How to Audit Your Medical Billing Company: 7 Questions and the Reports That Answer Them

Most practice administrators know something is wrong with their billing long before they can prove it. These seven questions give you the proof.

I ran a four-location occupational medicine practice in Hawaii for years before I started A-Z Medical Billing. During that time I hired three different billing companies, and each one gave me the same monthly summary: total charges, total payments, a handful of aging buckets. None of them told me what I actually needed to know, which was where the money was disappearing.

When I finally pulled the raw data myself, I found $1.1 million in claims that had been billed to payers and never followed up on. Not denied, not rejected. Just sent into the void with no carrier response, sitting in electronic limbo while my billing company reported a "97% submission rate" on their dashboard.

That number was technically accurate and completely useless.

My billing companies were not dishonest. I was asking the wrong questions. "How are we doing?" should have been "show me the report that proves it."

If you run a multi-location practice and you suspect your billing vendor is underperforming, these seven questions will tell you. Each one names the specific report you should request and describes what a bad answer looks like.

A good billing company will produce every one of these within 48 hours. A company that cannot produce them is not tracking the data, which means they are not doing the work.

1 What is our denial rate by payer, broken out by denial reason code?

Every billing company will give you an overall denial rate. Very few will break it out by payer and by CARC code without being asked. The overall number hides the one payer that is generating 60% of your denials, the one reason code that keeps repeating, and the one location where the front desk is collecting the wrong insurance information.

Report to Request
Denial rate by payer for the trailing 12 months, with the top five denial reason codes per payer ranked by dollar volume. Ask for it as a spreadsheet, not a PDF summary. You need to be able to sort and filter.
What a Bad Answer Looks Like
A single denial percentage with no payer breakdown. Or a report that lists denial codes by count instead of by dollar amount. A practice can have 200 CO-45 denials (contractual adjustments, generally expected) and 12 CO-197 denials (missing prior authorization on high-dollar imaging studies), and the CO-197 denials can cost ten times more. Count-based reporting buries the expensive problems. For a full reference on denial codes and their fixes, see the Denial Code Reference.

2 What percentage of our A/R is over 90 days, by location?

Accounts receivable aging is the most common metric in billing, and the most commonly reported in a way that hides problems. A group-level A/R aging summary is meaningless for a multi-location practice.

One location can have 8% of A/R over 90 days while another has 34%. Averaged together, the number looks acceptable. Separated, one site is hemorrhaging.

Report to Request
A/R aging by location with dollar amounts in each bucket (0-30, 31-60, 61-90, 91-120, 120+). The percentage of total A/R over 90 days should be calculated per site. Ask for the trend over the last six months so you can see whether the oldest bucket is growing.
What a Bad Answer Looks Like
A single aging summary for the entire practice. Or a report that shows percentages without dollar amounts. A site with $40,000 in total A/R and 30% over 90 days is a different problem than a site with $800,000 and 12% over 90 days. You need both numbers to know which one to fix first.

3 How many claims have been billed with no carrier response after 30 days?

This is the question that cost me $1.1 million before I learned to ask it. A claim can be submitted, acknowledged by the clearinghouse, and then never adjudicated by the payer. The clearinghouse acknowledgment looks like success in most billing dashboards, but it is not.

The claim is sitting in a queue somewhere, and unless someone is actively tracking the gap between submission and adjudication, it will age past timely filing and become uncollectable.

When I audited my own practice across 42,020 documents, I found that the claims with no carrier response were concentrated in three payers and two locations. The pattern was invisible in the summary reports my billing company was sending me.

Report to Request
A list of all claims submitted more than 30 days ago with no ERA (electronic remittance advice) or EOB on file. Grouped by payer and by rendering location. Include the date of service, date billed, CPT code, and billed amount. This report should update weekly.
What a Bad Answer Looks Like
"We don't track that separately." Or a report that only shows claims with a formal denial on file. The most dangerous claims are the ones that never generated any payer response at all. If your billing company cannot isolate them, they are not following up on them.

4 Can I see the write-off log with the reason for each adjustment?

Write-offs are where billing companies hide their failures. A denied claim that was never appealed gets written off. A claim that aged past timely filing gets written off. One that was coded wrong and rejected three times gets written off.

In each case, the write-off makes the problem disappear from the A/R report. The revenue is gone, and unless you review the log, you will never know why.

Report to Request
Complete write-off and adjustment log for the trailing 90 days. Each entry should include the patient account number (not the name), payer, CPT code, billed amount, written-off amount, and a free-text reason field explaining why the adjustment was made. Contractual adjustments (CO-45, CO-253) are expected. Everything else should have a documented reason.
What a Bad Answer Looks Like
A log with no reason codes, or reason codes only (no explanation). "Timely filing" as a reason on more than a handful of claims means the billing company is letting claims age out. "Patient responsibility" as a blanket category with no supporting EOB means balances are being written off without verifying insurance adjudication.

5 What is our clean claim rate at first submission?

The clean claim rate measures how many claims are accepted and paid on the first submission without edits, resubmissions, or corrections. It is the single best indicator of coding accuracy and front-end data quality. A rate below 90% means that more than one in ten claims is being reworked, and each rework adds 15 to 45 days to the payment cycle.

For specialty practices, the clean claim rate is especially revealing. Nuclear cardiology claims routinely involve five or more line items per study: the imaging code, the stress test, the pharmacologic agent, the radiopharmaceutical, and potentially the quantitative flow add-on (78434). A modifier error on any single line item can reject the entire claim.

Appending modifier 26 to 93015, for example, triggers an immediate CO-4 denial on the stress component. Specialty coding errors drag the clean claim rate down faster than primary care volume can pull it up.

Report to Request
First-pass clean claim rate by month for the trailing 12 months, measured as claims paid without rejection, denial, or resubmission divided by total claims submitted. Break it out by provider and by location if your PM system supports it.
What a Bad Answer Looks Like
A rate reported as "submissions accepted by the clearinghouse." Clearinghouse acceptance means the claim passed basic formatting checks (valid NPI, valid CPT code structure). It does not mean the payer accepted the claim. A 99% clearinghouse acceptance rate can coexist with an 80% clean claim rate at the payer level.

6 What is the average number of days from encounter to claim submission?

Lag time between the date of service and the date the claim is submitted tells you how quickly your billing operation is processing encounters. Every day of delay pushes the payment further out and increases the risk of timely filing denials. For payers with 90-day filing limits, a practice that takes 14 days to submit already loses 15% of the filing window before the claim even reaches the payer.

Authorization-dependent specialties feel this more acutely. A cardiac PET study that requires eviCore prior auth for Aetna or Cigna can sit for days waiting on documentation before the auth is submitted. If the billing company does not track submission lag separately from auth lag, the slow step stays invisible.

Report to Request
Average days from date of service to claim submission date, by month, for the trailing six months. Break it out by location. Flag any claims submitted more than seven business days after the encounter, with the reason for the delay.
What a Bad Answer Looks Like
An average that includes claims held for legitimate reasons (pending operative notes, pending auth) without separating them from claims that were simply sitting in a queue. The meaningful number is how long a clean, complete encounter takes to reach the payer. If that number is over five business days, the billing operation has a throughput problem.

7 Who physically calls the payer on a claim that has been outstanding for 45 days?

This is not a report question. It is a staffing question, and it reveals more about your billing company's operation than any dashboard.

Automated claim status checks (276/277 transactions) verify whether a claim was received and whether it is in process. They do not resolve denials, dispute underpayments, or chase claims that disappeared. At some point, a human being has to pick up the phone, call the payer's provider services line, get a reference number, and document what the rep said.

That work is where most of the revenue recovery happens, and it is the first thing a billing company cuts when they are trying to protect their margins.

Report to Request
A call log or follow-up activity log for all claims over 45 days old, showing the date of each outbound call, the payer, the representative's name or ID, the reference number, and the outcome. This is not the same as the notes field in your PM system. You are looking for evidence that someone is making calls, not just updating statuses.
What a Bad Answer Looks Like
"We use automated eligibility and claim status tools." Automation handles verification. It does not handle recovery. If your billing company cannot show you a call log with payer reference numbers, the follow-up is not happening. The $1.1 million I found with no carrier response at my own practice sat there because nobody was calling.

Need Help Running This Audit?

We build the per-location reports that surface the problems most billing companies are not tracking. The audit is free for multi-location groups.

Request a Free Billing Audit →

A billing company that produces all seven of these reports within 48 hours is doing the work. Keep them. A billing company that cannot produce them, or that pushes back on the request, is telling you something important about how they operate. Listen to it.

These are the same questions I ask when a multi-location group brings us in to evaluate their current billing. The reports are not proprietary, and every PM system can generate them. The only question is whether anyone is looking at them.

Dr. Zain Vally

Founder of A-Z Medical Billing & Consulting. Previously operated Vally Medical Group, a four-location occupational medicine practice across Hawaii, where he discovered $1.1M in unbilled and unworked claims by auditing his own billing vendors. He built A-Z to give multi-location groups the reporting and follow-up their current billing companies are not providing.

Frequently Asked Questions

How often should I audit my medical billing company?

Request the seven reports above quarterly at minimum. Denial rate by payer and A/R aging by location should be reviewed monthly. Claims with no carrier response should be tracked weekly, because those claims are aging toward timely filing deadlines in real time.

What denial rate should I expect from a good billing company?

A well-run billing operation for a multi-specialty or cardiology practice should maintain a denial rate below 5% by dollar volume. Above 8% signals a systematic coding, authorization, or data-quality problem. The rate varies by specialty: primary care typically runs 3% to 5%, while nuclear cardiology and interventional procedures run higher due to prior auth and modifier complexity.

What is a good clean claim rate?

Industry benchmarks target 95% or higher for first-pass clean claims measured at the payer level, not at the clearinghouse level. Below 90% indicates that more than one in ten claims requires rework before payment, adding 15 to 45 days per claim to the payment cycle.

Can my billing company refuse to provide these reports?

Your billing contract should guarantee access to detailed reporting. If the contract does not specify reporting deliverables, that is a negotiation point for renewal. A billing company that cannot produce payer-level denial rates or location-level A/R aging is likely not tracking those metrics internally, which means they are not managing to them.

What is the most common sign that a billing company is underperforming?

A growing A/R over 90 days bucket paired with a stable or increasing write-off volume. This combination means claims are aging out and being written off rather than worked. The second most common sign is a high clearinghouse acceptance rate reported alongside declining collections, which means claims are being submitted but not followed up on after payer rejection.

Should I audit the billing company myself or hire someone?

Start with the seven reports in this guide. If the billing company produces them and the numbers look reasonable, you can manage the oversight yourself with quarterly reviews. If the reports reveal problems or the company cannot produce them, bring in an outside billing consultant who can pull the raw data from your PM system independently.

How long should it take a billing company to produce these reports?

A billing company that is actively tracking these metrics should be able to produce all seven reports within 48 hours. Reports that take two weeks to compile are being built manually from scratch, which means the data is not part of the company's routine workflow.