Insurance claim denials: a follow-up queue for clinics

Why denials cost a clinic money

A denied claim that nobody opens is simply unpaid. The American Hospital Association's 2025 Cost of Caring report estimates that hospitals spent nearly $18 billion in 2025 on overturning claim denials alone, out of a wider $43 billion spent trying to collect payments insurers owed for care already delivered. Those figures cover hospitals, not clinics. The mechanics are the same, though: someone has to find each denial, understand it and push it back through the right channel.

For physician practices, the benchmark is lower. MGMA's 2026 benchmark, as summarised by WeBill Health, puts the typical first-submission denial rate for medical practices at 7% to 8%, and says well-run practices can get under 5%. The American Academy of Family Physicians calls 5% to 10% typical and treats under 5% as the better goal. Those two sources sit close enough to use 5% to 10% as a planning range. Your own rate, measured from your own claims, is the figure that should drive your decisions.

Denials also tend to get worse when they wait. A 2019 analysis of Medicare Advantage claims, published in Health Affairs in 2024, found that 14.6% of physician claims were denied at first submission. About 52% of those denials were later overturned, which left a net loss of 7.0% of physician claims. The denied claims that get paid often do so only after rework, so the cost is the staff time and the delay, not only the write-off.

Here is a simple way to size your own exposure. Suppose a clinic sends 1,000 claims a month and sees a 7% first-pass denial rate. That is 70 denials before any appeals. If each one takes a biller about 20 minutes to read, look up and resubmit, the work comes to roughly 23 staff hours a month. The 20-minute figure is an assumption you should replace with your own timing. Multiply your hours by your fully loaded hourly staff cost, and you have a cost figure that belongs to your practice.

Experian's 2025 State of Claims survey, reported by Physicians Practice, found that 41% of providers now report denial rates above 10%, up from 30% three years earlier. The same survey base is spread across providers and hospitals, so treat it as a sign of trend rather than a benchmark for a single clinic.

The manual way and the automated way

Most clinics already do each of these steps. The difference is whether a person has to go looking for the work every day, or whether the work arrives in a queue that a person then decides on.

Step-by-step: build the denial queue

The steps below start with the cheapest change and end with the one that needs the most care. Do them in order. Each step should be working and checked before the next begins.

1. Export one month of denied and unpaid claims. Pull the list from your practice management system, or from the billing module you use, into a spreadsheet. Include the payer, the date of service, the claim number, the billed amount, the denial reason as written, and the date it was first denied. Count the items by reason. This list is your baseline, and it tells you where automation will help.
2. Set up electronic claim status checks. The federal standard for these checks is the X12 276 request and the 277 response. The Centers for Medicare and Medicaid Services describes these as the provider's inquiry and the health plan's answer. Ask your practice management system or clearinghouse whether it sends 276 requests in real time or in batches, and which payers it reaches. Some payers only answer through their own portal, so keep a manual list for them.
3. Read the payment and denial details. Electronic remittance files, usually in the 835 format, carry the payment and adjustment codes for each claim. Many clinics find that payers send paper letters or scanned explanations of benefits for some claims. For those, optical character recognition, such as the text recognition in Adobe Acrobat or a cloud service like Amazon Textract, can pull out the claim number, amount and reason text. Check the extracted text against the original on a sample each week before you trust it.
4. Map reason codes to your own categories. Payers use claim adjustment reason codes and remark codes that are hard for non-billing staff to read. Build a table that turns them into eight to twelve categories your team uses, such as eligibility or registration, missing authorisation, coding or modifier, medical necessity, timely filing, duplicate, and coordination of benefits. The Medical Group Management Association has noted that registration and eligibility issues account for nearly 27% of denials, so that category is often the first place to look.
5. Route each item into a queue. A workflow tool such as Zapier or Microsoft Power Automate can create a task for each denial in a shared spreadsheet, a task board or your practice system, with the owner, the category and the due date filled in. Send an email to the billing lead for any item above a dollar threshold you set. Keep the rules visible so staff can see why an item landed where it did.
6. Use a language model only for free-text notes. Some payer remarks are written as short sentences. A language model can group those notes into your categories and summarise them in one line, which saves reading time. It should label its output as a suggestion, and a biller should confirm the category before any action is taken. Do not let it write appeal letters that cite clinical facts without a staff member checking those facts against the chart.
7. Use WhatsApp only for patient balance reminders. If patients get a balance after insurance pays, a WhatsApp Business API message can remind them to pay or call. Keep the message free of claim details, diagnoses and procedure names. Confirm the setup with your compliance lead before any patient information goes into a messaging tool.
8. Hold a weekly review. Once a week, the billing lead opens the queue, checks the oldest items, samples ten or so closed items to confirm the outcome, and updates the category table when a payer changes its codes.

Sorting denials by reason

Sorting is where the queue earns its keep. A flat list of 70 denials looks like one job. Sorted by reason, it becomes four or five different jobs, each with its own owner. Missing or wrong registration data, for example, is usually fixed at the front desk and never needs an appeal. A missing prior authorisation is a different problem, and practices often see denials where the authorisation exists but the number, dates, units or rendering provider on the claim do not match what the payer approved. Those are corrections, not appeals.

Medical necessity and coding denials need a clinician or a certified coder. The queue can put them in front of the right person, but it cannot judge whether a service was necessary or whether the documentation supports the code. Sort those into their own lane from the start, so they do not sit behind simple corrections and age out.

Put a deadline column next to every category. Filing limits and appeal windows vary by payer and by contract, so the category table should record the deadline your payer contract gives for each type of denial. Check those deadlines in the payer contract or the payer's own provider manual, not in a general guide.

What breaks and how to prevent it

The most common failure is a mapping table that falls out of date. Payers change codes and remarks, and a queue that keeps labelling new codes as unknown will bury the work. Prevent this by keeping the table in one place with a date on each change, and by sending any unknown code to a person on the same day it appears.

The second failure is a status check that returns nothing useful. A claim number that does not match what the payer holds, or a request sent before the claim has been processed, can come back as not found. Treat a not-found result as a flag for review, not as proof that the claim was never received. Check the claim number on the original submission first.

The third failure is duplicated work. If the status check, the remittance import and a manual check all create tasks for the same claim, staff will stop trusting the queue. Give each claim one open task at a time, and close the task only when someone records the outcome.

The fourth failure is a wrong authorisation or appeal sent out in good faith. Keep a human sign-off on every outgoing appeal, correction or payer message. Ask any vendor that will handle patient information whether it will sign a business associate agreement, and have your compliance lead review the answer before you start.

Finally, watch the numbers you report. A falling denial count can mean the queue is working, or it can mean denials are being closed without resolution. Sample closed items every week to tell the difference.

When not to automate

Automation cannot decide whether to appeal, whether a service was medically necessary, or whether a chart supports a code. Those decisions stay with your billing staff, providers and coders, and no software setting should override them. The queue only makes sure they see the right item at the right time.

Do not automate if your denial volume is small. If you see a handful of denials a month, a spreadsheet, a weekly hour and a clear owner will usually do more good than a set of connected tools that someone must maintain. Likewise, if your practice management data is incomplete, with missing payer IDs, wrong dates of service or unlinked claims, fix that first. Automation will copy the errors into the queue faster.

Do not connect a payer portal that your staff use only occasionally and that changes its layout often. In that case, a short manual check each week may cost less than the repair work.

What it typically takes

The main cost is time, not software. Expect a first month of logging and sorting by hand, so you have a baseline. Then plan several working sessions with the billing lead to agree on the reason categories, the owners and the deadlines. Testing the status checks and the remittance reading against a sample of real claims takes longer than most clinics expect, because each payer has its own quirks. How long this takes depends on your practice management system, the number of payers and how clean your claim data is. The free automation audit measures those things for your practice before any build begins.

Plan for staff to keep working the queue during the change, and do not switch off the manual process until the queue has matched it for several weeks.

The federal standard for claim status requests and responses is set under HIPAA rules, and CMS publishes a plain summary of how the 276 and 277 transactions work. This explains the rule in general terms; it is not legal or tax advice for your case.