What a Good Billing Function Looks Like in Behavioral Health: Verification Before the First Session, Clean Claims, the Denial Queue Worked on a Clock, Patient Balances Collected at the Door, and the Six Numbers That Prove It

In a behavioral health group, every session is delivered before anyone knows whether it will be paid. The billing function is the machine that converts delivered care into collected cash, and most groups judge it by whether the biller is busy rather than by whether the machine works. The anatomy of one that does.

The front end, before the first session. Verification of benefits — confirming the client's coverage is active, that outpatient behavioral health is a covered benefit, the copay or coinsurance and the remaining deductible, whether prior authorization or a session limit applies, whether the clinician is in-network for that plan and credentialed at that location, and whether telehealth is covered at parity — completed and documented for every new client before the first appointment is kept; groups that measure it find that 20 to 40 percent of eventual denials trace to a first session delivered without this step, and that the client who learns of a US$1,800 deductible at intake behaves very differently from one who learns of it on a statement eight weeks later; the metric is the share of new clients fully verified before session one, and a good function runs above 95 percent.

Intake data quality. Demographics, insurance ID, subscriber relationship, date of birth, and the authorization number captured exactly as the payer holds them, because a single transposed digit is a rejected claim that costs rework time and ten to twenty days of cash; the metric is the first-pass claim acceptance rate at the clearinghouse, and a good function runs above 97 percent.

Charge capture and coding. Every completed session produces a charge the day it is delivered, with the correct procedure code for the service and duration, the diagnosis code the clinician documented, the correct rendering and billing provider, place of service, and any required modifier — telehealth, supervised billing for an associate under the payer's rules, interactive complexity where it applies; the lag between date of service and charge entry is a metric in itself, and anything beyond two business days means revenue is being recognized late and claims are aging before they are filed.

Claim submission. Daily electronic submission through a clearinghouse with scrubbing rules tuned to the group's payers, rejections worked the same day, and a reconciliation that proves every completed session in the practice management system became a submitted claim — the gap between sessions delivered and claims filed is the first place revenue disappears, and in groups without the reconciliation it is routinely two to four percent of sessions.

Payment posting. Electronic remittance advice posted within two business days of receipt, line by line, with contractual adjustments separated from denials and from patient responsibility — because a payment posted as a lump "adjustment" hides a denial that could have been appealed and a patient balance that should have been billed; the metric is posting lag and the share of remittances posted automatically versus by hand.

The denial queue, worked on a clock. Every denial categorized by reason — eligibility, authorization, timely filing, coding, medical necessity, duplicate, coordination of benefits, credentialing — with an owner for each category; denials worked within five business days of receipt; appeals filed inside each payer's window, which ranges from 30 to 180 days; and a monthly review of everything over 90 days with a decision — appeal, resubmit, transfer to patient responsibility, or write off with a reason code; the metrics are denial rate by reason, appeal overturn rate, and days from denial to resolution, and a good function runs an initial denial rate under 8 percent and an overturn rate above 50 percent on the denials it chooses to appeal.

Authorization management. For payers that authorize sessions in blocks, a tracker by client that shows authorized sessions, used, remaining, and expiration, with a reauthorization request triggered two to three sessions before exhaustion — because a session delivered beyond the authorization is unbillable, and the clinician who delivered it did not know; the metric is the unbillable-session rate from authorization lapses.

Patient balances collected at the door. The estimated copay, coinsurance, or deductible amount collected at the time of service using the verification data; a card on file with written consent to charge for balances and for late-cancellation fees where the payer contract and state law permit; statements for residual balances within a week of the remittance, not at month-end; and a policy on scheduling with an outstanding balance — groups that collect at the time of service recover 85 to 95 percent of patient responsibility, while groups that bill after the fact recover 40 to 60 percent, and patient responsibility is the fastest-growing component of behavioral health receivables as high-deductible plans spread.

Credit balances and refunds. Overpayments by clients or payers identified monthly and refunded within the timeframes state law and payer contracts set — because unrefunded credit balances are a liability, a compliance exposure, and in several states subject to unclaimed-property rules; and because the refund process is the single most common vehicle for misappropriation in a small office, it runs with dual approval.

Timely filing discipline. Each payer's filing window — 90 days to a year — tracked by claim, with a report of claims approaching the limit, because a claim that misses the window is a total loss regardless of merit.

Credentialing coordination. The billing function is the first to see a credentialing failure as a denial; it keeps the clinician-by-payer credentialing matrix current and tells scheduling which payers each clinician can see, because a session scheduled with a payer the clinician is not credentialed for is a denial waiting to be filed.

Payer behavior intelligence. The billing function knows, from remittance data, each payer's actual net per session by code, days to pay, denial patterns, and documentation quirks — and feeds that to the CFO's payer file, because the negotiation case for a contract renewal is built from exactly this data.

The structure

In a group of ten to twenty clinicians, one full-time biller per 1,500 to 2,500 claims a month is a reasonable load with automation; below that the function is a shared role, above it a billing lead with specialists by payer group; outsourced billing companies at 4 to 8 percent of collections can work if the group keeps the front end — verification, intake quality, charge capture — in-house and holds the vendor to the metrics below, because outsourcing the back end without controlling the front end outsources the symptom and keeps the cause.

The six numbers that prove it works

Verification-before-first-session rate (above 95 percent), first-pass acceptance (above 97 percent), initial denial rate (under 8 percent), days in receivables (under 40 for a commercial-heavy mix, under 50 with Medicaid), patient-balance collection at time of service (above 85 percent of estimated responsibility), and net collection rate — cash collected over net expected revenue after contractual adjustments — which is the one number that summarizes the machine, and a good function runs above 96 percent.

What a weak function costs is not visible on the income statement, because it arrives as revenue that never happened: sessions never billed, denials never appealed, patient balances never collected, claims that missed the window — in a US$3 million group a weak function routinely leaves 5 to 10 percent of expected revenue uncollected, which is US$150,000 to US$300,000 a year, more than the cost of running the function well.

Worked example

A 20-clinician outpatient group, US$4.1 million in net expected revenue, with an outsourced billing company at 6 percent of collections and no front-end controls. The finance function measured the machine:

Illustrative figures for a hypothetical organization; not a client's data.

The vendor was not the primary problem. The group's own front end — verification, intake data, charge capture, time-of-service collection — was feeding the vendor claims that could not be paid, and the vendor's contract paid it on collections, not on the revenue it failed to collect. The fix: a verification gate in the scheduling workflow; card on file at intake; a daily sessions-to-claims reconciliation owned in-house; a denial-reason dashboard reviewed weekly with the vendor; and a renegotiated vendor agreement with the six metrics as performance terms. Net collection rate reached 95 percent in two quarters — about US$250,000 a year recovered, against a front-end cost of one additional intake coordinator.