A multi-site FQHC network: what the first 70 days found in front of the payer

The organization

The client is a multi-site, nonprofit FQHC network in the Mid-Atlantic with medical, dental and behavioral health lines. Its payer mix is dominated by Medicaid managed care: in 2026, Medicaid produced 88% of cash, most of it paid as PPS encounters rather than per-code fees. That matters for how we measured it. Because PPS pays a fixed amount per qualifying visit, we did not use gross collection rate as a performance measure. We used receipts per business day, net AR, cash turnaround and the size of each workable inventory.

We took over the full revenue cycle in early June 2026: coding, billing, posting, AR follow-up and credentialing support. This case covers what the first 70 days found and what moved. It is a work in progress, not a finished outcome, and we say where the numbers did not move.

What we found in the first 30 days

The receivable looked like a collections problem from the outside. Inside, three findings pointed somewhere else.

• A single workqueue holding claims returned for credentialing carried $1.035M at go-live, the largest workqueue in the book. These were encounters furnished by practitioners whose payer enrollment was not complete, so the claims could not be paid until enrollment caught up.
• One code carried most of the denials. When we validated the denial history in July, the FQHC encounter code T1015 accounted for $1.113M of $1.951M of 2026 denial dollars, 57%. Because the encounter is paid as a unit, a denial on T1015 is a whole-visit loss, not a partial one. Some of what was labelled a coding denial on T1015 was in fact a credentialing problem under a different reason code.
• Coding went out without a pre-bill check. Until mid-July, physician coding billed automatically, and errors surfaced only when a payer edit or rejection came back.

The front-end leak

By the end of July, cash was strong but net AR had stopped falling. The reason was in front of the payer. Claims in claim-error status held $746K. Open charges with no first claim date, meaning no claim had ever gone out, held $900K. Those two groups overlap: $678K sat in both. Counting each dollar once, $968K of open AR had not reached a payer.

The AR production file showed where it came from. About $718K of worked claims were parked in internal handoffs between coding, billing and posting. Because we own all three functions end to end, these were our own queues, not payer delays, and they were the source of the next week's claim-error and never-claimed growth.

What we did

• Worked the credentialing return queue weekly against enrollment status, releasing claims as each practitioner's enrollment cleared.
• Put a 100% pre-bill coding review queue live in the week of July 13, so encounter claims are checked before they leave rather than after a rejection.
• Set an internal handoff rule that work moving between coding, billing and posting is not parked, with daily release of claim-error claims and a direct attack on the sources of never-claimed charges.
• Rebuilt reporting around measures that fit PPS: receipts per business day, net AR after credit balances, cash turnaround, and separate inventories for claim error, never claimed, credentialing, no response and open denials, each counted once.

What moved by mid-August

The credentialing return queue fell from $1.035M at go-live to $679K at the July close, down 34%, and to $647K by August 11. The open denial inventory fell from $1.623M at the end of June to $1.215M at the July close. Net AR, after credit balances, went from $5.331M at June 29 to $4.837M at the July close.

Cash held up while the cleanup ran. Receipts per business day averaged about $42K from January to May; July closed at $52.7K per business day, the best month of 2026, with $1.30M of cash posted. August through the 14th ran at $50.9K per business day. Cash turnaround, measured as net AR over average daily charges, fell from 96 days at the March baseline to 68 days by July 20. Of the $8.10M collected in 2026 through mid-August, $1.81M was for 2025 dates of service, older claims recovered rather than written off.

What did not move yet

We report the misses with the wins. The front-end leak grew before it eased: by August 11 the de-duplicated stuck-before-payer figure had reached $1.16M as new billing ran ahead of release, before claim-error AR and never-claimed AR both fell in the week to August 18, to $780K and $1.037M. Net AR rose slightly in August, to $4.939M on August 18. The share of AR over 90 days stayed around 56 to 57%, because new cash clears young claims fastest and the aged tail needs its own decisions. And coding denials received rose from $54K in June to $97K in July, which told us June's dip was adjudication timing, not improvement. The first claims that went through the pre-bill review queue are the real test, and they adjudicate in August and September.

What this means for other health centers

In a PPS-heavy FQHC, revenue rarely leaks at the payer first. It leaks at enrollment, at the encounter code, and in the internal queues between coding, billing and posting. Measure the union of stuck inventories, not the sum. Treat a denied encounter code as a whole-visit loss and prevent it before billing. And put a clock on every internal handoff, because a parked claim earns nothing.