If you run billing for an ABA (applied behavior analysis, a structured treatment approach for autism) practice, you already know the denial rate feels higher than almost any other specialty. It is not your imagination. ABA billing has moving pieces that most other specialties do not deal with at all, and authorization units sit right at the center of it.
I have spent years untangling denial patterns for behavioral health practices, and the same three culprits show up again and again: authorization unit tracking, concurrent session rules, and credentialing lag. Let's walk through each one in plain terms, and then talk about where the real money hides.
Why ABA Denials Feel Different From Other Specialties
Most specialties bill a handful of codes per visit. ABA practices bill in 15-minute units, often multiple times a day, across multiple staff members, for the same client. A single child might have a BCBA (Board Certified Behavior Analyst, the clinician who designs the treatment plan), an RBT (Registered Behavior Technician, the staff member who delivers direct therapy), and a supervising clinician all touching the same authorization in the same week.
Every one of those touchpoints has to line up with what the insurance company approved in advance. When even one piece is off, the claim denies. This is part of why clean claim rate in ABA tends to run lower than in specialties with simpler visit structures. If you want a refresher on what that term actually measures, our clean claim rate guide breaks it down in plain English.
Authorization Units Are the First Place Claims Fall Apart
An authorization unit is the amount of therapy time an insurance company pre-approves for a specific date range, usually measured in 15-minute blocks. A typical authorization might approve, say, 80 units per month for direct therapy and 8 units per month for supervision. Once those units are used up, anything billed past that point denies automatically, no matter how medically necessary the session was.
The problem is that authorization units live in one system (the payer portal or a fax confirmation), while actual session delivery lives in a completely different system (your scheduling or EHR software). Nobody is watching both in real time. So a practice keeps scheduling sessions against an authorization that ran out two weeks ago, and every one of those claims denies in a batch once someone finally notices.
Common authorization unit failure points include:
- Units exhausted mid-month but sessions keep getting scheduled anyway
- Authorization date range mismatches, where the session date falls one day outside the approved window
- Wrong unit count entered at intake, so the system thinks there is more room than there actually is
- Renewal delays, where the old authorization expires before the new one is approved
Concurrent Session Rules Trip Up Even Careful Practices
Concurrent session rules govern how many billable hours can happen for the same client, or across a group of clients supervised by the same clinician, at the exact same time. Payers use these rules to prevent overlapping or duplicate billing.
Here is a simple example. Say a BCBA is supervising two RBTs who are each seeing a different client, but both sessions overlap by 30 minutes on the schedule. If the payer's rules cap how much overlapping supervision time can be billed, that overlap gets flagged and denied, even though the care was completely legitimate.
These overlaps are almost impossible to catch by eye on a busy scheduling grid, especially across multiple clinicians and multiple locations. That is exactly the kind of pattern-matching problem that software is good at and humans are not, which is part of why we wrote about how AI reduces medical claim denials for practices dealing with high-volume, high-complexity scheduling.
Credentialing Lag Creates Invisible Denials
Credentialing is the process of getting a provider formally approved to bill a specific insurance panel. Credentialing lag happens when a new RBT or BCBA starts seeing clients before their credentialing paperwork clears with every payer the practice takes.
ABA practices hire fast because demand is high, which means credentialing often trails behind actual hiring. A brand new RBT might start seeing three clients on day one, but if their enrollment with a specific payer is still pending, every claim tied to that provider denies until the effective date catches up. Worse, some of those sessions may never be billable at all, because the payer's effective date starts the day approval is granted, not the day the provider began working.
This is a quiet, slow leak. Nobody notices it until a batch of claims comes back denied weeks later, tied to a provider who has been seeing clients the whole time.
Where Leakage Hides in Unbilled Units
Leakage is the industry term for care that was actually delivered but never gets billed or gets billed for less than what was provided. In ABA, this shows up constantly around authorization units and session overlaps.
A common example: a session runs 15 minutes over the authorized unit count, so the front desk just does not bill the extra units rather than deal with the paperwork to request a retroactive authorization increase. Multiply that across dozens of clients and dozens of sessions a month, and the unbilled units add up to real revenue that simply evaporates. Nobody denies these claims, because nobody ever submits them.
This kind of quiet leakage is a close cousin to the direct cost of a denied claim, which we walked through in detail in what does a denied claim really cost your practice. The difference is that leakage does not even show up on a denial report. It just never shows up anywhere.
Fixing the Pattern: From Spreadsheets to AI Agents
Most ABA practices are still tracking authorization units, session overlaps, and credentialing status across separate spreadsheets, sticky notes, and someone's memory. That approach works fine until the practice grows past a handful of clients, and then it breaks down fast.
The AI agents we build at AutomatedRCM are designed to sit across your scheduling data, authorization records, and credentialing status at the same time, watching for the exact patterns described above: units running low before a session gets booked, overlapping concurrent sessions before they hit the schedule, and provider credentialing status before a claim goes out the door. You can see how this works in our live demos.
If you are not sure how much of this is already happening in your practice, the free Billing Health Check is a straightforward next step. It takes a look at where your denials and leakage are actually coming from, without any commitment.
Frequently Asked Questions
Why do ABA therapy claims get denied so often?
ABA claims deny frequently because of the sheer number of moving pieces involved, including authorization units, overlapping concurrent sessions, and provider credentialing status. Because ABA billing happens in small 15-minute units across multiple staff members for the same client, small mismatches between what was approved and what was delivered turn into denials more often than in specialties with simpler visit structures.
What are authorization units in ABA billing?
Authorization units are the specific amount of therapy time, usually measured in 15-minute blocks, that an insurance company pre-approves for a client over a set date range. Once those units run out or the date range expires, any additional billed units deny automatically, which is why tracking remaining units in real time is critical.
What is credentialing lag and how does it cause denials?
Credentialing lag happens when a new clinician, such as an RBT or BCBA, begins seeing clients before their approval with a specific insurance payer is finalized. Any sessions billed before the payer's official effective date typically deny, and some of that revenue may not be recoverable at all.
How does unbilled leakage happen in ABA practices?
Leakage happens when a practice delivers a service but never bills for it, often because a session ran past the authorized unit count and staff skip the extra paperwork needed to bill it correctly. This does not show up as a denial on any report, since the claim is never submitted, which makes it one of the hardest revenue problems to spot without dedicated tracking.