How to Improve Contractor Safety Performance in 2026

Almost every owner and general contractor screens subcontractors on TRIR. Set a threshold, disqualify anyone above it, move on.

There's a problem with that. In 2020, the Construction Safety Research Alliance published research by Hallowell and colleagues analyzing 17 years of data across 3.2 trillion worker-hours. The findings were blunt: no discernible relationship between TRIR and fatalities, recordable incidents occurring almost at random, and TRIR being statistically invalid for comparing companies, teams, or projects in nearly every circumstance. The researchers found it only becomes predictive with more than 100 months of data behind it.

Most prequalification packages ask for three years.

This doesn't mean safety metrics are useless or that prequalification is theater. It means the number you screen on and the things that actually make a contractor safer are not the same, and building a program around the first while ignoring the second produces paperwork rather than fewer injuries. This guide covers what a contractor safety program should include, what genuinely predicts performance, and how to structure a performance improvement plan when a contractor falls short.

Key takeaways

• TRIR, EMR, and DART are the three metrics prequalification runs on, and each measures something different: frequency, insurance loss history, and severity respectively.
• Research from the Construction Safety Research Alliance found TRIR statistically invalid for comparing contractors, though it remains the industry's default screening tool.
• Leading indicators, meaning near-miss reporting, inspection completion, corrective action closure, and training compliance, are what actually predict future performance. Contractors who track them consistently see lagging numbers improve within 12 to 18 months.
• A suspiciously low TRIR deserves as much scrutiny as a high one. An extensive, defensible OSHA log usually signals a better reporting culture than a near-zero rate.
• Under OSHA's multi-employer citation policy, a general contractor can be cited as a controlling employer for hazards created by a subcontractor, which is why contractor safety is not something an owner or GC can fully delegate.

The three metrics prequalification runs on

1. TRIR (Total Recordable Incident Rate) measures the frequency of OSHA-recordable incidents per 200,000 worker-hours, which represents what 100 employees would work in a year. The standardized denominator is what allows a 50-person firm to be compared against a 500-person firm. The BLS put the construction industry average at 2.3 per 100 full-time equivalent workers in 2024. A TRIR under 1.0 is generally treated as strong.
2. EMR (Experience Modification Rate) reflects workers' compensation loss history against businesses of similar size and classification, calculated by the National Council on Compensation Insurance or a state board. It's applied directly to workers' comp premiums, which makes it the metric with the clearest financial consequence. Baseline is 1.0. Above that means worse-than-average loss history and higher premiums; below means better. It runs on a rolling three-year experience period, so it moves slowly and can't be improved with a push before renewal season.
3. DART (Days Away, Restricted, or Transferred) measures cases serious enough to keep someone off the job or on modified duty. Because TRIR counts a minor recordable and a serious injury identically, DART adds the severity dimension TRIR lacks.

Together these give you frequency, cost, and severity. None of them tell you whether a contractor's safety program is actually functioning.

Why threshold screening falls short

Two problems with using these numbers as a pass-fail gate.

The first is the statistical one above. Small workforces produce noisy numbers. A ten-person crew that has one recordable incident in a year posts a TRIR that looks alarming next to a 400-person firm that had three. Neither number tells you much about hazard exposure or program quality.

The second is subtler and more important. A very low TRIR can reflect excellent safety performance, or it can reflect a reporting culture where incidents get discouraged, reclassified, or handled quietly. Experienced reviewers know this, which is why most prefer an extensive, defensible OSHA log over a near-zero rate. A contractor reporting near-misses actively will look statistically worse than one suppressing them, right up until something serious happens.

The practical response isn't to abandon the metrics. It's to treat them as the beginning of the conversation rather than the end of it. Look at trend direction across multiple years rather than a single snapshot. Ask what changed when a number moved. And when a contractor's EMR sits above 1.0, best practice is to let them explain the context or submit an improvement plan rather than automatically disqualifying them, which is how you avoid screening out a firm that had one bad year and has since fixed the underlying problem.

What actually predicts safety performance

Lagging indicators tell you what already happened. Leading indicators tell you what's likely to happen next, and they're where a safety program either works or doesn't.

The ones worth tracking:

• Near-miss reporting rate. Counterintuitively, more reports usually indicate a healthier program. A crew that reports near-misses is a crew that hasn't learned to stay quiet.
• Inspection completion rate. Whether scheduled inspections actually happen, on time, rather than getting signed off retroactively.
• Corrective action closure time. How long between identifying a hazard and confirming it's resolved. Open items aging past a threshold is one of the clearest warning signs available.
• Training compliance. Whether workers on site have current, role-appropriate certifications, tracked continuously rather than checked at mobilization.
• Safety observation frequency. Documented behavior observations and toolbox talks, with attendance recorded.

The connection between these and the metrics you get screened on is direct. Contractors who consistently track leading indicators generally see their lagging numbers improve over 12 to 18 months, which then flows into a lower EMR and better prequalification standing. Fewer incidents become recordable in the first place, which is the only durable way to lower TRIR.

For a broader look at tracking contractor performance beyond safety specifically, our guide to contractor performance monitoring covers schedule, quality, and documentation metrics.

What to require during prequalification

A defensible prequalification package captures more than three numbers. At minimum:

• Safety performance history: three years of TRIR and DART, EMR with the rating bureau documentation, OSHA 300A summaries, and any recent citations with resolution status.
• Written programs: a safety manual plus the hazard-specific programs relevant to the scope, typically fall protection, lockout/tagout, confined space, hot work, and job hazard analysis procedures.
• Training and authorizations: role-based training records for equipment like forklifts and aerial lifts, plus designated competent persons for the applicable hazards.
• Supervision structure: who holds safety responsibility, supervisor-to-worker ratios, the disciplinary policy, and whether stop-work authority is formally granted to workers.
• Insurance: current certificates with limits, endorsements, and waiver details.
• Continuous improvement evidence: the incident investigation process, which leading indicators they track, and how corrective actions get closed.

That last category is the one most packages skip and the one that tells you the most. A contractor who can describe their corrective action tracking process in specifics is running a real program. One who can only produce numbers is producing numbers.

Prequalification platforms

Several third-party services centralize this. ISNetworld, Avetta, Veriforce, Highwire, and Browz all maintain contractor safety data, verify documentation, and score or flag contractors against client-defined criteria.

The advantage is administrative: you're not chasing certificates and OSHA logs from forty subcontractors individually, and verification happens on the platform's side. Highwire and similar services automatically flag higher-risk contractors against thresholds you set.

The limitation is worth naming. These platforms verify that documentation exists and metrics fall within range. They don't tell you whether the safety program described on paper is the one being run in the field. That gap closes through orientation, site observation, and ongoing tracking, not through prequalification alone.

Running the program after the contract is signed

Prequalification screens who gets on site. What happens next determines whether anyone gets hurt.

Site-specific orientation matters more than generic training. A contractor's national safety program doesn't cover the particular hazards of your site, the location of your muster points, or which trades are working overhead this week.

Job hazard analyses should be scope-specific and reviewed before high-risk work begins, not filed and forgotten. Fall protection work and confined space entry in particular warrant confirming the rescue plan exists before mobilization, not after.

Stop-work authority only functions if workers believe using it won't cost them. That belief comes from how the first person who exercises it gets treated, which is a leadership question rather than a policy question.

Corrective action tracking is where most programs quietly fail. Identifying a hazard is easy. Confirming it was resolved, by whom, and when, and having that record available six months later is the part that requires a system.

How to structure a contractor performance improvement plan

When a contractor's safety performance falls below standard mid-project, a performance improvement plan is the formal alternative to either ignoring it or removing them. A workable one has five components.

1. Specific findings, not general concerns. "Safety performance is unsatisfactory" gives a contractor nothing to act on. "Three fall protection violations observed in the past 30 days, documented on [dates], in these locations" does.
2. Root cause, not just symptoms. Repeated violations of the same type usually trace to something structural: inadequate supervision ratios, workers trained on a different standard, schedule pressure creating shortcuts, or equipment that isn't available when needed. A plan addressing the violation without addressing the cause produces temporary compliance.
3. Required actions with named owners and dates. Each corrective action assigned to a specific person with a deadline, not distributed to the company generally.
4. Measurable verification criteria. What specifically has to be true for the plan to be considered satisfied, and who confirms it. Vague success criteria mean the plan never formally closes.
5. Consequences and review cadence. What happens if the plan isn't met, and when performance gets reviewed against it. A plan with no defined follow-up review tends not to get followed.

Documentation matters here beyond the immediate situation. If a serious incident occurs later, the record of what you identified, what you required, and what was verified becomes significant. That's true whether the outcome is improvement or removal.

Why owners and GCs carry more risk than they sometimes assume

OSHA's multi-employer citation policy recognizes several employer roles on a shared worksite, including creating, exposing, correcting, and controlling employers. A general contractor with general supervisory authority over a site can be cited as a controlling employer for hazards created by a subcontractor, even without having created or been exposed to the hazard.

The practical implication: contractor safety isn't fully delegable. Requiring a subcontractor to have a safety program doesn't transfer the obligation to exercise reasonable care in identifying and correcting hazards on your site. That's the legal reason contractor safety programs exist, separate from the more important reason.

This is general information rather than compliance advice, and specifics vary by jurisdiction and circumstance. Worth confirming with counsel or a qualified safety professional for your situation.

What tends not to work

Screening on a single metric. Covered above, but worth repeating because it remains the most common approach.

Orientation as a video and a signature. Documentation of training is not the same as training.

Safety programs that only exist in the prequalification binder. If site supervisors can't describe the program, workers aren't running it.

Punishing incident reporting. Any pressure toward lower reported numbers produces lower reported numbers, not fewer incidents.

Improvement plans without verification. A plan nobody checks against becomes a filed document rather than a change in behavior.

How INGENIOUS.BUILD supports this

Most of what's described here depends on documentation that stays organized and accessible: inspection records, corrective action status, training verification, and the trail showing what was identified and when it was resolved.

INGENIOUS.BUILD's Construction Administration module handles inspections, punch lists, and quality and safety tracking in the same connected system used for RFIs, submittals, and closeout, with owners, GCs, subcontractors, and design teams working from shared data. That matters for safety specifically because corrective action closure, which is the leading indicator most likely to slip, becomes visible in real time rather than requiring someone to compile it.

Teams using INGENIOUS.BUILD report 10x fewer change-order disputes, which comes from the same underlying property: a documentation trail that exists by default rather than being reconstructed after something goes wrong.

Book a demo to see how inspection and corrective action tracking works across your projects.

FAQ

What is a good TRIR for a construction contractor?

The BLS put the construction average at 2.3 per 100 full-time equivalent workers in 2024. Under 1.0 is generally considered strong, though research suggests TRIR is a weak basis for comparing companies directly.

What's the difference between TRIR, EMR, and DART?

TRIR measures the frequency of OSHA-recordable incidents. EMR reflects workers' compensation loss history and directly affects insurance premiums. DART measures the severity subset: cases involving days away, restricted duty, or job transfer.

Is TRIR a reliable measure of contractor safety?

Research from the Construction Safety Research Alliance covering 17 years and 3.2 trillion worker-hours found no discernible relationship between TRIR and fatalities, and concluded it's statistically invalid for comparing companies in most circumstances. It remains widely used for prequalification regardless.

What programs improve contractor safety performance?

Programs built on leading indicators, meaning near-miss reporting, inspection completion, corrective action closure, and training compliance, rather than lagging metrics alone. Contractors tracking these consistently typically see recordable rates improve over 12 to 18 months.

How do you write a contractor performance improvement plan?

Include specific documented findings rather than general concerns, root cause analysis, required actions with named owners and deadlines, measurable verification criteria, and a defined review cadence with stated consequences.

Should a contractor with an EMR above 1.0 be disqualified?

Not automatically. Best practice is allowing the contractor to provide context or submit an improvement plan, since a single bad year can elevate a rolling three-year rate even after the underlying issue is resolved.

What are the main contractor safety prequalification platforms?

ISNetworld, Avetta, Veriforce, Highwire, and Browz are the most widely used. They verify documentation and flag contractors against client-defined criteria, though they confirm that programs exist on paper rather than that they're being followed in the field.

Can a general contractor be cited for a subcontractor's safety violation?

Yes. Under OSHA's multi-employer citation policy, a GC with general supervisory authority can be cited as a controlling employer for hazards created by a subcontractor. Specifics vary, so confirm with qualified counsel.