Machine learning in payer contracting — what fee schedule analytics actually tells you

Every payer sends a fee schedule. Most practices review it at a high level — maybe check a handful of high-volume CPT codes, see that the rates look reasonable, and sign. The payer knows this. The fee schedule is written to pass that review.

ML-driven contract analysis changes the leverage. Instead of reviewing ten codes by hand, the platform compares every contracted code against Medicare benchmarks and regional market rates, surfaces the outliers, and tells you exactly which codes are underpaid and by how much.

The negotiation changes when both sides have data. Here is what the data says.

Why benchmarking matters

Commercial payer fee schedules are expressed as either fixed dollar amounts per CPT code or as a percentage of Medicare. The percentage-of-Medicare model is the easier one to benchmark — if the contract says 115% of Medicare, you can compute the exact reimbursement for any code using the published Medicare Fee Schedule.

But even in percentage-based contracts, there are complications: the Medicare base rate used in the calculation is often not the current year’s MPFS, the geographic GPCI adjustment may or may not be applied as written, and some codes are carved out with fixed rates that underperform what the percentage formula would yield. Fixed-rate schedules are harder to benchmark because you need external rate comparables — and that’s where ML-driven analysis earns its keep.

Metolius ingests fee schedules as structured data and runs each contracted rate against two benchmarks: the current Medicare Physician Fee Schedule for the same code and geography, and a regional market-rate model built from aggregated contracting data across the payer’s book. The output is a per-code variance table: contracted rate, Medicare rate, market median, and the delta between contracted and market.

What the CPT-code analysis actually surfaces

When you run ML benchmarking on a typical primary care or specialist contract, a few patterns appear consistently:

Evaluation and management codes (99213–99215) are usually close to market. Payers know these codes get scrutinized and price them competitively. Negotiating leverage here is limited unless your volume is high enough to justify a volume-based rate.

Procedure codes are where the gap lives. Minor surgical procedures, diagnostic codes with technical components, injection codes, and modifier-heavy billings are where per-code rates drift farthest from market. These codes get less visibility during initial contract review because they’re not the first thing a practice checks — but they can represent a significant share of gross revenue for specialty practices.

Global packages are frequently mispriced. When a payer’s fee schedule includes global surgical packages, the package rate often doesn’t reflect the component rates. Running the global package rate against the sum of its technical and professional components using Medicare’s RVU model sometimes reveals underpricing by 15 to 20 percent.

Telehealth and remote monitoring codes are the newest gap. Payers have been slow to price CPT 99453, 99454, 99457, 99458, and the telehealth-equivalent E/M codes competitively. Practices that added remote patient monitoring or expanded telehealth post-pandemic often signed fee schedules that predated the payer’s rate modeling for these codes. Benchmarking them now reveals material gaps in some contracts.

Payer-class variance: not all underpayments are negotiable

ML analysis doesn’t just tell you which codes are underpaid — it tells you which underpayments are negotiable. Payer class matters.

Medicare Advantage plans have limited rate flexibility. Many MA plans pay at or near traditional Medicare rates by design, and the contracting structure doesn’t leave much room for code-level negotiation. The leverage in MA contracts is usually in the administrative terms — clean claim submission windows, timely pay guarantees, and dispute resolution — not in the fee schedule.

Large commercial plans (BCBS, Aetna, Cigna, UHC) in competitive markets are where benchmarking creates real leverage. These payers have rate-setting authority and respond to data-backed negotiation. If your analysis shows that three procedure codes in your contract are at 85% of Medicare while the market median is 105%, that delta is a concrete ask. Payers negotiate from data. Bring data.

Small regional plans and self-funded employer plans are the highest-leverage opportunity. These plans often set rates based on what they paid the last provider in the same role, which may be years out of date. Market benchmarking here can reveal rates that are 20 to 30 percent below current market — and the plan sponsor often doesn’t know. The conversation becomes educational rather than adversarial.

Tracking rate changes across contract cycles

Fee schedule analysis isn’t only useful at signing. Rates change across contract cycles, sometimes without clear notice. Payers update fee schedules annually or upon contract renewal, and the changes are not always disclosed as a line item.

Metolius tracks contracted rates across contract cycles and surfaces the delta when a new fee schedule is loaded. If a payer’s 2025 fee schedule pays less for a code than its 2023 schedule did — which happens more often than practices expect — the variance report flags it. You know what changed before you sign the renewal, not six months after.

That tracking also establishes a negotiation baseline. When a payer proposes a new fee schedule and claims it represents a rate increase, the historical comparison tells you whether that’s true on a weighted-volume basis for your specific code mix — not just on the headline rate for a few anchor codes.

What to do this week

1. Pull your top 20 CPT codes by gross charge volume. These are the codes where benchmarking has the most dollar impact.
2. Compare contracted rates against the current Medicare Physician Fee Schedule. The MPFS is public and searchable by HCPCS code and locality. Even a manual comparison on 20 codes will surface outliers.
3. Flag any codes where your contracted rate is below 100% of Medicare. For commercial payers, that’s a floor, not a target — and it’s a starting point for a data-backed renegotiation ask.
4. Check your telehealth and RPM codes. If you’ve added those services in the last two years, your fee schedule may predate your payer’s current rate for them.

If you want a full contract benchmarking run on your fee schedules, talk to us. We’ll run the analysis in Metolius and walk through the variance report with you before your next renewal cycle.

— Medical Credentialing Services