Workers' Compensation Classification for Traffic Control Companies

Most owners of traffic control companies can tell you their workers' comp rate. Fewer can tell you which classification codes their payroll is sitting in, and that second number frequently matters more.

Classification is the step that happens before the rate is applied. It decides which bucket each dollar of payroll lands in, and the difference between buckets in construction-adjacent work is not small. A company that gets classification right, manages its experience modifier deliberately, and keeps clean payroll records is running a materially different cost structure from an identical company that does not — with the same crews, the same trucks and the same jobs.

This post explains how the mechanism works, where traffic control specifically sits awkwardly, and what to ask. It deliberately does not tell you which code applies to your business, because that answer depends on your state, your actual operations, and a rulebook that your carrier — not a website — is responsible for applying.

How the mechanism works

Workers' compensation premium is, in its basic shape, payroll in each classification multiplied by a rate for that classification, then adjusted.

The classification system. In most states, the classification system is administered by the National Council on Compensation Insurance. NCCI maintains the classification phraseology and the rules for assigning codes; its Classification Codes and Statistical Codes resources describe the structure, and its Classification Basics material explains how assignment is supposed to work. The Scopes manual — a supplement to the Basic Manual — is the interpretive guide carriers use to decide what a given classification actually covers.

The independent-bureau states. Several states run their own rating organizations rather than using NCCI's, with their own codes and their own rules. New York works through the New York Compensation Insurance Rating Board, Pennsylvania through the Pennsylvania Compensation Rating Bureau, California through its own rating bureau under the oversight of the California Department of Industrial Relations. If you operate across state lines, you are operating under more than one classification system, and they do not agree with each other.

The experience modifier. Your loss history relative to expected losses for a business of your size in your classes produces a multiplier applied to premium. It is calculated over a multi-year window, which means both that a bad year follows you and that improvement takes time to show up.

The audit. Policies are written on estimated payroll and audited against actual. The audit checks the amounts and the classifications, and it is where a classification assumption that was never tested gets tested.

The upshot: three levers, and the one most owners spend the least time on — classification — is the one that is set once at policy issuance and then quietly multiplies everything.

Why traffic control sits awkwardly

The classification system was built around describing operations. Traffic control is unusual in that one company routinely performs several operations that would be described differently, often on the same day, sometimes by the same person.

Consider what a mid-sized traffic control company actually does:

• Flagging — a person standing at the roadside directing traffic, exposed to live traffic, performing no construction.
• Device setup and removal on live roadways — crews placing cones, drums, barricades and signs in or adjacent to travel lanes.
• Sign installation and larger device placement — sometimes involving posts, sometimes equipment.
• Yard work — loading, unloading, washing, refurbishing devices, warehouse-type work.
• Shop and fleet maintenance — servicing arrow boards, attenuator trailers, light towers, trucks.
• Office and estimating — clerical and sales work with no field exposure.
• Plan preparation — drafting or engineering work, sometimes in-house.

Those are genuinely different risk profiles. Whether they can be separated for premium purposes is governed by division-of-payroll rules that vary by state and by the specific facts of how your people work. In construction operations, classification generally involves the concept of a governing classification, with payroll division permitted only in defined circumstances and typically requiring records adequate to support the split.

The practical implication is blunt: your ability to have payroll classified separately usually depends on whether your records can prove the split. An employee who flags in the morning, loads trucks in the afternoon, and whose time is recorded as a single daily total is, for classification purposes, one thing. The same employee whose time is recorded against distinct activities may be treated differently — if the rules in your state allow it, and if the records satisfy an auditor.

That is a records question before it is an insurance question, and it is one of the few places where installing a time system pays for itself in a line you can point at.

The exposure is real, and the data says where it sits

Classification is a pricing mechanism, but the pricing exists because the exposure is genuine.

BLS Census of Fatal Occupational Injuries data compiled by the National Work Zone Safety Information Clearinghouse shows between 82 and 143 fatal worker injuries at road construction sites annually from 2015 through 2024, representing 1.6% to 2.8% of all US worker fatalities.

It is worth knowing what that count includes, because the boundary is broader than "highway construction". The Clearinghouse states that "BLS defines a 'road construction site' as a location of construction, maintenance, and utility work on a road, street, or highway" — which covers the short-duration utility work and maintenance operations that make up a large share of many traffic control companies' schedules, not just the long capital projects people picture. The same source is candid about the limits of the data, noting that it is possible some fatalities and injuries occurring at road construction sites were not recorded as such and so would not appear in the statistics at all.

The mechanism is consistent. Averaged across 2022 to 2024, 52.7% of those worker fatalities were workers on foot struck by a vehicle, and a further 24.8% were workers as drivers or passengers in motor vehicle crashes — roughly 78% vehicle-related between them, leaving 22.5% for falls, struck-by-object, electrocution and everything else combined.

The same source shows the share of worker fatalities involving workers on foot struck by vehicles rising from 35.4% in 2015 to 59.8% in 2023 before easing to 47.5% in 2024.

On the traffic side, the Clearinghouse records 763 fatal work zone crashes and 850 work zone fatalities in 2024, and FHWA's tabulation for 2022 records 891 work zone fatalities including 94 highway worker occupational fatalities.

Two things follow for an owner reading a premium invoice. First, a class that prices roadside exposure heavily is pricing something real, and arguing the classification down is not the lever — reducing the exposure is. Second, because the dominant mechanism is a worker on foot being struck, the controls that move your loss record are specific and known: positive protection where it is warranted, high-visibility apparel actually worn and actually in serviceable condition, flagger station setup with real escape routes, and work zone intrusion prevention and response as a planned-for event rather than a surprise.

Where misclassification comes from

Misclassification is rarely fraud. It is usually one of five ordinary things.

The business changed and the policy did not. A company that started as a flagging service and now runs barricade rentals, sign installation and a maintenance shop is performing operations the original classification did not contemplate. Nobody revisits it because the renewal is a renewal.

Payroll records cannot support the split that was assumed. The policy was written on the understanding that yard staff and field crews were separate; the timesheets show one total per person per week. At audit, the assumption fails.

Subcontracted labour lands in your audit. Uninsured subs are commonly picked up as your payroll. A certificate of insurance that expired mid-policy, or covers a different entity than the one that sent people, produces an audit bill you did not budget for. There is also a distinct legal question underneath — whether those people are properly independent contractors at all — which is covered separately in flagger employee vs independent contractor classification.

Overtime is reported wrong. Rules exist for how overtime pay is treated in classified payroll, and applying them incorrectly moves real money in both directions. Overtime and FLSA basics covers the wage side of the same records.

Nobody read the class description. The phraseology for a classification is specific, and reading it against what your people actually do is a thirty-minute exercise that most owners have never done. NCCI itself notes that asking appropriate questions about an employer's operations at assignment reduces later reclassification.

What to actually do

None of the following is advice about your specific policy — your carrier, agent and the rating organization in your state are the people who apply the rules. This is the preparation that makes that conversation productive.

1. Get your current classifications in writing, with the phraseology. Not just the code numbers. The description is the thing to compare against reality.

2. Write an honest operations description. One page: what your people do, roughly what proportion of payroll sits in each activity, whether individuals cross between them, and what equipment is involved. Be accurate rather than flattering — a description written to produce a cheaper answer is the one that fails at audit.

3. Ask your carrier three questions. Are these the right classifications for the operations described? Under the rules in force in this state, is any division of payroll available to us? And if so, exactly what records would an auditor need to see?

4. Then build those records. This is the step that converts the conversation into money. If the answer requires activity-level time records, you need a time system that captures activity, not a weekly total on a paper sheet. A GPS time clock that records where and on what each entry occurred produces exactly this as a by-product of running the business, and the same data is what makes certified payroll and job costing possible.

5. Manage the modifier deliberately. Frequency drives the experience modifier harder than most owners expect. A rigorous approach to near-miss reporting, intrusion response and return-to-work generally shows up in the modifier before it shows up anywhere else. Measuring safety performance covers what to track.

6. Treat the modifier as a sales asset. It is used as a prequalification screen by owners and primes, and it appears in DOT prequalification and private-sector qualification packages alike. Safety records and EMR in prequalification covers how it gets read.

7. Prepare for the audit before it arrives. Payroll by class, overtime segregated, subcontractor certificates for the whole policy period, and job records that corroborate the activity split. An audit you can substantiate is a routine event; one you cannot is an invoice.

The seasonal-workforce problem

Traffic control has a workforce shape that interacts badly with every part of this, and it is worth treating as its own topic rather than a detail.

A company that runs eighteen people in February and forty-six in August is presenting the classification and audit system with a moving target. Three specific consequences follow.

Estimated payroll is nearly always wrong. Policies are written on an estimate, and an estimate built from a slow month understates the year badly while one built from peak overstates it. Both produce a surprise at audit — one a bill, the other a refund you financed for twelve months. The fix is unglamorous: build the estimate from last year's actual annualised payroll by class, adjusted for known changes, rather than from current headcount.

New hires land in whatever class the payroll system defaults to. When you are hiring eleven people in a fortnight, the classification assignment on each one is made by whoever is doing data entry, usually by copying the last record. This is the most common source of quietly wrong payroll allocation in a growing company, and it is invisible until audit because nothing about it looks wrong on a payslip.

Turnover erodes the training that drives the modifier. The controls that actually reduce struck-by exposure — escape route discipline, intrusion response, apparel condition — are behavioural, and they degrade every time the crew composition changes. A company with high seasonal turnover is re-establishing them continuously, which is a real cost and one that shows up in claim frequency rather than in any line of the budget. Hiring and retaining flaggers is the commercial side of the same problem.

None of these are solved by picking a better classification. They are solved by the same operational hygiene — accurate hours against activities, a real onboarding sequence, and records that exist without anyone having to remember to create them.

Why this connects to your operating system

The recurring theme above is that almost every classification and audit outcome rests on records that either exist as a by-product of how you run jobs, or do not exist at all.

Payroll split by activity, hours tied to jobs and locations, subcontractor usage per job, and an accurate picture of what each crew did on a given day are the same data that produces job costing, certified payroll, and a defensible daily ticket. Companies that keep this in a time-clock app that is separate from their dispatch system end up reconciling two records and trusting neither.

Traffic OS runs the GPS time clock against the same job record that carries the crew, the equipment and the signed daily ticket, so activity-level hours come out of the normal work rather than a parallel process. It is priced in flat monthly tiers — $499, $949 and $1,499 as of September 2026 — with no per-user charge, which matters specifically here: the employees whose classification you most need to evidence are seasonal and part-time field staff, and a per-seat licence creates an incentive to leave exactly those people off the system. The features page has the detail, and a walkthrough is the fastest way to see whether the records it produces would satisfy your auditor.

What this costs you if you ignore it

It is worth being concrete about the shape of the loss, without inventing numbers for a business we cannot see.

The loss is not usually a single dramatic event. It is three slow ones running in parallel.

You bid against a cost you have not measured. Labour burden — comp, taxes, insurance, benefits — is a multiplier on every hour you quote. A company that carries its burden as a rough percentage inherited from three years ago is bidding with a number that has drifted, and drifted in a direction it cannot see. Flagger labour cost per hour is the exercise that fixes this, and the comp component is usually the piece people have never broken out.

You discover the audit result after you have spent the money. An audit bill arrives against a year that is already closed, priced into jobs that are already delivered. There is no recovering it. The only defence is getting the estimate and the classification right at the start of the policy period, which requires knowing your payroll split before you need it.

Your modifier prices your next three years. Because the experience modifier runs on a multi-year window, a bad year is not a bad year — it is a surcharge on everything you bid for a while afterwards, plus a prequalification problem with owners who screen on it. Conversely a deliberate improvement takes time to appear, which is the argument for starting before you need it rather than after losing a bid over it.

Against those, the work described above is a page of writing, three questions to your agent, and a time system that records activity. That is a favourable trade for almost any company past a handful of crews, and it is one of the few cost levers in this business that does not require winning an argument with a customer.

The standing caveat

Workers' compensation classification rules, rates, and the organizations that administer them vary by state, and the rules change. Nothing here is insurance advice, a determination that any particular code applies to your business, or a substitute for the judgment of your carrier, your agent, or the rating organization with jurisdiction. Verify everything against your own policy and your own state — and where the answer is expensive either way, get it in writing before the policy period starts rather than after the audit.