Data provenance · methodology

The data behind every number.

Every RateScore, every peer median, every documented opportunity traces to a real federal filing, with its own sample size and refresh date. No PHI required.

n=1,247 · source TiC · refreshed Jul 2026 Insufficient sample (n<11)
The TrustChip

Every number carries its own receipt.

The same chip you see inside the app rides under each figure. It answers three questions before you ask them: how many observations, from which source, refreshed when. Nothing renders without it.

Anatomy of the chip
n=1,247
The count of documented local-peer observations behind this figure. A bigger n means a denser local market. When n is under 11, the sample is suppressed: we show that it is too small, never the value.
source TiC
Which federal filing this number is composed from. Source is one of TiC, MPFS, or Blend. It is never left blank, and never inferred.
refreshed Jul 2026
The file date, written atomically with the data itself, so freshness can never lie. If the source has not refreshed, the chip says so plainly rather than aging in silence.

Trust triplet shape: n= / source / refreshed. When two sources combine, source reads Blend and n reports the Transparency-in-Coverage leg, so a blended number never overstates its own depth.

Two scales, never conflated

RateScore is not the Index.

Two separate measures do two separate jobs. We keep them distinct so a strong row-level Index is never mistaken for a strong RateScore, or the reverse.

RateScore
300–850
A per-NPI, per-CPT score against the local-peer median and your own payers. The target is P90, the top of the local market, never a national average. This is the headline measure of where a practice stands in its own market.
Row Index
0–100
A separate per-row metric on a 0 to 100 scale. It describes a single line, not the practice. It shares no math with the RateScore band and must never be read as one.
No PHI required. Everything on this page is modeled from public federal filings. We are not HIPAA certified, we do not operate under a BAA, and nothing here is a guarantee of recovered dollars. It is a documented reimbursement opportunity: modeled, not guaranteed.
Provenance manifest

Every source, its cadence, what it feeds.

Five public federal datasets, each on its own refresh clock, each feeding a specific part of the engine. When two combine, the result is labeled a Blend.

TiC MRFsMonthly
Transparency-in-Coverage machine-readable files. Schema 2.0 is in force for files posted on or after Feb 2, 2026. These negotiated commercial rates are the backbone of the local market picture.
Feeds: local-peer medians & P90
MPFS RVU / GPCI / CFAnnual
Medicare Physician Fee Schedule components. The CY2026 conversion factor is $33.4009 non-APM and $33.5675 APM. This is the Medicare anchor every commercial rate is compared against.
Feeds: Medicare anchor
DMEPOS fee scheduleQuarterly
The CMS durable medical equipment, prosthetics, orthotics, and supplies fee schedule, updated four times a year, so DME benchmarks track the current quarter.
Feeds: DME rates
NCCI PTP editsQuarterly
National Correct Coding Initiative procedure-to-procedure edits, for example v322r0 effective Jul 1, 2026. These drive the bundling and modifier logic behind a clean comparison.
Feeds: bundling / modifier logic
Medicare utilization (Part B)Each CMS release
CMS Medicare Physician & Other Practitioners utilization by provider and service: the real service volumes behind every annualized figure. CY2013-2024 published by CMS; CY2020-2024 held in the Reddenda lake (48.5M rows); CY2023 and CY2024 live-served per NPI.
Feeds: per-NPI volume & service mix
When two sources combine

A Blend is what we call a figure composed from more than one dataset. The chip reports source='Blend', and its n reports the Transparency-in-Coverage leg only. That keeps the sample honest: a blended number never borrows depth it does not have.

Zero PHI by architecture

What we pull ourselves. What you provide.

The score exists before you send us anything. We build it from public federal data. Your own numbers are optional, and they never include PHI.

What we pull (public, no PHI)
  • NPI Registry (NPPES): the public identity and taxonomy for every NPI. No patient data.
  • Transparency-in-Coverage MRFs: payer-published negotiated commercial rates, the local market picture.
  • CMS Physician Fee Schedule: RVU, GPCI, and the CY2026 conversion factor, the Medicare anchor.
  • Medicare utilization (Part B PUF): each NPI's public service volumes; CY2023 and CY2024 live-served, five vintages held (CY2020-2024).
  • DMEPOS & NCCI: the quarterly DME fee schedule and the correct-coding edits behind a clean comparison.
What you optionally provide
  • Your contracted fee schedule or payer list, if you want the comparison sharpened to your exact payers. Optional.
  • Never PHI: no patient names, no dates of birth, no member IDs, no identified claims. The Free Snapshot never asks for any of it.
  • Nothing here requires a signed BAA, because nothing here touches protected health information.
Market context

The macro numbers, each attributed.

We cite market-level figures the same way we cite a rate: with a source. These frame the space Reddenda operates in. They are not claims about any one payer or practice.

$172B → $308B
Revenue-cycle-management total addressable market, at a 10.1% CAGR across 2024 to 2030.
per Grand View Research
1–3%
Commercial underpayment as a share of net revenue, the documented gap the local benchmark surfaces.
per MD Clarity
~20%
Average in-network claim denials, ranging from 1% to 54% by insurer.
per KFF, 2023
314M+ records · 500+ payers · 50 states · 9.2M+ NPIs
The live scope of the index. Every headline figure you see is drawn from the normalized dataset above, with its own n, source, and refresh date attached. The mother company publishes the full estate in the open, dataset by dataset, at reddenda.health ↗.

Want to see the provenance chip riding under your own numbers? Book a short discovery call, or open the app and run a public rate benchmark. Optional, no pressure, no PHI.

Book a discovery call →