Benchmarking Is a Question, Not an Answer
A benchmark can be useful. It can tell a CEO or CFO that a question is worth asking. It cannot tell that leader what to do next.
That distinction matters in health benefits. A peer comparison may show a cost level, a funding arrangement, or a common benefit practice. It does not show whether two employers have the same workforce, geography, network access, pharmacy mix, claim volatility, stop-loss terms, or vendor contracts. Treating a benchmark as a verdict can turn a helpful signal into an expensive shortcut.
The source idea: national data has boundaries
The U.S. Department of Labor’s 2026 report to Congress on self-insured group health plans uses Form 5500 information and other financial data to describe the self-insured market. It separates self-insured, fully insured, and mixed-insured arrangements and notes that its large-plan findings are based on available filing data through 2023. The report also describes limits in matching filings to outside financial information.
That is not a weakness to dismiss. It is a useful reminder about what a responsible comparison looks like: first understand the population behind the number, then decide whether it is comparable enough to prompt a closer review.
Start with the decision, not the spreadsheet
Before asking whether a plan is above or below a benchmark, name the business decision that the answer is supposed to improve. Is the company deciding whether to remain fully insured, evaluate self-funding, change stop-loss structure, review the PBM contract, or investigate a sharp renewal increase? Each question needs a different comparison set and a different level of detail.
A broad national figure may be fine for a board conversation about market direction. It is not enough to price a renewal or select a vendor. For that, leaders need plan-specific evidence: enrolled population, paid claims, large-claim patterns, pharmacy detail, network and site-of-care utilization, administrative fees, stop-loss provisions, and the actual services included in competing proposals.
Ask what the number leaves out
Two employers can report a similar per-employee cost and have very different plans. One may have a concentrated workforce near a high-cost hospital system. Another may have a geographically dispersed population. One may include a carved-out pharmacy arrangement or a different stop-loss deductible. A number that looks comparable may be measuring different things.
The practical work is not to reject all benchmarks. It is to annotate them. What year is the data from? Which employers are included? Is the measure paid claims, total plan spend, premium equivalent, or something else? Are large claims and pharmacy included? What contract terms or employee contributions are outside the figure?
Turn a benchmark into a review agenda
For Paul H. Flowers Jr. at Superior Insurance Advisors and Paul.Health, a good benchmark is an invitation to a more disciplined conversation, not a sales slogan. It should help an employer request the right reports, define the comparison set, and identify where an independent review could add clarity.
That approach also keeps the leadership team from chasing a single “average” while overlooking the decisions it can actually control. A CFO may not be able to change the local care market overnight. The company can still improve its data access, contract questions, vendor accountability, and process for evaluating major purchasing decisions.
Source: U.S. Department of Labor, “2026 Report to Congress: Annual Report on Self-Insured Group Health Plans” and its Appendix B. This article is original educational commentary, not legal, tax, investment, or fiduciary advice. Plan documents, contracts, and applicable law govern.
Five questions before using a benchmark
- What decision will this comparison help us make?
- Does the benchmark population actually resemble our workforce and care market?
- What costs, contract terms, and benefits are included or excluded?
- Which plan-specific reports would confirm or challenge the comparison?
- Who will own the next review step and document the conclusion?
At your next benefits meeting, could your team explain what a benchmark is measuring—and what it is not?