Most behavioral-health practice owners we've talked to know they lose money to insurance denials. Few have an exact number. Here is the one that surprises every owner who works it out for themselves:
A 5-clinician commercial behavioral-health practice loses on the order of $216,000 a yearto denied claims it never appeals.
Two-hundred-sixteen thousand dollars is not a small leak. It is bigger than most BH practices spend on rent. It is, in many cases, larger than the practice owner's take-home pay. So the right question isn't whether the number is real — the AMA's 2025 prior-auth analysis and KFF's 2024 Medicare Advantage data both anchor it within a fairly tight range. The right question is why it persists.
The math nobody runs
Five clinicians, each seeing roughly 25 sessions a week, 48 weeks a year, at roughly $130 average reimbursement for a 90834 in-network: about $780,000 in billed revenue. A 16–22% denial rate (industry-standard for BH; two to four times the rate for medical/surgical care) puts $125k–$170k of that into the “initially denied” bucket each year. Roughly half of appealed BH denials are overturned. So if appeals were free, the practice would recover ~$65k–$85k. Compound that across the share that gets billed off the bad note, the share that gets written off entirely, and a 5–10% downstream rate for newer denials introduced after partial recovery, and the annual revenue leak from unappealed denials lands at the $200k–$240k range. We call it $216k because that's where the median 5-clinician practice we've sat with actually lands.
Why nobody appeals
Not because anyone is lazy. Because the per-claim economics are upside down.
- An appeal takes an hour of someone's time — a biller at $55, an admin at $35, the owner-clinician at $150+ if they do it themselves.
- The recovery on a single 90834 is roughly $130, and only ~50% of appeals win.
- Expected value is ~$65 minus an hour of $55–$150 staff time — break-even to actively negative on the small claims. So the rational decision, claim by claim, is “don't appeal.”
That decision is the right one for a single claim, evaluated in isolation, and the wrong one for the practice. Across hundreds of claims per year, the negative-EV-per-claim math swings sharply positive — provided the per-claim friction comes down. Most practices never make it across that swing because the friction never comes down.
The behavioral-health–specific reasons it's worse here
Three structural reasons BH practices denial rates run 2–4× higher than the rest of medicine:
- Documentation gaps that are easy to make and hard to defend after the fact. A 90837 requires 53+ documented minutes; a 90834 requires 38–52. Payers routinely downcode 90837 to 90834 (~25% of dollars per claim) when the note doesn't pin the duration. Add-on codes (interactive complexity 90785, family therapy 90847) get denied when the intervention's rationale isn't in the note. None of this is a clinical failure — it's a documentation format problem.
- Prior-authorization complexity disproportionately falls on BH. A 2025 AMA study found behavioral-health services face ~3× the PA burden of medical/surgical services. Every PA is another point where a clean claim becomes a denial because the auth was missing, expired, or for the wrong CPT.
- Parity enforcement is structurally weak. Federal mental-health parity law (MHPAEA) is real on paper, intermittent in practice. Until CMS-0057-F starts forcing public denial-rate reporting in March 2026, payers have had no public-disclosure pressure to align BH denial rates with the rest of their book.
What changes the math
For the unappealed-revenue leak to actually close, three things have to happen at the same time:
The documentation has to arrive appeal-ready. Not a SOAP note that the clinician then edits to include the duration, the modality, and the medical-necessity language a payer's criteria call out. The note has to already include those things, with the transcript moments cited, the first time it's drafted.
The appeal letter has to be drafted automatically when the denial lands. Not someday-when-someone-has-time. The denial code arrives via the 835 remittance file; the system reads the CARC code, finds the payer's criterion for the underlying CPT, drafts a letter that cites that criterion plus the relevant moment in the transcript, and presents it to the clinician for one-click sign + submit.
The practice owner can't be the bottleneck. If the workflow requires the owner-clinician to spend an hour pulling records for each appeal, nothing changes. The whole point is the friction goes away.
That's the system Trellova is building: ambient capture that produces an appeal-ready note, a denial-recovery layer that drafts the appeal the moment the 835 lands, and a billing model where the contingency on recovered revenue is small enough that the practice keeps the vast majority of what comes back. The point isn't the AI. The point is the per-claim friction goes to zero. Once that happens, the $216k leak is no longer a leak — it's recovered revenue.
If you run a 2–10 clinician behavioral-health practice and any of this maps to what you see, we'd like to talk.