Safety Records and EMR in Traffic Control Prequalification

There is a moment in a lot of traffic control companies' growth where the safety record stops being an internal matter and becomes a commercial gate. A general contractor asks for an EMR before adding you to a bid list. An agency prequalification form has a section for OSHA rates. A large customer's procurement portal will not let you register without uploading three years of loss runs.

At that point, safety performance is not a values conversation. It is a revenue one.

What an EMR is, mechanically

An experience modification rate is a multiplier applied to workers' compensation premium, produced by a rating bureau from your claims history compared against expected losses for a company of your classification and size.

A few properties matter for how it should be interpreted.

It is externally produced. You do not calculate it and you cannot adjust it. That is the entire reason it carries weight with people evaluating you — unlike a self-reported safety statement, it is hard to dress up.

It is lagging and multi-year. The calculation draws on several past policy periods rather than the current one. A bad year affects you for years, and a genuine improvement takes years to appear. This is the property most companies discover at the worst time.

Frequency usually hurts more than severity. Most rating formulas weight the number of claims more heavily than the total dollars, on the theory that frequency is more predictive of future losses than one catastrophic event. Several small recordables can move an EMR more than one large claim.

It is size-sensitive. Small companies have volatile EMRs, because a single claim is a large fraction of their expected losses. This is a real unfairness in how small subcontractors get screened, and it is worth raising directly when a threshold is applied mechanically.

The practical implication of all of it: an EMR is a slow-moving asset. It cannot be fixed in a quarter, which is exactly why it is worth managing before anyone asks for it.

What OSHA recordkeeping produces

The other externally-anchored numbers in a prequalification package come out of injury and illness recordkeeping — recordable rates and lost-time rates computed from your own logs.

These are self-maintained, which makes them less trusted than an EMR, but they are also subject to inspection and to a legal obligation to be accurate. Underrecording is not a strategy; it is an exposure.

Where the regulatory floor sits for this trade is worth restating, because it is often misunderstood. OSHA does not maintain a separate flagging rulebook — it incorporates the traffic engineering manual by reference. 29 CFR 1926.201(a) states that "Signaling by flaggers and the use of flaggers, including warning garments worn by flaggers, shall conform to Part 6 of the MUTCD." OSHA's highway work zones hub confirms that "The MUTCD is referenced in 1926 Subpart G," with those broader requirements at 29 CFR Part 1926 Subpart G.

That structure means a work zone setup deficiency can become an OSHA citation, and a citation history is visible to the people evaluating you.

The context an EMR does not capture

Here is the fairness problem with screening traffic control companies on loss history, and it is worth understanding well enough to articulate.

Work zone exposure is not evenly distributed and it is not fully within a contractor's control. The Associated General Contractors of America, reporting its annual survey with HCSS, states that "Sixty percent of highway construction firms report having at least one car crash into their work zones during the past year".

Sixty percent, annually, is a base rate — not a distinguishing characteristic. And the crashes are overwhelmingly caused by people who do not work for you. FHWA reports speeding as a factor in 281 of 821 work zone fatal crashes in 2022, roughly 34 percent, with commercial motor vehicles involved in 246, roughly 30 percent. The Work Zone Safety Information Clearinghouse records 850 work zone fatalities in 763 fatal crashes in 2024, the large majority of them road users rather than workers.

A company that runs high-speed night closures on truck-heavy corridors is exposed to a different hazard than one doing daytime residential utility work, and no rating formula adjusts for that.

The productive response is not to complain about the metric. It is to bring context alongside it: your exposure profile, your leading indicators, and evidence of what you actually do. Evaluators who receive only a number treat it as the whole story. Evaluators who receive a number plus a credible program have something else to weigh.

What a credible program looks like on paper

Prequalification reviewers and GC safety departments look for a consistent set of artifacts. If you are building toward this, build these.

A written program that matches what you actually do. A generic downloaded manual is transparently generic, and reviewers read a lot of them. A short program that describes your real setup procedures, your real training, and your real escalation path is more persuasive than forty pages of boilerplate.

Training records with dates and content. Flagger certification is table stakes — see flagger certification requirements. What distinguishes a program is evidence of ongoing training tied to your own incidents.

Incident investigation with corrective actions closed out. Not an incident report filed. An incident report, a cause analysis, a change, and evidence the change happened.

Leading indicators. This is where a traffic control company can genuinely differentiate. Near misses and intrusions appear in no external dataset, which means a company that measures its own intrusion rate has data nobody else can produce. Our post on work zone intrusion prevention and response covers how to build that reporting so crews actually use it.

Setup documentation. Timestamped, located evidence that setups were built as planned. This is the artifact that answers "how do we know your crews follow the plan" with something other than an assertion.

The reporting trap

One warning, because it defeats more safety programs than any other single thing.

If reporting a near miss produces a conversation about what the crew did wrong, crews stop reporting near misses. The leading-indicator data you need for prequalification is the same data your people will withhold if reporting is costly to them.

Blameless-by-default reporting is not a soft management preference. It is the only mechanism by which the data exists at all. A company with zero reported near misses does not have a safe operation; it has an operation where reporting is discouraged, and any experienced safety evaluator reads it that way.

Practical sequencing

If prequalification is on your horizon in the next eighteen months:

Get your loss runs and your current EMR now. Not when a form asks. Understanding what is in your history takes time, and occasionally there are claims coded incorrectly that can be corrected.

Start leading-indicator reporting immediately, because it needs months of data before it is worth showing anyone.

Close the loop on open corrective actions. A reviewer who finds three incident reports with no documented resolution has learned something specific about your management system.

Assemble the package once, properly, and keep it current. Prequalification renewals are recurring; rebuilding the package from scratch annually is a self-inflicted cost. Our post on bidding municipal contracts covers where prequalification sits in the wider procurement timeline.

Standing caveat: prequalification requirements, safety thresholds, and rating methodologies vary by agency, by state, and by rating bureau. Nothing here is legal, insurance, or actuarial advice. Verify with the specific agency and with your insurance broker.

Where the documentation comes from

Every artifact above depends on field records that were captured at the time and can be retrieved later. Setup photographs attached to a job. Tickets with times and signatures. Incident notes tied to a location and a date. Certification expiry visible before someone is dispatched rather than after.

When those live in four systems, assembling a prequalification package is a two-week archaeology project. When they live on the job record, it is an export. That is the reasoning behind how Traffic OS handles field capture and crew records — the pricing page shows what is at each tier, and you can book a walkthrough to see it on real screens.

An EMR is a number somebody else calculates about your past. The program you build now is the only input you control, and it takes years to show up. That is an argument for starting before the form arrives, not after.