Billed, allowed, collected: the number nobody puts on your dashboard

Ask an operator how the business is doing and you will usually get a big number. "We billed two million last quarter." It is the wrong number. Nobody pays billed.

Billed is not allowed is not collected

Three numbers run a treatment center's finances, and they are not the same.

• Billed is the sticker price you submit. It is inflated by out-of-network write-offs, and almost nobody pays it.
• Allowed is what the payer actually agreed to pay. This is what you collect against.
• Collected is what actually landed in the bank.

A dashboard that reports billed is reporting the one figure that never touches your account. The gap between billed and collected is where the real business lives, and it is widening.

Denials are rising, and most are never reworked

The initial claim denial rate hit 11.8% in 2024, up from around 10% a few years earlier (Kodiak Solutions, via HFMA). 41% of providers now run denial rates of 10% or higher (Experian Health, 2025 State of Claims). The money left on the table is staggering. An estimated 60% of denied claims are never reworked (HFMA), even though reworking one costs only $25 to $181 (MGMA, 2024).

The pressure is compounding. 68% of providers say submitting clean claims is harder than a year ago, and 43% report being understaffed (Experian, 2025).

The fourth number: what is expected, and when

Even collected is not the whole picture. The question every owner actually asks, "how are we doing," is really about the future. What is still expected, from which payers, and when will it arrive? That is a cash-flow forecast, and it is exactly what most treatment-center dashboards cannot produce. Meanwhile every unoccupied bed is a permanent loss of that day's revenue, with residential occupancy targets of 85 to 95% (industry benchmark), so timing is not academic.

Why the number is usually wrong

Billed is easy to display and flattering to look at. The real economics live in the remittance and adjudication detail, the line-level record of what each payer allowed and paid, which is what most systems skip. The fix is not another report. It is sourcing the money from the adjudication detail itself, measuring collection rate against allowed rather than billed, and turning the accounts-receivable picture into a forecast. McKinsey estimates that automating the revenue cycle could cut cost-to-collect by 30 to 60% (McKinsey, 2025), but only if the automation is reliable enough to trust.

You collect against what payers allow, never what you billed. A system that does not know the difference is telling you a comfortable story about a business run by three other numbers.