Introduction
Workers' compensation data often shows costs or claim outcomes drifting the wrong way long before the renewal, the loss run, or the financial report makes it obvious. By then, the fix is slower and more expensive.
Benchmarking is how you spot that drift early. It compares your program, claims, safety, and operational results against relevant external peers and your own history.
Even national trends can diverge: NCCI's October 2026 update reports preliminary 2025 frequency down 2% while lost-time severity rose 4%. A single headline number would hide that.
This article covers what to benchmark, why it matters, and how to do it, while accounting for state rules, class codes, payroll, claim maturity, program design, and data quality.
Key Takeaways
- Benchmarking goes beyond premiums to the drivers of cost, risk, and claim outcomes.
- Useful comparisons need matched peers, consistent definitions, and normalized data.
- Pair outside peers with your own year-over-year trends.
- Value comes only when a variance gets a root cause, an owner, an action, and a follow-up.
What Is Workers' Compensation Benchmarking, and Why Does It Matter?
Benchmarking compares your results with a chosen standard: comparable employers, industry data, state or class-code data, and your own prior performance. NCCI's experience rating is a built-in version. It compares each employer with the average employer in the same classification.
"Benchmarking" covers four different exercises:
- Program: premium, structure, and retention choices
- Claims outcomes: frequency, severity, and closure patterns
- Safety: injury rates and preventive activity
- Claims handling: speed, medical management, and litigation
What a Variance Can Tell You
A cost gap might come from payroll or classification changes, loss experience, claim severity, carrier pricing, program structure, or operational practice. Benchmarking helps you isolate which of these is driving the difference.
The goal is to find meaningful variance, understand its cause, and decide whether it is acceptable or actionable, not simply to post the lowest number in every category. A low-cost outlier can signal under-reserving or underreported claims.
Comparisons can also mislead. State regulations, rating bureau methods, class-code mix, claim development, deductibles, retentions, and coverage structure all change what a number means.
This article is general information, not insurance, actuarial, legal, or claims-handling advice. Work with a qualified broker, actuary, carrier, TPA, or attorney on your specific program.
What Should You Benchmark?
Premium and Pricing Drivers
NCCI illustrates classification premium as payroll divided by 100, multiplied by the class rate. Classification premiums are summed before the experience mod applies. The mod usually uses the latest three years of experience, and a January 2026 renewal excludes 2025 policy experience.
State details differ:
- New York: the approved rate is the approved loss cost times a carrier-specific multiplier
- California: advisory pure premiums exclude insurer overhead, and insurers set their own rates
- Independent bureaus: California, Delaware, Indiana, Massachusetts, Michigan, Minnesota, New Jersey, New York, North Carolina, Pennsylvania, and Wisconsin
- Monopolistic states: North Dakota, Ohio, Washington, and Wyoming run their own plans
Verify the formula and terminology for each state you operate in.
Loss Experience and Claim Outcomes
Define every measure before comparing it:
- Frequency: claim count divided by exposure (payroll, head count, or hours)
- Severity: average cost per claim
- Paid: amount received by the claimant as of the evaluation date
- Case-incurred: paid plus case reserve
- Ultimate: estimated value once all claims for the period are reported, paid, and closed
Also track:
- Open versus closed claims
- Reserve changes and claim age
- Injury type and cause of loss
- Litigation status
Safety and Leading Indicators
Backward-looking measures include the e-mod and injury rates. BLS puts 2024 manufacturing at 2.7 recordable cases per 100 full-time workers, calculated as cases divided by hours worked, times 200,000. That is an injury measure, not a dollar-loss measure.
OSHA describes leading indicators as proactive measures of safety activity. Useful examples include:
- Near-miss reporting and safety observations
- Training completion and corrective-action closure
- Inspection results and return-to-work activity
Claims Operations and Program Structure
Operational measures worth tracking include:
- Report lag and time to first contact
- Closure and litigation rates
- Medical management outcomes
- Allocated expenses and administrative charges
Be careful with lag: one WCRI-linked study found that more than 30 days from disability onset to first indemnity payment was associated with an 18-point higher likelihood of attorney representation. That is payment delay, not injury-report lag. Don't swap them.
Program structure matters too. Compare retained losses and premium separately under each model:
- Guaranteed cost: claims filed during the policy period don't change final premium
- Large deductible: the carrier pays and the employer reimburses
- Retrospective rating: premium adjusts as losses develop
Types of Benchmarking
APQC's categories map well to workers' compensation:
| Type | Question it answers |
|---|---|
| External peer | How do we compare with similar employers? |
| Internal year-over-year | Are we improving against ourselves? |
| Process | How do our claim practices compare? |
| Financial/program | Is our structure and spend still appropriate? |
Use them together. No outside employer is a perfect match, and a single ranking hides too much.
How Do You Conduct Workers' Compensation Benchmarking?
1. Define the Question and Scope
Start with a specific objective:
- Explaining a renewal increase
- Investigating rising severity
- Evaluating a claims administrator
- Prioritizing safety spend
- Testing program structure
Then lock the scope: time period, states, business units, class codes, claim maturity, coverage structure, and measures.
2. Assemble and Document the Data
Pull the files you will rely on:
- Payroll records and policy documents
- Loss runs, claim notes, and reserve histories
- OSHA logs and safety reports
- Vendor reports and claims-system exports
Record each data owner, definition, valuation date, and missing field.
3. Choose Peers and Sources
Sources include rating bureaus, carriers, brokers, TPAs, actuaries, government data, and your own history. Each has limits. For example, WCRI's CompScope covers state systems, not your claims administrator's service levels. WCRI's overall 2026 edition covers 18 states, with its medical studies using different date ranges.
Milliman's analysis points to firm size, risk tolerance, industry, geography, loss control, and claims management as matching factors. Also match on:
- Class-code mix and payroll scale
- Retention and program type
- Claim maturity
A broad industry average alone won't do.
4. Normalize the Comparison
Label or adjust for:
- Payroll exposure and class codes
- State rules and inflation
- Loss development
- Deductible or retention treatment
- Whether loss-adjustment expenses are included
Bureau loss costs applied to your payroll, per $100, give a fairer comparison than raw totals.
Never mix paid, incurred, and ultimate losses without a consistent basis. Incurred development factors don't apply to paid losses. Immature claims will look better than they are.
5. Analyze Position and Trend
Compare rates, ratios, distributions, and trend lines. Avoid unsupported claims about what counts as "good." Then segment outliers by:
- State, location, and class code
- Cause of loss and claim type
- Litigation status and administrator
- Policy year
That separates systemic problems from isolated events.
6. Make It a Repeating Cycle
Write a report recording the benchmark definition, sources, assumptions, variance explanations, actions, owners, and review dates. Have brokers, actuaries, carriers, TPAs, or counsel help interpret results where appropriate.
How Do You Turn Benchmark Findings Into Action?
From Variance to Root Cause
An unfavorable result is a prompt to investigate, not a conclusion. Check:
- Classification accuracy
- Delayed reporting
- Reserve adequacy
- Claim-handling practices
- Safety hazards
- Medical management and litigation strategy
- Return-to-work processes
- Program design
Assign Interventions and Indicators
For each priority issue, define:
- Corrective action and owner
- Expected business effect
- Leading indicator and lagging outcome
- Review cadence
Separate what you control from external market or regulatory factors.
For example, if severity is rising in one state, a leading indicator might be time to first medical visit, and the lagging outcome is average incurred cost at a fixed maturity.
Connect Claims Data to Portfolio Decisions
Structured historical claims data lets claims professionals and defense teams spot recurring patterns, compare outcomes, and surface high-risk segments.
OraClaim automatically structures and benchmarks past cases so teams can anticipate outcomes and refine strategy. For workers' compensation defense, it adds medical chronology, MMI and disability-rating flags, apportionment analysis, and reserve exposure tuned to state statutes.
Common Mistakes to Avoid
Using a Poorly Matched Peer Group
Mismatched peers produce false conclusions. Avoid comparing across different:
- States or jurisdictions
- Class-code mixes
- Payroll exposures
- Claim maturities
- Program structures
State your comparison boundaries explicitly. Published loss costs are not rates; approved multipliers produce the rates.
Inconsistent Definitions and Immature Claims
Keep definitions consistent. Do not mix:
- Paid and incurred values
- Immature and developed claims
- Different retention treatments
- Inconsistent expense inclusion
Document the loss basis and validate source data before calculating variance.
Treating Rank as Diagnosis
A benchmark shows a signal, not its cause. Follow with analysis, stakeholder review, an action plan, and a later measurement of whether the intervention moved the trend.
Frequently Asked Questions
How are workers' compensation rates determined?
Rates reflect state rules, class code, payroll exposure, loss costs or filed rates, carrier pricing factors, experience modification, and program structure. In New York, for instance, rate equals loss cost times a carrier multiplier. The exact calculation varies by state and policy.
How do you do workers' compensation benchmarking?
Define the objective, gather consistent claims and exposure data, and choose comparable peers. Normalize for relevant differences, compare current position and history, then assign corrective actions with owners.
What are the types of benchmarking?
External peer, internal year-over-year, process, safety, claims-outcome, and financial or program benchmarking. Use external for context, internal for progress, and process for operational gaps.
What metrics should workers' compensation benchmarking include?
Premium and rate drivers, e-mod, frequency and severity, developed losses, report lag, closure and litigation rates, safety indicators, return-to-work, and claim expenses. Availability and consistent definitions limit what you can use.
How often should workers' compensation benchmarking be updated?
Align reviews with renewals and internal reporting, and add reviews after material changes in claims trends, payroll, operations, states, program structure, or claims administration. No single cadence fits every organization.
