Nearly all medical eligibility checks are automated. ABA benefit design defeats the standard, which is why software built for other specialties breaks here.
Key Takeaways
- The electronic eligibility transaction exists and does not settle an ABA benefit: The federally adopted 270 and 271 pair returns whether coverage is active and what the patient owes. It does not reliably return hour limits, supervision requirements, concurrent billing rules, or the code-level exclusions that decide whether an ABA authorization is worth pursuing.
- The gap is ABA benefit design, not technology: Roughly two-thirds of covered workers are in self-funded plans, which are governed by federal law and are not bound by state autism mandates. Benefit terms are customized employer by employer, in documents a payer’s eligibility system was never built to expose.
- The contrast with the rest of medicine is stark: Ninety-six percent of medical eligibility verifications were fully electronic in 2023, according to the CAQH Index. In ABA, a complete verification for a new client commonly runs 20 to 30 minutes and still includes a call.
- This is what breaks general healthcare software on contact with ABA: Vendors arriving from other specialties assume an electronic path that resolves the question. Operators should ask how a product handles a commercial plan that returns nothing useful, because that is the ordinary case rather than the edge case.
Jeffrey Morelli was answering a question about how to pressure-test a software vendor when he reached for the example that explains a decade of failed entries into the ABA market. Morelli is Chief Executive Officer and co-founder of Silna, which builds front-end revenue cycle tools for behavioral health, and he was one of five vendors on a panel at ABA CARES in Boston last month about the cost of bad technology. Too many companies, he said, solve a problem for general healthcare, decide ABA looks interesting, and try to port the solution over.
Then he named the specific place it snaps. “Insurance verifications outside of ABA are very easy to solve,” he said. “You can just wrap payers and get electronic EDI 271 responses. In ABA, unless it’s Medicaid or an MCO, you have to call 100 percent of the time.”
His advice to buyers followed from it: pressure test the vendor, and make sure the team on the other side of the table knows more about the process than you do.
The claim is a vendor’s, made about the market he sells into, and the round number invites skepticism. The underlying mechanism holds up, and it is more interesting than the number. The electronic path Morelli describes is not merely available in ABA. It is federally mandated, widely implemented, and used constantly. It just does not answer the question.
What the 270 and 271 Eligibility Verification Transactions Actually Return
The transaction Morelli named is one half of a pair. Under HIPAA, the government adopted a standard for asking a health plan about a patient’s coverage (the 270) and for the plan’s answer (the 271), both maintained by the standards body ASC X12. A set of federally mandated operating rules published by CAQH CORE sits on top of them, requiring real-time responses, the return of patient financial responsibility, defined error reporting, and system uptime of 86 percent.
The result is one of the genuine success stories in healthcare administration. By CAQH’s count, 96 percent of medical eligibility and benefit verifications were fully electronic in 2023, a figure the industry treats as saturation. The pipe carries millions of transactions a day. It has cost the industry billions to build and is close to universal.
What the pipe carries is coverage status and cost-sharing: is this member active, what is the deductible, what has been met, what is the copay for a given service type. That is sufficient for most of medicine because most of medicine bills a discrete encounter against a benefit that is either open or closed.
Why an ABA Benefit Does Not Fit Inside an Eligibility Verification Response
ABA is not that. Whether a plan covers applied behavior analysis is the beginning of the inquiry, not the end of it. The operative facts sit downstream: whether the plan requires an autism diagnosis from a specified evaluator type and how recent it must be, whether an annual or weekly hour cap applies, whether 97155 supervision is billable concurrently with 97153 direct treatment, whether a Registered Behavior Technician can deliver services under the plan’s credentialing rules, whether telehealth-delivered parent training is payable, whether an age limit applies, and whether a network adequacy gap makes a single case agreement realistic. None of those are hypothetical constraints. Hour caps in particular are ordinary: Virginia caps ABA at 20 hours a week, and age limits produce the services cliff families hit when a child ages out.
Practitioners who do this work every day describe the split in exactly those terms. Practice management platforms do support the electronic check, and the check is useful for confirming active coverage, deductible position and cost-sharing. Everything past that starts with a phone call to the payer, a portal login where a portal exists, and a form filled in by hand.
The structural reason is that the answer often does not live at the payer. Under the Employee Retirement Income Security Act, self-funded employer plans are regulated federally and are not subject to state autism insurance mandates, which means the employer designs the benefit within federal limits. That is not a marginal segment. In the Kaiser Family Foundation’s 2025 Employer Health Benefits Survey, 67 percent of covered workers were enrolled in self-funded arrangements, including 80 percent of those at large firms. The insurer whose name is on the card is frequently administering someone else’s plan document, and the terms that decide an ABA case sit in that document rather than in the eligibility system.
So the phone call is not evidence that ABA missed the automation wave. It is what happens when a standardized transaction meets a benefit that was never standardized. The staff member on hold is not recovering something the system dropped. She is retrieving something the system was never designed to carry.
Why Medicaid and Managed Care ABA Verification Is the Exception
Morelli’s carve-out is the part of his claim that proves the mechanism. Medicaid and managed care organizations, he said, are the cases where the electronic path works, and the reason is not that state programs have solved a technical problem the commercial market has not.
It is that in Medicaid, the benefit is defined by published policy rather than by an employer. State programs issue fee schedules, coverage bulletins, and companion guides specifying exactly how their eligibility responses behave, and the ABA benefit itself is set out in state plan and waiver documents rather than negotiated plan by plan. Coverage for children flows through the Early and Periodic Screening, Diagnostic and Treatment requirement, which obliges states to cover medically necessary services for enrollees under 21. There is variation across states, and Acuity has documented plenty of it, including an Amerigroup rate cut in Georgia where silence counts as consent. But the variation is published, stable, and machine-readable in a way that several hundred separately designed employer plan documents are not, and the standardization now moving through state Medicaid systems will widen that gap rather than close it.
That distinction has a practical consequence for operators weighing payer mix. A Medicaid-heavy book carries rate risk, policy risk, and audit exposure. A commercial book carries a permanent administrative burden the electronic path was not built to lift. Neither is free, and the trade-off is visible in the intake queue.
What ABA Software Buyers Should Ask About Insurance Verification
The buying implication is why Morelli raised it at all. A vendor that solved eligibility verification for orthopedics or primary care built a product around an assumption that holds in those settings and fails in this one. The failure mode is not that the electronic call breaks. It is that the electronic call succeeds and returns nothing useful, and the workflow assumes it did.
The question to put to a vendor is therefore not whether it supports electronic eligibility checks. Every vendor supports electronic eligibility checks. It is what the product does when the response comes back thin: whether it routes the case to a work queue, whether it tracks which payers reliably return ABA-specific detail and which never do, whether it stores what the last call established so the next intake coordinator is not starting over, and what portion of the vendor’s existing ABA customers are commercial rather than Medicaid. Joe Burst, Director of ABA at Vivantium, made the adjacent point on the same panel about multi-state operations, where the compliance surface multiplies with every jurisdiction added. Verification behaves the same way, and the tooling rarely accounts for it. The same discipline is visible in how revenue cycle vendors sequence process before automation.
The burden also lands somewhere other than the finance line. Shridhevi Veerappan, Clinical Solutions Manager at RMS ABA and a Board Certified Behavior Analyst, put the clinician’s version of it on the same panel, describing what happens when documentation, verification, and billing are treated as separate problems handled by separate teams: cases where a BCBA writes clinically appropriate goals against a plan that does not cover them, and where an intake team confirms coverage without confirming the constraints the clinical team will run into. The verification failure and the documentation failure are the same failure, priced twice.
There is a downstream cost operators tend to book in the wrong place. Thin verification produces authorizations built on assumptions, and assumptions produce denials and recoupments later. Federal and state auditors have already spent the past year reading ABA files closely. Office of Inspector General audits of Medicaid ABA claims flagged notes too vague to substantiate active treatment, notes signed before the scheduled session ended, and codes billed that did not match the activity described, and Massachusetts providers are contesting a $19 million MassHealth recoupment built on supervision ratios. A benefit misunderstood at intake becomes a claim defended eighteen months later by someone who was not on the call. At the extreme, the administrative side of the payer relationship is what closes companies: the Indianapolis provider that shut down this year cited payment holds, a network termination, and credentialing backlogs in its letter to families.
None of this argues that the burden is fixed in place. The work is automatable in the sense that most repetitive telephone work is now automatable, and vendors including Morelli’s are selling against exactly that gap. What the electronic transaction cannot do, a well-scoped agent or a disciplined workflow can approximate, provided the buyer understands what it is approximating. The mistake to avoid is the one Morelli was flagging: buying a general healthcare product that assumes the electronic path answered the question, and discovering after go-live that in ABA it never does.
