The short answer
Keep the work that needs to happen inside the building: eligibility at check-in, coverage and demographic capture, charge entry, and patient conversations. Hand off the payer work that piles up: rejections, denials, appeals, aged AR and underpayments. Your biller stays, and the patient relationship stays with the practice.
The split works only with three things in place: a written task map, one owner per queue, and shared reporting both sides can see. Without them, work falls between the two teams and the filing clock keeps running.
Why practices split the work
Experian Health's 2025 State of Claims survey, released 22 September 2025 and based on 250 revenue cycle decision makers, shows where the pressure sits:
43%
of providers report being understaffed
54%
say claim errors are increasing
68%
find clean claims harder to submit than a year ago
90%
of denials are reworked with at least some human review
Read together, those numbers describe a workload problem, not a skills problem. Back-end payer work grows with denials: each one needs a person to read it, fix it, and resubmit or appeal. The front desk does not shrink to make room. Check-in, eligibility, coverage questions and patient calls all still happen every day, at the same volume.
An in-house biller sits between those two loads. When denials rise, something gets deferred, and it is usually the payer follow-up, because the patient in front of the desk can't wait. Splitting the work lets the biller keep the part that needs to be in the building. Before deciding, check how your billing is performing.
The task map
This is a starting framework, not a rule. The split depends on your specialty, your volume and what your biller does best. What matters is that every row has one answer, written down.
| Task | Who | Why |
|---|---|---|
| Eligibility and benefits at check-in | Keep in-house | Needs the patient there |
| Coverage, COB and demographic capture | Keep in-house | Same, and the root of most front-end denials |
| Charge entry and coding | Either | Depends on specialty and staffing |
| Claim submission | Either | Either works, as long as the owner is named |
| Clearinghouse and payer rejections | Hand off | Must be worked daily or the filing clock wins |
| Denials and appeals | Hand off | Phone time and appeal deadlines |
| AR over 90 days | Hand off | Deadline-driven triage |
| Underpayments | Hand off | Needs fee schedule comparison |
| Patient statements and conversations | Keep in-house | The relationship stays with the practice |
The front-end rows stay in-house for a practical reason: they need the patient, the insurance card and the conversation at the desk. A remote team can't scan a card or ask whether a patient changed jobs. The biller who knows the patients is the right person for that work.
The back-end rows go out because they are queue work with deadlines. Rejected claims never reached the payer, so the filing clock keeps running while they wait. Denials need hold time on payer lines and appeals with their own windows. Aged AR needs sorting by deadline before anything is reworked. Underpayments need someone to compare each payment to the contracted or fee schedule amount. None of it needs to happen at the desk, and all of it suffers when it is done only at month end.
The two "Either" rows are where practices differ most. A biller who codes well for your specialty should keep coding. Claim submission can sit on either side, but only one side.
Handoff rules that make it work
A split model fails at the seams, not in either team. These six rules close the seams.
- 01
One named owner per queue
Rejections, denials, aged AR and underpayments each have one person responsible. A queue shared by two teams goes unworked, because each side assumes the other has it.
- 02
Rejections worked within days, not at month end
A rejection hasn't been filed yet. Every day it sits counts against the payer's filing limit.
- 03
Root causes fed back to the front desk every week
Each denial reason that starts at registration goes back to the people at check-in, in plain terms. Without that loop, the back end fixes the same error every month.
- 04
Work inside your existing EHR
The partner works in your system, with no data migration and no second system of record. Your biller sees every note and every status change.
- 05
One shared view of AR aging and denial reasons
Both sides read the same aging report and the same denial categories. Disagreements about the numbers end when there is only one set.
- 06
A written escalation path for anything near a deadline
When a claim is within weeks of its filing or appeal limit, the path to a decision is written down: who calls whom, and by when.
When splitting won't help
If nobody owns eligibility at the front desk, handing off the back end only moves the rework. Denials that start at registration will keep arriving, and the partner will keep fixing them. The fix has to start with who checks coverage, and when.
If claim volume is small and your biller is keeping up, splitting adds coordination cost for little gain. Look at the rejection and denial queues first. If they are short and current, the model you have is working.
A split model is for a practice that keeps its biller. It is a different decision if your biller has just left.
One practice that kept its biller
One AR recovery engagement. A family medicine practice in Minnesota. Individual results vary. One engagement is not a forecast.
AR. The practice had an in-house biller. Rejections had gone unworked for about a year, and 245 claims were in scope.
Found. Roughly $40,000 to $50,000 in billed charges was sitting unworked in the rejection queue.
Limit. UnitedHealthcare's 90-day window had closed on most affected claims, with no proof of timely submission on file.
Proof. About $5,500 was recovered, across 50 to 55 paid claims, the bulk within 45 to 50 working days.
Front desk. 70% of denials originated at the front desk.
Change. The practice kept its in-house biller and changed the front desk. Eligibility is now verified with software before arrival, every patient is checked the night before, and walk-ins stopped. The biller learned directly from the biller on the account which mistakes to stop making.
This was an AR recovery engagement and an audit, not an ongoing billing relationship. The practice acted on the diagnosis itself. Read the full engagement write-up.