Denial Management

Effective Claim Denial Management Workflow Design

Treating denials as real-time events rather than monthly reports recovers the most revenue.

Altair Health 4 September 2026

Denial management runs on a sequence of time-bound steps. Every gap between those steps is where money goes missing.

Most practices measure denials in aggregate, at month-end, after the appeal clocks have already started. By the time a spike shows up on a report, some of the denials behind it have aged past the point of correction.

The practices losing the most are rarely the ones with bad billers or a hard payer mix. They are the ones treating denial management as a monthly reporting exercise instead of an hourly one.

The scale is documented. Initial claim denials reached 11.81% of claims in 2024, per Kodiak Solutions data covering more than 2,100 hospitals and 300,000 physicians. On the marketplace side, HealthCare.gov insurers denied 19% of in-network claims in 2024, according to KFF.

Unworked denials do not announce themselves. That is what makes this expensive: it compounds without ever showing up as an incident.

What the 835 contains, and why it has to be read on arrival

The 835 is the payer's adjudication record. It states what was paid, what was adjusted, what was denied, and why.

Three things live inside it. Claim Adjustment Reason Codes, the CARCs, give the payer's official reason for reducing or denying payment. Remittance Advice Remark Codes, the RARCs, add what the CARC leaves out, and often name what the payer wants next: more documentation, a corrected code, a prior authorization number. Then the payment figures themselves, including contractual adjustments and the split between payer responsibility and patient responsibility.

None of it helps sitting unread. Post 835s by hand, or batch them for end-of-day review, and the filing and appeal clocks run before anyone has looked at the denial reason.

Automated ingestion changes the order. The file arrives, gets parsed, and every denial is classified before a person touches it. Practices posting ERAs manually find denials days or weeks later, once staff attention has moved to the next batch.

Ingestion speed is not one priority among several. It decides how much of the denied revenue is still recoverable, which makes it the first fix rather than the fifth.

Sorting denials into action lanes with CARC and RARC

Denials do not all need the same response. Working them as one undifferentiated queue is the most common mistake in this workflow.

Four lanes cover almost everything. Correctable and resubmittable denials, a wrong modifier, a transposed NPI, a missing field, get fixed and resent with no formal appeal. Appealable denials, medical necessity disputes and coverage determinations that contradict the payer's published policy, need a documented appeal. Payer error and contractual issues, underpayments and incorrect adjustments, usually need a reconsideration request rather than an appeal. Patient liability, meaning non-covered services and confirmed patient responsibility, moves to patient billing rather than staying in the denial queue.

The CARC usually tells you which lane. A CO-4 is a modifier problem. A CO-50 is medical necessity. They need different work, and confusing them wastes the speed that early classification is supposed to buy. Our denial code reference covers the common CARCs and what each one requires.

One complication: the same CARC does not mean the same thing at every payer. The fix that clears a denial at one insurer can do nothing at another.

A properly sorted queue has no orphans. A denial with no assigned action ages by default.

Correcting and resubmitting, where the recoverable money goes stale

Correctable denials should be the easiest revenue you have. The payer already told you what was wrong, in writing, with a code attached.

In practice this is where the most recoverable money is lost to delay.

The failure modes are familiar to anyone who has worked a billing desk. Denials sit in a worklist with no guaranteed turnaround. Corrections get batched instead of worked on arrival. The biller who touched the claim changes roles, and it sits unassigned. Nobody writes down the fix for a recurring denial at a specific payer, so the next biller solves it again from scratch.

Speed matters for two reasons. Timely filing runs anywhere from 90 days to a full year depending on the payer.

A deterministic fix should close the loop with no person in the path. A missing modifier, a transposed digit, a blank field: correct and resend the moment classification finishes. No worklist, no queue.

The practices recovering the most from this category removed the gap between classification and resubmission. They did not hire more people to work the queue faster.

Building an appeal that survives review

An appeal is a formal argument that the payer's own policy, or the patient's coverage terms, does not support the denial. It is a document built for scrutiny.

A complete package needs the original claim and the ERA showing the denial with its CARC and RARC. It needs clinical documentation that maps to the stated denial reason, which for a medical necessity denial means notes answering the payer's coverage criteria rather than general clinical narrative. It needs the payer's published policy, cited by version and date, showing where the denial departs from it. It needs a written letter naming the denial reason, the policy language, and the documents attached. It needs the prior authorization number if the service required one.

A thin appeal is easy to deny a second time: a generic letter, clinical detail that does not answer the criteria, a citation to a policy version that is no longer current.

The appeal evidence comes from prior authorization rather than claims, but the pattern is stark. Medicare Advantage insurers denied 4.1 million prior authorization requests in 2024, and only 11.5% of those prior authorization denials were appealed, with 80.7% of the appeals overturned, per KFF. KFF notes these are service determinations rather than payment decisions on care already given. Four in five challenged denials still failed to survive the challenge, and most were never challenged.

Commercial claims tell a similar story. A 2022 Premier survey of 516 hospitals found that 54.3% of private payer denials were eventually overturned, at an average of $43.84 per claim in appeal labor.

Payer-specific knowledge separates an appeal that wins from a form letter. Knowing what documentation that payer has accepted before, for that procedure, beats knowing what sounds reasonable.

Deadlines, and why sorting by dollar value loses money

Every step above runs inside a deadline, and the deadlines are not uniform.

Timely filing for resubmission, the initial appeal window, and any secondary appeal or external review each carry their own clock. Each payer sets its own lengths. Medicare generally allows a year from the date of service for filing, with appeal levels running on shorter separate clocks. Commercial windows vary widely. Medicaid varies by state, often shorter than commercial. Our payer guides hold the current filing and appeal windows by payer.

The operational requirement is simple. Every denial in the queue needs its deadline attached, visible, and sortable.

Sorting by dollar value, or by which biller owns the account, is the common default and the wrong one. It lets smaller claims expire while attention sits on the big ones. Every one of those expirations was avoidable.

A missed filing deadline is not an error you correct later. It is permanent.

Root-cause analysis, turning one denial into prevention

Working a denial recovers one claim. Understanding why it happened prevents the next several hundred that share the same conditions.

Ask one question of every denial: front-end failure, mid-cycle failure, or a payer policy change nobody caught?

Front-end failures are eligibility never verified before the visit, a missing or mismatched prior authorization, a registration error in demographics. Mid-cycle failures are claims submitted without a check against the payer's current policies, modifiers applied wrong, documentation left off. Payer-side changes are the hardest to catch, because the policy updates and the billing process does not register it before the next claim goes out.

The mix shifts, which is why last year's fixes stop working. Kodiak's 2024 data shows authorization-related denials falling 7.7% while medical necessity denials rose 5% and requests for more information rose 5.4%.

Pattern is what you are looking for. The same CARC, at the same payer, for the same procedure means a process problem upstream, not another round of claim-by-claim correction.

Group denials by reason code, payer, procedure, rendering provider, and submission date, so the pattern surfaces in days instead of in a month-end summary.

The findings matter less than where they travel next. They belong back in eligibility verification, in prior authorization, and in the scrubbing logic that checks claims before they go out.

Every worked denial teaches you how a payer behaves, which is not the same as what its published policy says. Let that knowledge live in one person's head and you relearn it from zero when they leave. Altair keeps it in the system instead, so the next claim to that payer already carries what the last one taught.

Scrubbing before submission

The cheapest denial is the one that never happens. No rework hours, no appeal labor, no deadline pressure.

Practices underinvest here relative to what it saves, mostly because prevention has no line item the way rework staffing does.

Scrubbing catches four categories before a claim reaches a payer. Eligibility mismatches, where coverage on file will not adjudicate on the date of service. Authorization gaps, where the auth on record covers a different procedure or date range. Payer policy conflicts, a missing modifier, a diagnosis that does not support the procedure, a wrong place-of-service code. And clearinghouse-level errors, formatting problems, missing required fields, NPI errors.

The word that matters is current. Payers revise policy on their own schedule, often without useful notice. A scrub checked against last quarter's policies misses exactly the denials this quarter's change is generating.

Front-end information errors are among the most common preventable causes. Experian Health's 2025 State of Claims survey found that 26% of providers trace at least one in ten denials to intake errors, and they are among the simplest to stop, with an eligibility check run before the patient is seen.

This is where root-cause analysis pays for itself. A recurring front-end pattern belongs in the scrubbing layer, not in a rework queue that solves it again every time.

The measure here is clean claim rate: the share accepted on first pass with no correction and no appeal.

Real-time visibility

A monthly denial report describes what already happened. Some of those correction windows are closing while the report sits on a desk.

Real-time visibility replaces the lag. Every claim tracked by status, from submitted through adjudicated, denied, appealed, and paid, with denials surfacing when the 835 arrives.

It also means the owner sees the same numbers the billing team sees, without requesting a report and waiting for someone to run it.

Response time changes. A denial that would have sat for two weeks gets classified and routed within hours. That shows up in days in A/R.

There is a governance point here too. A practice that can see every claim and every dollar as it moves does not wait for monthly reconciliation to know whether revenue is on track. See how Altair runs denial management for how this works end to end.

The four action lanes

Common questions