and contractor interfaces. The site fixes each individual issue, but no one asks why the issue is recurring.

That's where trend review matters.

Recurring findings are rarely a worker problem alone. They usually expose a planning problem, a supervision problem, or a control design problem.

What to change when patterns emerge

When recurrence shows up, the response needs to move up a level. Don't just brief the crew again. Ask which of these needs to change:

• Training content: Is the current package too generic for the actual task?
• Procedure design: Does the procedure match the field conditions, or is it being bypassed for practical reasons?
• Equipment and layout: Is the plant setup creating the unsafe workaround?
• Supervisor routine: Are field checks focused on the right failure points?
• Procurement decisions: Did the business buy a cheaper option that introduced extra risk?

That's how audits and investigations become useful. Not by producing neat reports, but by changing the operating conditions that produced the finding in the first place.

Manage Contractor and Subcontractor Safety

Contractor management breaks down when the principal contractor or host business treats it as a document collection exercise. Licences are checked. Insurance is filed. SWMS arrive by email. Then the actual work starts, interfaces change, and no one has a clear live view of who is competent, inducted, approved, or drifting off method.

For a PCBU, that's not a minor admin issue. It's a visibility issue.

Control the contractor lifecycle

The cleanest way to manage contractor WHS is to set controls across the full lifecycle, not just at induction.

A practical checklist looks like this:

• Pre-qualification: Check whether the contractor can manage the risks of the work. Review licences, insurances, relevant competencies, and the quality of their WHS system, not just whether they have one.
• Scope review: Confirm who controls what. A lot of contractor incidents start with assumptions about isolation, permits, traffic control, lifting plans, or access.
• Onboarding: Verify worker competencies, complete inductions, and confirm the approved SWMS or task method before site access.
• Work execution: Monitor field compliance through joint inspections, supervisor checks, permit reviews, and observation of actual work, not just submitted paperwork.
• Post-job review: Capture performance issues, strong practices, unresolved actions, and whether the contractor should be approved for future work.

Watch the interfaces

The highest contractor risk often sits between parties, not within them. One crew assumes another has isolated the asset. The host business assumes the subcontractor has briefed the revised sequence. Traffic management changes mid-shift and no one updates the delivery team.

Those failures don't usually come from ignorance. They come from fragmented control.

A centralised system helps because it gives supervisors one place to check contractor status, documents, competencies, and outstanding actions instead of chasing spreadsheets, inbox threads, and shared drives. That matters most on multi-site work and in industrial shutdowns, where the coordination load increases quickly.

Don't confuse submission with compliance

Receiving a SWMS doesn't mean the work is controlled. A strong contractor review asks:

Contractor management gets easier when the process is structured, current, and visible. It gets harder when the business relies on email chains, memory, and assumptions.

Close the Loop with Continuous Improvement

There isn't a finished version of safety performance. If your program feels finished, it's probably gone static. Conditions change. Crews change. plant changes. Supervisors change. Production pressure changes. A WHS system has to keep learning or it starts drifting out of step with the work.

That's why continuous improvement needs a review rhythm, not good intentions.

Review what matters and ignore the noise

Improving performance against high-severity risks requires targeting the most prevalent and dangerous hazard types rather than relying on simple counts, with screening and threshold-based prioritisation helping management focus on the few issues most likely to cause serious harm, as set out in this FHWA safety screening guide.

That principle matters in construction and manufacturing. Many businesses overreact to volume and underreact to consequence. They chase every low-value issue equally, which clogs the system and spreads effort too thin.

Run a deliberate review cycle

Your monthly or quarterly review should force decisions. Not just discussion.

Use it to answer questions like:

• Which hazards or task types keep recurring across sites or shifts?
• Which corrective actions are repeatedly overdue or ineffective?
• Where are we relying too heavily on admin controls when stronger controls are available?
• Which exposures have the clearest path to serious harm if they fail?
• What needs to change in training, supervision, plant, or planning before the next review?

The point of review isn't to prove effort. It's to decide where management attention goes next.

Prioritise severity, not volume

Some of the most dangerous failures won't produce lots of entries in your system. They may appear only occasionally. But when they do, the outcome could be catastrophic. Think uncontrolled energy, mobile plant interactions, lifting failures, work at height, confined spaces, and interface breakdowns during non-routine work.

That means your review process should separate:

• High-volume low-consequence issues that still need basic control
• Low-frequency high-severity exposures that deserve disproportionate management attention

Good organisations don't try to fix everything at once. They screen, rank, and act where the harm potential is highest.

A business that wants to improve safety performance year after year needs one discipline above all others. It needs to keep turning field information into changed conditions. When that happens consistently, audits become useful, training gets sharper, supervisors get better signals, and workers see that reporting leads to action.

Safety Space helps Australian businesses connect audits, inspections, incidents, contractor management, and corrective actions in one system so nothing gets lost between the field and management review. If you want a clearer way to turn WHS data into action across construction, manufacturing, or industrial services, have a look at Safety Space.