How to Reduce Claim Denials: Fix Them Where They Start

A claim comes back denied. Someone on the billing team corrects it, resubmits it, and moves on to the next one. A week later the same denial arrives on a different claim. Most revenue-cycle teams know their denial rate. Knowing how to reduce claim denials is harder, because each denial is a symptom of something that went wrong earlier: at scheduling, at registration, in the chart, or in a payer rule nobody updated.

The way out is to stop treating denials one claim at a time and start fixing them where they begin. Here's how healthcare organizations can do that, in the order that usually pays off first.

Where do claim denials start?

Denials are common, and providers report they are rising. In Experian Health's State of Claims 2025 survey of 250 healthcare professionals, 41% of providers said they face denial rates of 10% or higher (Experian). Premier's analysis of respondents representing 516 hospitals found that nearly 15% of claims submitted to private payers in 2022 were initially denied (Premier).

The causes cluster at the front of the revenue cycle. Experian's top three reasons for denials were missing or inaccurate data (50%), authorizations (35%) and incomplete or inaccurate patient registration data (32%) (Experian Health). In a 2026 Inovalon survey of more than 400 industry leaders, about four in five tied denied claims to at least one front-end workflow (TechTarget).

That's good news, because it means much of the problem is preventable. Change Healthcare's 2020 Revenue Cycle Denials Index, built from 102 million transactions across more than 1,500 U.S. hospitals, found that 86% of denials were potentially avoidable. It also found that half of denials came from front-end issues such as registration and eligibility, authorization and service not covered (HFMA). The index is older than the surveys, but the pattern is the same: most of the work to prevent a denial happens before the claim is sent.

What should you fix before a claim goes out?

Prevention costs less than recovery. Premier's respondents that fought denials spent an average of $43.84 per claim doing so (Premier), and that's for a claim the organization may still lose. Four front-end checks catch most of what can be caught.

Verify eligibility and authorization early, then again

Check coverage when the visit is scheduled, again at registration, and again shortly before the date of service for scheduled patients, because coverage changes. Ask the payer about the specific service, not just whether the plan is active. For anything that may need prior authorization, check the payer's current rules, record the authorization number, and confirm the authorized service, site and dates match what is actually delivered. A procedure that moves from outpatient to inpatient, or from one facility to another, often needs a new or updated authorization. We cover how eligibility verification and cost estimation work best as one workflow in a separate post.

Get registration data right the first time

A wrong date of birth, a misspelled name, an outdated member ID or the wrong subscriber relationship can each stop a claim. So can listing payers in the wrong order when a patient has more than one. Scan the insurance card and photo ID, validate member-ID formats against the payer, and confirm which plan pays first. Then send every registration-related denial back to patient access as a regular report by location, so the people who enter the data see what it cost.

Close coding and documentation gaps

Medical necessity and level-of-care denials usually trace to documentation that does not support the service billed: observation versus inpatient status, missing clinical detail, or a diagnosis that does not match the order. Clinical documentation review and coder queries before final billing catch many of these. For level-of-care questions, involve a physician advisor before the claim drops, not after the denial arrives.

Scrub every claim against payer-specific edits

Generic claim edits catch formatting errors. Payer-specific edits catch the rules that differ by contract, plan and service. Each time a new denial pattern appears, add an edit so the same error is flagged before the next claim goes out. Watch filing deadlines too: Medicare requires claims within one calendar year of the date of service, and a claim denied for late filing cannot be appealed (CMS). Commercial payers set their own limits in contracts and provider manuals.

How do you find the root cause of a denial?

Every adjustment on a remittance carries a Claim Adjustment Reason Code (CARC). X12, which maintains the code set, says these codes "describe why a claim or service line was paid differently than it was billed" (X12). Remittance Advice Remark Codes (RARCs) add detail. Together with the payer, they are the raw material for root-cause work. A practical loop looks like this:

1. Capture every denial in one place, with payer, CARC, RARC, service line, department, location, provider and dollars.
2. Group codes into categories your team can act on. HFMA's Claim Integrity Task Force suggests categories such as eligibility, authorization, medical necessity and level of care, and notifications (HFMA).
3. Rank by dollars and by count, payer by payer. A reason that looks minor overall can dominate one payer or one service line.
4. Read a sample of claims end to end for each top category, from scheduling to remittance.
5. Send each cause back to the department where it starts, with an owner, a fix and a date, then report the trend to that department every month.

Three misreads come up often:

• The reason code is not the root cause. CARC 16, "Claim/service lacks information or has submission/billing error(s)," says what the payer objected to, not which step failed. Only the claim sample shows that.
• Claim-level counts hide partial denials. That's why HFMA's task force calls line-level reporting best practice.
• Counting every resubmission as a new denial inflates the rate. HFMA's definitions count only the first denial made for a given claim (HFMA).

The same data also shows where money is slipping away without a denial at all. We walk through that wider review in how to find revenue leakage.

Which denials should you appeal, and what should you work first?

Appeals are worth the effort when the denial is wrong. Premier found that more than half (54.3%) of denials by private payers were ultimately overturned and the claims paid (Premier). Not every denial can be won, though. The Change Healthcare index found that 24% of potentially avoidable denials were not recoverable (HFMA), which is one more reason prevention comes first.

Sort each denial into one of three paths:

• Correct and resubmit, for errors such as a wrong member ID or a missing modifier.
• Appeal with evidence, for medical necessity or level-of-care denials that the chart supports.
• Write off and trace, for denials that cannot be recovered, so the cause is fixed for the next claim.

Track appeal deadlines by payer. For Medicare, a first-level appeal (redetermination) must be filed within 120 days of receipt of the initial claim determination (CMS); commercial windows vary by contract. Build an appeal template for each top denial category, and track overturn rates by payer and category so you know which appeals pay off.

For A/R follow-up, oldest-first is rarely the best order. Rank the worklist by dollar value, days left before a filing or appeal deadline, and how likely the claim is to be paid, then work high-value, recoverable claims near their deadline first. Batch low-dollar items with the same cause so one fix clears many of them.

Where does AI help, and where doesn't it?

AI is useful where the volume is too high for people to see patterns. It can score each claim for denial risk before submission, using payer rules and your own denial history. It can group thousands of remittances by root cause, rank A/R worklists, and draft appeal letters from the denial reason and the chart. Only 14% of Experian's respondents were using AI; of those, 69% said it reduced denials or increased the success of resubmissions (Experian).

It does not fix a registration process that keeps entering bad data, and it cannot write the documentation that supports medical necessity. Its output also needs people: Experian found that 90% of claim denials are reworked with at least some human review before resubmission (Experian Health). The best results come when the AI flags the risk and suggests a fix, and the department that owns the cause makes the change.

Revenue Intelligence, a module of OptraHEALTH's HealthFAX AI platform, is AI denial management software that scores claims for denial risk before submission, applies payer-specific rules, traces denials to their root cause, builds A/R follow-up worklists and automates appeals. One multi-site provider organization cut preventable denials by 35%.

How do you know it is working?

Pick a small set of measures, set a baseline before you change anything, and report them monthly:

• Initial denial rate. HFMA's task force defines it as initial denials divided by claims submitted, measured by count and by dollars (HFMA).
• Clean claim rate. The share of claims that pass your edits and go out without manual correction. Definitions vary between organizations, so write yours down and keep it fixed.
• Denials by category and department. The preventable categories should fall first, starting with the ones you fixed.
• Overturn rate and time to appeal, by payer and category.
• Denial write-offs as a share of net patient service revenue, which HFMA defines as MAP Key AR-6 (HFMA).

Where to start this month

Pick the denial categories that are both frequent and preventable, usually eligibility, registration and authorization. Trace them to their source, fix them there, and measure the change. Then move to coding and documentation, and make sure every recoverable denial is appealed on time. Denials won't disappear, but the ones that keep coming back will.