Insurance eligibility verification for therapy practices

Eligibility verification confirms a client's coverage, benefits, and cost share before the appointment, so the practice is not billing blind. Behavioral health makes it harder than most specialties: separate carve-outs, session limits, and authorization rules. Most groups verify manually a few days ahead. The workable alternative is to automatically check against a schedule and escalate only the exceptions.

TL;DR

• Verification is three different questions wearing one name: is the plan active, what does it cover, and is approval required. Practices that conflate them result in preventable denials.
• Behavioral health is harder than most specialties: carve-outs, session limits, authorization rules, and telehealth variation, all moving independently.
• Five fields captured on every check cover what the front desk, the clinician and the biller each need.
• The inversion that scales: checks run automatically against the schedule, and only the exceptions reach a human, as a worklist.
• Keragon connects the EHR, clearinghouse, and verification tools without code, with a BAA on every paid plan. It is never the EHR and never the clearinghouse; it moves the data between them.

What is insurance eligibility verification?

Key definition

Insurance eligibility verification is the process of confirming, before care is delivered, that a patient's coverage is active and what it will pay. It bundles three distinct questions. Eligibility: is the plan active on the date of service? Benefits: what does it cover for this care, at what copay, coinsurance and deductible, with what session limits? Authorization: does the payer require approval before this service is delivered? Practices routinely conflate the three, and the distinction is the first real value of getting the process right, because each question fails differently and each failure produces a different denial.

A note on register before going further: this article serves both therapy practices, where the person in the chair is a client, and psychiatry and billing contexts, where payer and billing language refers to a patient. Both appear below deliberately, matched to their context.

Why behavioral health verification is harder than other specialties

Five structural reasons, all general patterns rather than any payer's current policy, because those change too often to print:

Carve-outs. A patient's behavioral health benefit is often administered by a different entity than the medical plan on the front of the card. The medical plan's portal can show active coverage, while the behavioral benefit lives elsewhere entirely, with its own network, its own rules, and its own phone number. Verifying against the wrong administrator produces a confident answer to the wrong question, which is worse than no answer.

Session limits. Behavioral health benefits frequently cap visits per year or track them against review thresholds. A plan verified as active in January can run out of covered sessions in September, meaning verification for ongoing therapy is not a one-time event but a recurring check against a depleting balance.

Authorization rules. Whether approval is required can vary by plan, service code, and level of care, and it moves independently of eligibility. Higher levels of care carry greater authorization weight, and the requirement can be attached mid-treatment when the level of care changes.

Telehealth variation. Coverage for virtual sessions and the cost share attached to them still vary by plan and do not always match the in-person benefit, which matters enormously for practices delivering care both ways.

EAP sessions. Employee assistance program sessions often sit entirely outside the insurance benefit, with their own count, their own authorization, and their own billing path. A client transitioning from EAP sessions to insurance-covered therapy crosses a boundary that the verification process must see.

The five things to capture on every check

A verification that answers the phone-call question ("are they covered?") but captures nothing durable has to be repeated by the next person who needs it. Five fields, captured every time, make the check an asset instead of a moment:

When to check: a working cadence

The question "how often should we verify?" has a structural answer: different client situations deplete or change on different clocks, so the cadence follows the situation rather than a single rule:

Manually, this cadence is aspirational; automated against the schedule, it is just configuration, which is the honest reason most groups only achieve it after automating.

What the manual process actually costs

The manual version is a scramble with a schedule attached: a staff member works the appointment list a few days out, portal by portal and payer by payer, retyping the same demographic details into each one, transcribing results into wherever results go, and phoning the payers whose portals disagree with themselves. It is honest, necessary work, and it scales exactly as badly as it sounds.

At group scale, the arithmetic turns severe. A multi-state telepsychiatry group described an insurance team manually checking coverage for every patient three to five days ahead of roughly 5,000 appointments a month, and things inevitably getting missed. That is the shape of the problem everywhere: not that any single check is hard, but that thousands of them, under deadline, by hand, guarantee a miss rate, and every missed check is a candidate denial or an awkward cost conversation after the session instead of before it.

Run the arithmetic transparently, with no invented statistics required: 5,000 appointments a month mean 5,000 checks a month, and even a few minutes of portal work per manual check means hundreds of staff hours monthly spent confirming what is usually true, to catch the minority of cases where it is not. That ratio is the tell. Work where the overwhelming majority of outcomes are "fine, as expected" is exactly the work to hand to a machine, precisely so the minority that is not fine gets human attention while there is still time to act on it.

The misses cluster where the structure is hardest: the carve-out that was never identified, the session limit that nobody was tracking, the authorization that lapsed when the plan year rolled over. Which is to say, the misses cluster exactly where behavioral health differs from the specialties the portals were designed around.

How to automate the checks and escalate only the exceptions

The fix is not checking faster; it is inverting who touches what. In the manual model, a person processes the whole queue and judgment is spread thin across hundreds of routine confirmations. In the automated model, the queue processes itself, and judgment concentrates on the cases that need it. Humans handle exceptions, not the queue. That inversion is the entire idea, and it is worth stating plainly because everything else is plumbing.

Workflow diagram

1. The schedule triggers checks on a cadence: new appointments at booking, upcoming appointments a set number of days ahead, and ongoing clients on a re-verification rhythm
2. Each check runs against the clearinghouse or verification service, using the HIPAA-standard eligibility transaction the industry runs on (the 270 request and 271 response)
3. Clean results write back to the EHR or billing record automatically, dated, with the five fields captured
4. Only the exceptions reach a human, as a worklist: inactive plans, carve-outs needing a different administrator, exhausted or near-limit session counts, missing authorizations, and responses that do not reconcile
5. Each resolved exception updates the record, so the same surprise does not recur next month

The exception types recur predictably, which is what makes the worklist manageable. In practice the queue is dominated by a handful of shapes: the carve-out discovered mid-check, where the response points to a different administrator and the record needs the correction once; the near-limit session count, where the right move is a conversation with the client before the covered sessions run out rather than after; the inactive plan at renewal season, usually a new card nobody has sent in yet; the authorization gap on a level-of-care change; and the response that contradicts last month, which is the one genuinely worth a phone call. A team working that list works minutes per day; a team working the whole queue works hours.

The connector categories, named honestly: clearinghouses and verification services answer the coverage question; the EHR and billing records are where the answers live; and the automation layer moves data between them to the rhythm of the schedule. Keragon is that layer, and only that layer: it is never an EHR and never a clearinghouse. It connects them, runs the cadence, writes the results back and builds the exception worklist, without code, with a BAA on every paid plan and a 14-day free trial. Eligibility work sits within its broader billing and insurance automation, and for the software selection side of this decision, the insurance eligibility verification software guide covers the tool landscape in depth. See Keragon Workflows and Agents for how the worklist pattern runs.

What still needs a person

Plenty, and saying so is what makes the rest believable. Interpreting an ambiguous benefit is judgment. Calling a payer to resolve a contradiction is judgment plus patience. Deciding what to tell a client about their cost share, and how, is care work, and it stays with a person: the client conversation is always a person, full stop. Anything involving a coverage appeal is a human process with documents attached.

And one boundary belongs in writing: automated verification is neither a payment guarantee nor a compliance guarantee. It confirms what the payer's systems report at the moment of the check, which is the best available answer and still not a promise. The practice's obligations and the payer's final determination are what they are; the automation's job is to ensure nobody is surprised by them. Nothing in this article is billing, legal, or payer-contract advice.

If your EHR has no API

Verification automation is exactly where a closed EHR bites hardest, because the schedule that should trigger the checks lives inside a system nothing can subscribe to. The pattern still works; the trigger moves. On SimplePractice, TherapyNotes, and the rest of the closed tier, the synced calendar becomes the schedule feed and results land in the billing layer rather than the chart, with a person completing the EHR writeback. The closed-EHR automation patterns cover the mechanics, and the EHR integration readiness comparison maps which platforms make this easy, hard, or manual.

Rolling it out in a group practice

The rollout that works is narrow first: one location or one clinician panel, the highest-volume payers first, running the automated cadence in parallel with the existing manual process for a few weeks. The parallel period does two jobs: it builds trust in the results before anyone depends on them, and it surfaces payer quirks (the plan whose responses need interpretation, the carve-out arrangement peculiar to one employer group) while a human safety net remains in place.

Measure the before honestly so the after means something: denial categories for a recent quarter, the hours the team spends on checks, and the miss rate you can reconstruct. Then expand payer by payer and site by site. Do not promise yourself a denial-reduction percentage in advance; measure your own, because your payer mix and caseload are the only benchmark that matters.

FAQ

What is the difference between eligibility and benefits verification?

Eligibility asks whether the plan is active on the date of service. Benefits verification asks what the active plan actually covers for this care: copays, coinsurance, remaining deductible, session limits, and network status. A plan can be perfectly active and cover far less than anyone assumed, which is why the two questions are checked together and recorded separately.

What is a behavioral health carve-out?

An arrangement in which the behavioral health benefit is administered by a separate entity from the medical plan, with its own network, rules, and contact points. Practically speaking, it means the plan on the card may not be the one you verify with, and identifying the carve-out administrator is step one in an accurate check.

How far in advance should we verify?

A working default for groups: at booking for new clients, again a few days before the appointment, and on a monthly or per-visit-cluster cadence for ongoing clients whose benefits deplete. The right cadence is the one that catches changes before the session does; automation makes the cadence cheap enough to run properly.

Do we need to re-verify returning clients?

Yes, on a rhythm. Coverage changes at plan renewals, job changes, and level-of-care changes, and session limits are depleted continuously. The re-verification check is lighter than the initial one, which is exactly why it is the first thing practices skip manually and the easiest thing to automate.

What causes most behavioral health denials?

Structurally: services delivered against inactive or misidentified coverage, exhausted session limits nobody was tracking, and missing authorizations. All three are verification failures before they are billing failures, which is why the exception worklist earns its keep at the front of the revenue cycle rather than the back.

Can eligibility checks be automated?

Yes; the checks themselves are the industry's most standardized transaction, and running them on a schedule, with results written back, is an established practice. What should not be automated is the judgment layer: ambiguous benefits, payer calls, appeals, and the client cost conversation stay with people.

Is an eligibility check a HIPAA transaction?

Yes. The eligibility inquiry and response are HIPAA standard transactions, the 270 request and 271 response, which is why clearinghouses and verification services can interoperate across payers at all. It also means the whole chain handles PHI and falls under BAA end-to-end.

Do we need a clearinghouse?

Most groups reach payers through one, because a clearinghouse turns hundreds of payer connections into a single interface for eligibility and claims alike. Some verification services bundle that access. What matters for this process is that some service can answer the 270/271 question for your payer mix; which one is a stack decision.

Who should own verification in a group practice?

Ownership of the process belongs to the billing or revenue cycle lead; ownership of the daily work should belong mostly to the automation, with the exception of the worklist as the human queue. The failure mode to avoid is diffuse ownership, where front desk, billing and clinicians each assume someone else checked.

Can this work if our EHR doesn't have an API?

Yes, with the trigger moved: the synced calendar feeds the schedule instead of the EHR, and a person completes the chart writeback. The closed-EHR patterns cover exactly how.

Does verification guarantee the claim will be paid?

No. It reports what the payer's systems say at the time of the check, which prevents preventable denials but cannot guarantee the final determination. Treat it as the best available answer, date-stamped, and keep the record; that record is also your position if a determination is disputed.

For the wider set of mental health operations this process sits inside, from intake to no-shows, the mental health automation hub is the place to start.