The population needs more than one total
I would start by agreeing on what unapplied means locally. It may include a deposit without remittance, an unmatched payment, a payment posted to a clearing account, or a patient payment waiting for liability. Some items are temporary timing differences. Others require research, correction, or a refund. Combining them in one aging total makes ownership difficult.
I would retain the difference between insurance cash and patient payments and between HB and PB. A provider-level adjustment also needs its own treatment because it may explain a difference between an electronic payment and the sum of claim payments without belonging to a single patient. The reconciliation should identify those categories rather than force every amount onto an account.
The request should connect the bank and the ledger
I would request bank deposit or electronic payment references through finance's approved process, cash receipts, remittance files and runs, posting batches, clearing account activity, and unresolved matching exceptions. I would need dates and identifiers that let the team connect those sources without relying only on equal dollar amounts. Two payments for the same amount are not necessarily the same receipt.
I would ask who owns each reconciliation point and how differences are currently documented. Payment posting may not control missing payer remittance, merchant settlement timing, or an accounts payable offset. Finance, treasury, interfaces, and operational staff should agree on the division of work. The request should include the evidence behind resolved discrepancies so the new process can distinguish repeat defects from legitimate timing.
The investigation should explain where the link broke
For an unmatched insurance receipt, I would trace whether the payment arrived before the ERA, whether the identifying information changed, whether the file failed processing, or whether claim matching left an undistributed amount. For patient payments, I would inspect the encounter and expected liability before concluding that immediate application is appropriate.
In Epic, I would review remittance matching, payment allocation, relevant posting methods, clearing account behavior, and the credit workflow used by the organization. I would not apply PB credit assumptions indiscriminately to HB accounts. I would also investigate provider-level adjustments separately from patient claim activity, with finance confirming their intended accounting treatment. The goal is a supported disposition for the receipt, not merely a way to make the unmatched list shorter.
The operating process needs a controlled roll-forward
I would create a daily or otherwise agreed reconciliation view showing opening unresolved items, new receipts, amounts applied or otherwise resolved, and ending differences. The view should preserve receipt-level identity. Summing repeated snapshots of an unresolved receipt should never increase the apparent cash balance.
Each difference needs a category, owner, required evidence, and expected review point. A timing item should have an anticipated event, such as a remittance arrival, rather than an indefinite defer date. A correction needs a traceable link to the original and revised posting. If an item belongs to another department, the handoff should define acceptance and return status instead of transferring the problem without confirmation.
A receipt that crosses a month-end boundary should keep the same identifier and ownership in the next period. I would reconcile that carry-forward explicitly so it cannot disappear between a daily cash report and the finance close.
Testing should include offsets and duplicate-looking receipts
I would test a deposit with several remittance runs, a remittance containing multiple claim payments, an amount reduced by provider-level adjustments, a duplicate file, and a reversal after the original item reconciled. I would include two equal-value receipts with different identifiers to verify that the process does not clear them simply because the numbers match.
For automated matching, I would confirm the supported identifiers and evaluate the consequences of ambiguous matches before expanding the scope. A manual exception path remains necessary. Acceptance should include the cash team and finance, with evidence that the receipt, posting, and accounting views remain reconcilable after the correction. An automated match rate without a review of incorrect matches would give an incomplete picture.
The outcome is accurate application and less unresolved work
I would measure unresolved receipts by category and age, the proportion reconciled within the agreed interval, and the number requiring repeated research. I would report timing differences separately from unexplained differences and identify whether the clock begins at bank receipt, file availability, or posting. That definition prevents inconsistent aging comparisons.
Applying money already received does not create an equal amount of new cash collections. It can improve account accuracy, reporting, staff effort, and downstream patient or payer follow-up. Where the work identifies genuinely missing receipts or preventable losses, I would measure those separately. The operating review should show that new unapplied items are reaching an owner quickly and that repeat causes, such as missing remittance or an identifier mismatch, are becoming less common.