What Is a Machine-Readable File? A Facility's Guide to MRFs

If you've started hearing that "the rates are public now," the thing people are pointing at is the machine-readable file — the MRF. It's the raw material behind every claim you'll read on this blog about payer transparency. Before you can use that material, it helps to know exactly what it is: what's in it, what isn't, and why having the file is a long way from having an answer.

What the Transparency in Coverage rule requires

The federal Transparency in Coverage rule requires health plans and insurers to publish the negotiated rates behind their networks in machine-readable files. The in-network rate files became a broad requirement in 2022, and payers now post them publicly on an ongoing basis.

"Machine-readable" is the key phrase. These files are built for software to parse at scale, not for a person to open and read. That single design fact explains most of what makes them powerful and most of what makes them frustrating.

What's actually in the file

At its core, an in-network MRF lists negotiated rates — the amounts a payer has agreed to pay — tied to two other things:

• Billing codes, which identify the service the rate applies to; and
• Provider references, the identifiers that tie a rate to the provider or group it was negotiated with.

That's the spine of the file: a rate, the code it's for, and a pointer to who it belongs to. Published across an entire payer's network, that becomes an enormous table — a single payer's file can be staggeringly large.

What's not in the file

Just as important is what an MRF leaves out. It is not an explanation of your benefits or a member's bill. In particular, the files generally don't tell you:

• Benefit design — what a given plan actually covers, or under what conditions;
• Member cost-sharing — deductibles, copays, coinsurance, the amounts a patient owes;
• The context around a rate — the contract structure it sits under, whether it's still active, or which real-world entity stands behind the provider reference.

So an MRF answers "what rate is on file for this code and this provider reference?" It does not answer "what will this facility actually be paid?" Those are different questions, and the gap between them is where most misreadings happen.

Why the raw files are so unwieldy

Three things make raw MRFs hard to use directly:

1. Scale. The files are big enough that opening one the normal way isn't an option; they're meant to be processed programmatically.
2. Noise. Because they're machine-generated and comprehensive, they carry a lot that isn't useful — including rates that are duplicated, outdated, or no longer active. (We cover those ghost rates separately.)
3. Identity. The provider references don't come with a clean, human-readable "this is Facility X." Resolving a reference to the real organization behind it is genuine work, not a lookup.

None of this is a flaw you can complain your way out of — it's just what the raw material is. The value isn't in having the file. Everyone can download the file.

What you can — and can't — conclude from a raw MRF number

Here's the trap. You find your code, you find a provider reference that looks like a peer, and there's a number. It feels like an answer. It usually isn't — at least not yet.

A raw MRF number tells you a rate exists on file. It does not, on its own, tell you:

• whether that rate is still a live contract;
• how it's structured (a per-diem, a case rate, a percent-of-charge — all of which change what the number means);
• which real entity the provider reference resolves to; or
• whether a behavioral health carve-out means the rate you're reading isn't even the one that governs the care.

Treat the number as a settled payment and you'll walk into a negotiation confident and wrong. The better mental model — the one worth internalizing before anything else — is that an MRF rate is a signal, not a settled payment. We make that case in full in MRF data isn't your reimbursement rate.

From file to intelligence

The MRF era genuinely changed things: the raw negotiated-rate data is public now in a way it never was. But "public" and "usable" are not the same word. Turning a giant, noisy, identity-ambiguous file into a number you can defend in a payer conversation takes resolving it to the real facility, reading it in context, filtering what's dead, and scoring what's left. That's the work TierBench does — see how the methodology works, or browse the rate-data FAQ for the short answers.

When you want to know where your rates actually stand — resolved, in context, and scored for confidence — see which payers we hold rate data on or book a 20-minute call.

---

← Back to all posts