A construction safety program is two things at once: a written plan that spells out how your company finds and controls hazards, and the daily field habits — toolbox talks, pre-task planning, inspections — that prove the plan is real and not a binder on a shelf. Inspectors, general contractors, and insurers all judge you on the gap between the two. This guide maps the whole program: the elements that belong in it, the leading indicators that predict incidents before they happen, and the metrics that tell you whether it's working.
This is the hub for our safety-program coverage. Start here for the structure, then go deep where you need to.
What OSHA actually expects a program to contain
OSHA's Recommended Practices for Safety and Health Programs (2016) is the clearest map of what a working program looks like. It's guidance, not a citable standard on its own — but it's the framework OSHA, ANSI Z10, and ISO 45001 all converge on, and it's built around seven core elements:
1. Management leadership — leaders set the policy, fund it, and are visibly accountable. 2. Worker participation — the crew helps find hazards and isn't punished for reporting them. 3. Hazard identification and assessment — you systematically look for hazards (inspections, JHAs, incident data). 4. Hazard prevention and control — you fix what you find, following the hierarchy of controls. 5. Education and training — everyone knows the hazards of their work and how to control them. 6. Program evaluation and improvement — you measure the program and change what isn't working. 7. Communication and coordination on multi-employer sites — the GC and subs coordinate so nobody's hazard becomes someone else's injury.
If your program addresses those seven honestly, the specific paperwork (written plans, training records, inspection logs) tends to fall out of them naturally.
Find the hazards before they find you: the JHA
The engine of hazard identification is the Job Hazard Analysis (JHA), also called a Job Safety Analysis (JSA). You break a task into steps, name the hazard in each step, and decide the control before the work starts. Done well, a JHA is where most of your incidents get prevented — on paper, the day before, instead of in the field.
The discipline that makes JHAs work is the hierarchy of controls: eliminate the hazard first, then substitute, then engineer it out, then use administrative controls, and only last rely on PPE. A JHA that answers every hazard with "wear gloves and be careful" isn't controlling anything.
> Free tool: Get a defensible, printable field record with the free daily log template — the day-to-day companion to your pre-task planning.
The daily habits that keep a program alive
A written program is worthless without the field cadence that runs it every day:
- Toolbox talks — short, frequent, hazard-specific crew briefings. Consistency matters more than length: a five-minute talk every morning beats a monthly seminar.
- Pre-task planning / daily huddles — the crew walks the day's specific work and its specific hazards before tools come out. This is the JHA made live.
- Site safety inspections — a regular walk against a checklist, with findings tracked to closure. An inspection that names a hazard but never confirms the fix is just a to-do list.
The pattern across all three: they're only worth anything if the findings are captured and closed, not just discussed and forgotten.
Leading indicators beat lagging ones
Most contractors measure safety only after someone gets hurt — injury rates, lost days, claims. Those are lagging indicators: real, but they tell you about a failure that already happened.
Leading indicators measure the activity that prevents injuries, so they move before the incident rate does:
- Toolbox talks held vs planned
- JHAs completed for high-risk tasks
- Inspection findings closed within target
- Near-miss reports submitted
Near-miss reporting is the highest-leverage leading indicator in construction. A near miss is a free lesson — the same failure that could have caused an injury, minus the injury. Programs that make it easy and blame-free to report near misses surface the conditions that would otherwise become the next recordable. Behavior-based safety (BBS) and structured safety observations work the same way: they generate a stream of leading data about how work is actually being done.
Measure whether it's working
When a GC prequalifies you or an insurer prices you, three numbers come up:
- TRIR (Total Recordable Incident Rate) — recordable cases × 200,000 ÷ hours worked. Recordables per 100 full-time workers per year.
- DART (Days Away, Restricted, or Transferred) — the subset of recordables serious enough to keep someone off their normal job. Always ≤ TRIR.
- EMR (Experience Modification Rate) — set by your workers'-comp rating bureau from three years of actual-vs-expected losses. You don't calculate it; you earn it. An EMR under 1.0 means you're safer than average for your class, and it directly discounts your premium.
TRIR and DART you compute yourself, straight from the OSHA 300 log.
> Free tool: Calculate your TRIR and DART rates from your own log with the free TRIR & DART rate calculator.
When something does go wrong: investigate the cause, not the person
When an incident or a serious near miss happens, a real investigation looks for the root cause — the system condition that allowed it — not just the proximate act. "The worker wasn't tied off" is a proximate cause; "we had no anchor points planned for that work and no one checked" is a root cause. Fixing proximate causes gets you the same incident again with a different name; fixing root causes is how the program improves (core element #6).
How the pieces fit together
| Layer | What it is | Why it matters |
|---|---|---|
| Written program | The plan: policy, roles, controls, training | What an inspector or GC asks to see |
| JHA / pre-task planning | Hazards named + controlled before work | Where most incidents are actually prevented |
| Toolbox talks + inspections | The daily field cadence | Keeps the program live, not shelved |
| Leading indicators | Near misses, JHAs, closed findings | Predict incidents before they happen |
| Lagging metrics (TRIR/DART/EMR) | The scoreboard | How you're prequalified and priced |
| Incident investigation | Root-cause, not blame | How the program improves over time |
The honest summary: a safety program isn't a document you finish, it's a loop you run — find hazards, control them, train on them, measure, and improve. The contractors who win at it are the ones whose leading indicators are strong long before their injury rate has anything to say.
Start with the OSHA compliance guide for the recordkeeping and standards that sit underneath the program.
Where this breaks down is the tracking — toolbox talks, JHAs, inspections, and near misses scattered across texts, paper, and three supervisors' trucks. Jobsite keeps the whole loop in one place and turns your field activity into leading indicators automatically, so you can see the program working before the incident rate ever moves. See how it works →
Articles in this guide
See how Jobsite handles this in practice: Safety & compliance on the platform →