Pull one insurer's published price file for imaging in a single metro and it reads like a crowded market. In BlueCross BlueShield of Tennessee's home-network rate file for the Nashville area, hundreds of imaging providers appear — each with a name, an NPI, and their own rows of negotiated rates for CT, MRI, ultrasound, X-ray, and mammography. Every provider looks like a separate seller who sat across a table and negotiated a price.
Most of them didn't.
We took ten common imaging codes and, for each provider on them, built a rate vector — the exact set of negotiated component rates the file lists for all ten procedures. Then we asked how many providers share the identical vector, matched to the cent.
228 providers carry one byte-identical ten-code rate vector. Same numbers, same components underneath each one, across every procedure. Not approximately. Identically.
Identical rates are usually a shared contract. These aren't.
When two providers share a rate schedule to the cent, the ordinary explanation is that they share a contract — a group, an employed practice, a negotiated network deal. That story survives right up until you look at who the 228 are.
They resolve to at least ten different provider taxonomies. 195 are radiologists — no surprise for imaging codes. The other 33 are not radiologists at all. Among them:
| Specialty | On the identical vector |
|---|---|
| Radiology | 195 |
| Nurse practitioner | 10 |
| Speech-language pathology | 4 |
| Cardiology | 3 |
| Neurosurgery | 3 |
| Physical medicine & rehab | 3 |
| Physician assistant | 3 |
| Nuclear medicine | 2 |
| Dermatology | 1 |
| (student registrants) | 4 |
A speech-language pathologist and a radiologist do not sit down together to negotiate a joint CT and MRI contract. Neither do a dermatologist, a cardiologist, and a nurse practitioner. These providers have nothing to negotiate with each other on imaging — several of them don't perform these procedures at all. Yet the file gives every one of them the same ten imaging prices, to the penny.
There is only one thing that puts the identical imaging rate on a radiologist, a cardiologist, a dermatologist, and a speech-language pathologist: it is the payer's standard fee schedule — the default rate table loaded for any provider who never negotiated individual imaging terms. It is not 228 contracts. It is one schedule, applied 228 times.
Why this breaks rate benchmarking
Rate benchmarking almost always starts by counting providers. You pull the file, you see how many NPIs are listed at what price, and you build a distribution: this practice is at the 80th percentile, that one is at the median, here is where the market sits. The provider count is the denominator for everything that follows.
The provider count is the wrong denominator. In this file, 228 line items are one price decision. Treating them as 228 independent data points doesn't make the benchmark more precise — it makes it confidently wrong, because it stacks a single payer choice up as if it were a market consensus. The distribution looks dense and settled. It's mostly one schedule wearing 228 name tags.
The real question a benchmark should answer is not "how many providers are near this rate," but "how many distinct schedules exist, and which one am I on." Those are very different numbers. Here, 228 apparent negotiating parties collapse into a single one — and that one is the payer, showing up as a default schedule assigned to providers who never negotiated individual imaging terms. Every additional cluster like it shrinks the real count of pricing decisions further below the provider count.
For context on what that default is worth: this standard schedule is not a cheap one. For a routine CT of the head, the shared rate is the highest rate in the file for that code — roughly 1.75 times the market median for that scan. Across the ten codes the schedule runs on the order of three to five times the 2026 Medicare rate depending on the procedure, with screening mammography at the low end near three times. In other words, the "default" most of these providers are on is a high, payer-authored commercial schedule they likely assume they negotiated.
The tell: cross-specialty rate collisions
Here is the reusable part, worth keeping:
When providers in unrelated specialties share a rate to the cent, you are not looking at a negotiation — you are looking at a payer default.
Two radiologists at the same MRI price could plausibly be one practice. A radiologist and a speech-language pathologist at the same MRI price cannot be — they have no shared contract to explain it. The collision across specialties is the fingerprint of a standard schedule. Once you see it, you stop counting providers and start counting schedules, and the market gets much smaller and much clearer:
- Count schedules, not providers. The number of distinct rate vectors is the real number of pricing decisions in a file. It is a small fraction of the NPI count.
- Cross-specialty collisions flag the payer's default. If a dermatologist and a radiologist carry identical imaging rates, that vector is the standard schedule, not a contract.
- Ask which schedule you are on before you ask where you rank. Being at the 60th percentile of a distribution that is mostly one default rate tells you almost nothing. Knowing whether your rate is the default — or something the payer agreed to only for you — tells you everything.
How we know
This is a correction as much as a finding. An earlier version of this analysis ran on a derived extract of the same file — and that extract had silently dropped the component modifier and mislabeled every rate, which inflated the apparent provider and schedule counts. So we went back to the raw source: the public BCBS-TN 890 in-network-rates file as published on 2026-06-27, read directly rather than through our own pipeline. We matched the exact raw negotiated rates that define the shared vector, resolved the file's provider references to NPIs, and joined those NPIs to the national provider registry (NPPES) to recover each provider's taxonomy. The cross-specialty spread — a radiologist, a cardiologist, a dermatologist, a speech-language pathologist, and a student on one identical ten-code vector — is what the raw file says, not what a summary of it said.
No provider names appear here, and none are needed. The point is not who these 228 are. The point is that a price file can list them as 228 separate sellers when they are one schedule, and if you count the sellers instead of the schedules, you will draw a market that does not exist.
That is the whole job of reading a payer's rate file well: telling the negotiations apart from the defaults. It is most of what we do.