To find the median commercial rate for a procedure in your state, filter negotiated-rate records to one billing context, then take the 50th percentile of what is left. Match the CPT or HCPCS code, provider type, setting of care, payer, and geography first. Compare the result with Medicare only after confirming both numbers describe the same service in the same setting and place.
A defensible median comes from comparable records. Not every rate in the state that carries the same CPT code.
Filter to a consistent procedure, setting, geography, and contract population, then calculate the median plus the 25th and 75th percentiles of the remaining negotiated rates.
For the broader workflow behind each filter, see our guide to benchmarking negotiated rates by insurer, provider, CPT code, and location.
The median is the midpoint of a comparable set of disclosed commercial negotiated rates. Half the retained records sit at or below it, half above.
That is the same idea behind any 50th-percentile benchmark. FAIR Health, for example, explains that a price at the 50th percentile means half of the fees billed by other providers for the same service were that amount or lower. FAIR Health builds its benchmarks from tens of billions of commercial and Medicare claim records dating to 2002. That is a claims-based source. Disclosed negotiated rates are a different source type, so the two should not be treated as interchangeable.
Know which number you are holding:
Matching on CPT code alone is rarely enough. Two records with the same code can describe different services, settings, and contracts.
Use these fields when they are present. Not every file carries all of them.
Hold everything constant except the one dimension you are testing. That gives you three clean views of the same market.
Fix the procedure, provider type, setting, and geography. Then calculate the median and percentile range for each payer. Report the record count beside each payer median, since a median built on a dozen records is weaker evidence than one built on hundreds.
Build a peer group: same service, same setting, same provider type, same local market. Place the target hospital, physician group, or billing entity against the peer median and the 25th to 75th percentile band. A rate above the 75th percentile is a negotiation conversation. A rate below the 25th is too.
A state median can hide large local differences between metro and rural markets. Start at the state level for coverage, then narrow to metro, county, or ZIP-based markets when the record count supports it. If a local market has too few records, say so rather than reporting a thin median.
Divide the commercial rate by the comparable Medicare amount and multiply by 100.
Percent of Medicare = commercial negotiated rate / comparable Medicare amount x 100
Hypothetical example: a $1,350 commercial rate against a $900 Medicare amount for the same code, setting, and locality is 150% of Medicare. These figures are illustrative, not observed market rates.
"Comparable" is the hard part. Medicare does not pay one national amount. CMS adjusts physician fee schedule payments with Geographic Practice Cost Indices calculated for Medicare payment localities, which can be a whole state, a metro area, or the rest of a state. Before dividing, confirm:
Gigasheet pairs commercial negotiated rates with geographically adjusted Medicare reimbursement, so the locality match is handled for you.
Yes, for facility rates. Hospital files support facility-rate benchmarking once they are filtered to the same service and billing context.
CMS has required every U.S. hospital to publish pricing online since January 1, 2021, including a comprehensive machine-readable file of all items and services. The two main public sources answer related but different questions:
Hospital files show one facility's contracts across payers. Plan files show one payer's contracts across providers, including physicians. Using both gives the fuller market view.
Gigasheet provides nationwide negotiated-rate data across payers and providers, from both Transparency in Coverage and hospital price transparency files, with commercial rates paired with geographically adjusted Medicare benchmarks. The data is refreshed continuously as payers and hospitals post new files.
Teams that use it include payment integrity, repricing, and cost-containment firms; TPAs, self-insured employers, and benefits consultants; care navigation and patient advocacy organizations; hospitals, physician groups, and MSOs; and healthcare technology, data, and AI companies.
See Gigasheet pricing, explore the data, or Request a Demo.
Last updated November 2, 2026.
Filter to comparable negotiated-rate records, then calculate the 50th percentile. Match the CPT or HCPCS code, provider type, setting, payer, and geographic market first, and report the record count with the result.
No. Federal Transparency in Coverage rules define a negotiated rate as the amount a plan has contractually agreed to pay an in-network provider. The allowed amount published in those files is an out-of-network figure: the maximum a plan will pay an out-of-network provider.
Yes. Hold the procedure, setting, provider type, and geography constant, then compare the median and percentile range for each insurer or plan in the data.
Yes. Build a peer group with the same service, setting, provider type, and local market, then report the provider's rate alongside the peer median and 25th to 75th percentile range.
Records with the same code can differ in setting, provider type, payer product, network, modifiers, units, and geography. Each is a separate contract, so rates are not expected to match.
The commercial rate is 1.5 times the comparable Medicare amount. Confirm both amounts use the same code, setting, unit, and Medicare locality before reporting it.
Treat it as a starting benchmark. A state median may not reflect the provider mix, payer mix, or setting in the local market where the rate will apply.
Yes, as long as the local market has enough records to support a median. Start at the state level, then narrow, and flag any market where the record count is thin.