Paying a bill or have a question about your account? Make a paymentConsumer help
Paying a bill or have a question about your account? Make a paymentConsumer help
Revenue cycleA vendor demo won't tell you what a partner would do with your denials. Here's how to set a baseline, score bidders on the same sheet and test one payer and denial class before you commit.
Judge a partner on outcomes measured on your own claims: overturns, paid yield, cash, time and cost per claim. Then look at how it works denials by payer and denial class, how it protects your data, whether the contract lets you test it on one carve-out first, and whether it gives you references you can call.
HFMA's Claim Integrity Task Force published standard denial measures so providers and health plans can compare like with like (HFMA guidance, 2022, updated February 2026; the page is sponsored by CloudMed, ReveCore and TruBridge). Use its definitions so every bidder reports the same way.
| Measure | How to calculate it | Why it matters |
|---|---|---|
| Initial denial rate | Initial denials ÷ claims submitted, by count and by dollars (HFMA) | Shows how much work enters the queue |
| Initial denials overturned | Share of initial denials overturned and paid (HFMA) | The core result, by payer and class |
| Days from denial to appeal | Days from the denial remittance to the appeal submission date (HFMA) | Slow appeals run into filing limits |
| Paid yield | Cash received on overturned claims ÷ expected payment | An overturn that pays little isn't much of a win |
| Cost per claim worked | Staff time, vendor fees and software share ÷ denials worked | Tells you which claims are worth appealing |
| Denial write-offs | Net dollars written off as denials ÷ average monthly net patient service revenue (HFMA MAP Key AR-6) | What slips through today |
Cost per claim is where many baselines stop short. Premier's survey of 280 hospitals put the average cost of fighting a denial at $57.23 for 2023 claims, up from $43.84 for 2022 claims (Premier, February 2025, vendor-published). Your own number decides the claim value below which an appeal loses money.
Payers deny at very different rates. Among insurers selling HealthCare.gov plans, in-network denial rates ranged from 3% to 36% in 2024 (KFF, March 2026). Overturn rates vary too: in Premier's survey of 2022 claims, 45.8% of initial Medicaid denials were overturned, against 60.5% for commercial plans (Premier, March 2024, vendor-published).
A vendor that reports one rate across all clients can't show you how it would do on your payer mix. Ask for results split by payer type and denial class, with the period and the number of claims behind each figure.
This is a worked example. The weights are ours; set your own before the first demo, and don't change them after.
| Criterion | Weight | What to ask for | Evidence to request |
|---|---|---|---|
| Results on claims like yours | 30% | Overturn rate, paid yield and days to appeal by payer and denial class | A report with dates and claim counts, from a client with a similar payer mix |
| Cost per claim | 15% | Fully loaded cost per claim worked, and how it changes with claim value | The pricing basis and a sample invoice |
| Workflow by payer and class | 15% | How appeals are prioritized, who reviews them, and how filing deadlines are tracked | A walk-through on a sample of your own denials |
| Data security | 15% | Scope and date of security audits, where your data sits, who can see it | Audit reports and the issuing body's name |
| Contract terms | 15% | A carve-out test, a written baseline, data return at exit, audit rights | Draft contract language |
| References | 10% | 2 or 3 clients you can call yourself | Names and phone numbers, with the client's permission |
A carve-out gives one payer or denial class to the new partner while your current team or vendor keeps the rest. It's the cleanest way to compare, because both sides work the same period.
Put the baseline in the contract: the measures above, the months they cover, and how results will be counted. Add who carries the risk when a claim nears its filing limit, how your claims history and appeal records come back at exit, and your right to audit.
A projection of your results before the vendor has worked any of your claims. That number can't be known yet.
Performance shown only as a blend across clients and denial types.
References who describe the relationship but can't give a number that changed.
Claims about automation without a plain answer to who reviews each appeal before it goes out.
Often, yes. Check whether your current agreement gives the incumbent exclusive rights to all denials or only to named work. If it names the work, you can carve out one payer or denial class for a new partner. Agree on the baseline and the counting rules before the test starts, so both sides measure the same way.
It depends on who does the work. Software helps a team that has the staff and payer knowledge but needs better sorting and tracking. Outsourcing, such as denial management, adds people who work the appeals. Compare both on the same measures: cost per claim worked, overturn rate by payer and class, and days to appeal. Some hospitals use software in-house and outsource the classes they can't staff.
You can keep the work in-house and add software, outsource one function such as denials, early-out or bad debt, or bring in a partner for a single payer or denial class. Each keeps more control than a full outsourcing deal. The early-out and bad-debt vendor RFP lists what to ask.
Start with the denial break-even worksheet. It shows the claim value below which an appeal loses money on your costs.
Results vary by payer mix and client program.
General information, current as of Nov 5, 2026. It isn't legal advice. Rules change and differ by state, so check the sources linked here and talk to your own counsel before you act on it.
The Denials Economics Sprint looks at one payer and one denial class, with defined claim-value bands and a baseline, and returns overturn rate, paid yield, cash, time, human effort, cost per claim, the economic threshold and the expansion case.
