Common problem
How to tell if your billing company is actually working your denials
Check three numbers and request six reports. If your net collection rate is under 95 percent, more than 15 percent of your AR is past 90 days, or your denial rate sits above 10 percent, something is not being worked. Then ask for a denial breakdown by reason code. A company managing your revenue cycle produces it the same day. A company that only submits claims will not. You can do all of this in an afternoon.
Start with three numbers
Under 95% net collection rate
Benchmark: MGMA. Below 95 percent means collectible money is not being collected.
Over 15% of AR past 90 days
Benchmark: MGMA 12–15 percent. Above this, claims are ageing faster than they are being worked.
Over 10% denial rate
Benchmark: HFMA targets under 5 percent. Kodiak Solutions recorded an industry initial denial rate of 11.81 percent in 2024.
If any of these sits outside range and has not improved in two quarters, keep reading. If all three look healthy, your billing company is probably doing its job.
What underperformance looks like in the data
Activity is not the same as results. A billing company can submit every claim on time and still leave money on the table. The real work happens after a denial arrives. These numbers show whether that work is happening.
| Metric | Healthy | Concern | Source |
|---|---|---|---|
| Net collection rate | 95% and above | Below 90% | MGMA |
| Days in AR | 30–40 | Over 50 | HFMA MAP Keys |
| AR over 90 days | Under 15% | Over 25% | MGMA |
| First-pass clean claim rate | 95% and above | Below 90% | HFMA MAP Keys |
| Denial rate | Under 5% | Over 10% | HFMA |
| Denial write-offs | Under 1.5% of net revenue | Above 3% | HFMA MAP Key AR-6 |
| Appeal overturn rate | 80%+ on appealed claims | Under 50% | HFMA |
What the numbers hide
Three failures rarely show up cleanly on a standard report.
Adjustments used as a hiding place.
Contractual adjustments are legitimate. An unworked denial can also be buried inside an adjustment code and disappear from the AR report without anyone appealing it.
Small balances left to die.
A $40 claim takes the same phone call as a $400 one. Nothing in an AR report tells you which claims were skipped because they were not worth the labour.
Timely filing.
Payer windows run from 90 days to 12 months. Once one closes, the claim is gone. This is the most expensive failure and the one least likely to appear unless you specifically ask for it.
Scale of the problem
Fifty to sixty-five percent of denied claims are never reworked. Forty-one percent of providers now run denial rates of 10 percent or higher (Experian 2025). Of 86 million in-network claims denied by ACA marketplace insurers in 2023, fewer than 1 percent were appealed. Roughly 44 percent of those that were appealed were overturned (KFF analysis of CMS data).
Most denials are not correct. They are simply unchallenged.
Free tool
Billing performance self-check
Enter five numbers from your own reports and see how they compare with published benchmarks. Runs in your browser. Sends nothing. Needs no patient data.
Informational only. Benchmarks vary by specialty, payer mix and practice size. These results do not prove anything about any particular billing company. Runs in your browser. Sends nothing. Needs no patient data.
The six reports to ask for
Request these six reports in writing. Note both what comes back and how long it takes. A company managing your revenue cycle produces all six from standard exports, usually the same day. A company that only submits claims will aggregate, delay, or explain why the system cannot do it. The response itself is the finding.
AR ageing, split by payer and by provider
Shows whether one payer or one provider is ageing disproportionately.
Good answerMatrix with 0–30, 31–60, 61–90, 91–120 and 120+ buckets, split both ways.
EvasionA single blended figure.
Denials by CARC and RARC reason code
Shows the root cause of your denials and whether anyone is preventing them.
Good answerDenials grouped by reason code with volume and dollars, trended over several months.
EvasionA total denial count with no code detail.
Appeals filed and overturn rate
Shows whether denials are being fought or absorbed.
Good answerAppeals filed, overturned, dollars recovered, broken out by payer.
EvasionNo appeal tracking at all.
Adjustments and write-offs by reason code
Shows whether unworked claims are being hidden inside adjustments.
Good answerEvery adjustment mapped to a reason code, contractual adjustments separated from denial write-offs.
EvasionLump-sum totals with no breakdown.
This is the report that most often surprises people.
Claims by status, with last-touch date
Shows the claims nobody has looked at.
Good answerWorklist showing the date each claim was last actioned, flagging anything untouched beyond 30 or 45 days.
EvasionOnly submitted and paid statuses.
Timely filing exposure
Shows claims approaching a filing or appeal deadline while they can still be saved.
Good answerOpen claims sorted by days remaining against each payer’s window.
EvasionNo such report exists.
You do not need all six. If a billing company cannot produce the denial breakdown and the timely-filing report, the other four will not save it.
Free tool
Report request generator
Tick the reports you want, add your practice and vendor names, and this writes the email. Tone is deliberately neutral — you are asking for standard reporting, not making an accusation.
Reports to request
Response window
Your request email
Tone is deliberately neutral — you are asking for standard reporting, not making an accusation. No patient data. Runs client-side.
Does your billing company have a reason to skip your denials?
Sometimes yes. It depends on how they are paid. Every pricing model creates an incentive somewhere. We charge a percentage of collections, so we will start with the problem in our own model.
Percentage of collections (what we charge)
The alignment is real on large, straightforward claims — we only get paid when you do. The misalignment is real on small and difficult ones. A $40 claim at a 6 percent fee earns the biller $2.40. The phone call to overturn the denial costs more than that in labour. Aged claims needing several touches have the same problem. The rational move, if nobody is watching, is to work the easy claims and let the hard ones age out.
Flat monthly fee
Removes the small-claim bias. Introduces a different one: the vendor is paid whether or not your claims are collected.
Per claim
Rewards submitting claims, not collecting on them. There is no structural reward for follow-up after a denial.
That is the most common complaint practice owners make about percentage billing. It is a fair one.
The mitigation is not a promise. It is visibility. If you can see the status of every claim, including the ones sitting unworked, the incentive becomes checkable. Ask any billing company you are considering — including us — for that view before you sign.
Honest conclusion
No model is clean. Percentage aligns on the big dollars and misaligns on the small ones. Flat and per-claim remove that bias but break the link between the vendor being paid and you being paid. What matters more than the model is whether you can see the work. A vendor that charges a percentage and will not show you unworked claims has the worst combination available.
What happens to your AR if you switch?
The claims already in flight are where practices lose money. Somebody has to keep working them. Either the outgoing vendor continues for a defined wind-down (usually 60–90 days) or the incoming vendor absorbs the aged AR as a separate workstream. If neither is agreed in writing, the claims sit and age past their filing deadlines. That money does not come back.
Run the two vendors in parallel for 30–60 days. New claims move to the new vendor from the cutover date. Legacy claims keep being worked in the background with monthly reconciliation.
A wind-down period alone does not guarantee recovery. MultiCare Health System alleges in a complaint against Change Healthcare and Optum that roughly $1.6 million in claims were not submitted or processed in time, about half of which had already passed their submission windows. These are allegations in ongoing litigation, not findings. The mechanism is the same at any size.
Your data
Your billing data is your asset, but contracts do not always say so. Check what an export includes, in what format, how quickly, and whether anyone charges for it. Also read the assignment clause. RCM vendors are being consolidated, and the contract you signed can end up owned by a company you did not choose.
What to get in writing before you sign with anyone
This applies to us as much as to anyone else.
- 1. You own your data. Complete export in a usable format within a defined number of days, at no charge.
- 2. Every open claim has an owner. Who works pre-cutover claims, for how long, and how it is reported to you monthly.
- 3. Notice period and any termination fee, stated plainly.
- 4. Whether the contract auto-renews, and the assignment clause.
- 5. A business associate agreement covering everyone who touches protected health information.
If a vendor resists putting any of these in writing, that is information.
Free 48-hour AR audit
Get a free 48-hour AR audit
We look at your claim data and tell you what is not being worked, where the denials cluster, and what it is costing you. No obligation. If your numbers are healthy, we will tell you that too.
What is your specialty?
How many claims a month?
Which of the six reports could your current billing company not produce?
You told us
— · — claims/mo · All reports available
Common questions
What is a good net collection rate?+
Ninety-five percent or above. MGMA treats 96–97 percent as effectively collecting. Below 90 percent points to a structural problem. On a practice collecting $5 million, each percentage point is roughly $50,000 a year.
What is a normal number of days in AR?+
Thirty to forty days is the HFMA MAP Keys target. Over 50 days suggests claims are not being followed up systematically. Read it alongside your AR-over-90 figure — a reasonable average can hide a large pile of very old claims.
How much AR over 90 days is too much?+
MGMA benchmarks 12–15 percent. Above 25 percent is a clear signal. Direction matters as much as the number. A figure that has climbed for two quarters without explanation is worth more attention than a single high month.
Can I audit my billing company without switching?+
Yes. Requesting the six reports is a routine reporting request. A vendor doing the job will produce them without fuss. If the request itself causes friction, you have learned something without changing anything.
How do I fire my billing company?+
Check your notice period and any termination fee first. Agree in writing who works the claims already in flight and for how long. Do not terminate before your new arrangement is operational. Run both in parallel for 30–60 days. The risk is not the switch itself — it is the claims left in the gap.
Who owns my billing data?+
You should, but the contract decides it. Check what an export includes, in what format, how quickly it is provided, and whether there is a charge. Negotiate this before you sign, not when you want to leave.
What notice do I have to give?+
Commonly 30–90 days. Check for auto-renewal clauses and early termination fees. Quilven works month to month with 30 days written notice.
What percentage of collections is normal?+
Generally 4–9 percent, with most quotes to practices of three to twelve providers between 5 and 7 percent. A lower rate that covers only claim submission is not cheaper than a higher rate that covers the whole revenue cycle. It is a smaller service.
Will my collections drop when I switch?+
Usually a short dip in the first cycle while claims move across, then recovery as the new pipeline fills. The larger risk is legacy AR going unworked during the handover. Agree who owns those claims before anything else is signed.
How long does switching take?+
Typically 30–60 days to be fully operational, longer if credentialing or system access is involved. Plan for both vendors running in parallel rather than a hard cutover on a single date.