Chapter 31 — Accounts Receivable: Aging, Work Queues, Follow-Up, Overpayments, and the Metrics That Run a Business Office

📍 Where you are

Chapter 29 classified the denial and Chapter 30 argued it. This chapter is about everything that is still sitting there.

Every claim that has been submitted and not resolved, every patient balance that has been billed and not paid, every payment that arrived and could not find its account — all of it accumulates in one place, and that place has a name: accounts receivable.

This is the chapter where the book stops following one claim and starts managing all of them at once — which is a different job, with different tools, and it is the job most business offices are actually doing all day.

One number in this chapter is load-bearing for the whole book. §31.7 publishes the fully loaded cost of a minute of denial-management staff time. Chapter 29 published the minutes. Chapter 40 puts them together, and nothing before Chapter 40 does.

Overview

Accounts receivable is the money you have earned and do not have, and the first thing to understand about it is that it is not a report. It is a standing inventory — hundreds or thousands of accounts, each somewhere in the hundred-day arc this book has been walking one account through since Chapter 1 — and it is the raw material of the entire back end of the revenue cycle.

The second thing to understand is that most of the numbers reported about it are softer than they look. Days in accounts receivable — the field's favorite metric — changes by double digits depending on definitional choices nobody writes down. An aging report can be made to look better by giving up — write off the old accounts, and the report improves on the day you collect nothing. A net collection rate rises when a payer silently underpays you, because the denominator shrinks with the numerator (Chapter 28 §28.8 — this chapter will not rebuild that argument, but its dashboard is built on it). A business office that manages to these numbers without knowing how they bend will be managed by them instead.

And the third thing to understand is that accounts receivable (AR) contains obligations as well as assets. Sitting inside the receivable — netted against it on most reports, which is the problem — are credit balances: money the practice holds that may belong to a payer or a patient. Some of those credits are overpayments, and a known overpayment from a federal health program carries a sixty-day legal clock that converts an accounting backlog into a False Claims Act exposure. The most dangerous number in this chapter is not a balance somebody failed to collect. It is a balance somebody failed to give back.

The chapter runs in three movements. First, the measurement: what AR is (§31.1), how it ages (§31.2), and how the two headline metrics — days in AR (§31.3) and AR over ninety (§31.4) — are computed honestly and gamed routinely. Second, the work: the queue that finds money (§31.5), the follow-up that gets it (§31.6), and the arithmetic of deciding when to stop (§31.7). Third, the obligations and the instruments: credits and unapplied cash (§31.8), overpayments and the sixty-day rule (§31.9), bad debt and charity care (§31.10), and the dashboard a manager should actually look at (§31.11).

The chapter's thesis: AR is not a pile to be shoveled. It is a portfolio to be managed — and the difference between the two is knowing what each account is worth, what it costs to pursue, and which clock is running against it.

31.1 What accounts receivable actually is

Chapter 1 defined accounts receivable and this chapter runs it as a business. The three-word reminder: money earned, uncollected.

Start with what the number is made of. When a claim goes out the door, the full charge posts to the account — and Chapter 23 §23.7 established that the charge is an opening figure with only an accidental relationship to what anyone will pay. Which means raw AR — "gross AR," AR stated at charges — is inflated by every contractual adjustment that has not been posted yet. A practice whose contracts allow 55 or 60 cents of the charge dollar is carrying AR of which a large share was never collectible by anyone, and never will be, and represents no failure when it disappears.

AR becomes honest as it adjudicates. Once the remittance posts, the contractual adjustment comes off, and what remains is real: an allowed amount somebody actually owes. This is why the same dollar figure means completely different things at different ages — young AR is mostly air and expectation; adjudicated AR is mostly money.

🧮 Run the Numbers

What one account contributes to AR, before and after adjudication. Account 10-4471, using the frozen figures from Chapters 1 and 28. (Constructed teaching file.)

```text DAY 2 — claim submitted. AR at charges: 367.00 billed -30.00 copay collected day 0 (not AR; it is in the drawer) ────── 337.00 "receivable" — of which the contracts will erase 150.72 no matter what anyone does

DAY 17 — first remittance posted (Ch. 28 §28.11): payment .................... 70.30 contractual, lines 2–4 ..... 94.12 contractual, line 1 (CO-45) 56.60 ────────────────────────────────── REMAINING AR ............... 145.98 insurance AR (line 1, denied, on appeal) ............. 128.40 patient AR (coinsurance, lines 2–4) ............. 17.58 ```

Checks: 367.00 − 70.30 − 94.12 − 56.60 = 145.98 ✓ · 128.40 + 17.58 = 145.98 ✓

Read what happened to the number. On day 2 the account "contained" \$337.00. On day 17 it contains \$145.98 — and no money was lost in between. The difference is mostly contractual adjustment that was never going to be collected, plus \$70.30 that was. A manager who watched gross AR fall from \$337.00 to \$145.98 and celebrated, or watched it and panicked, would be wrong both times.

And read what the \$145.98 is.** It is not one kind of money. **\$128.40 is a denied line with an appeal pending — it will resolve when a payer decides, and pushing on it means Chapter 30's work, not a statement. \$17.58 is a patient balance — it will resolve when a statement goes out and a person pays it, which is Chapter 32's work. Same account, two receivables, two workflows, two clocks. AR management is the discipline of never confusing them.

That split — insurance AR versus patient AR — is the first cut every AR analysis should make, because the two behave differently in every way that matters. Insurance AR resolves on adjudication schedules and dies by timely filing and appeal deadlines; it is worked by phone, portal, and appeal. Patient AR resolves on statement cycles and household budgets; it is worked by statements, calls, payment plans, and — carefully, and last — collection agencies. A combined number hides both.

Two more distinctions, quickly, because the rest of the chapter uses them.

AR is not revenue and it is not cash. It is a promise with a decay rate. The practice's accountant cares about when revenue is earned; the practice's payroll cares about when cash arrives; AR is the gap between them, and the size and age of that gap is what this chapter's metrics measure.

And AR is not all positive. Somewhere inside the total, on almost every system, are negative balances — credit balances — netted silently against the positive ones. Hold that thought until §31.8; for now, know that every AR total you read should make you ask whether credits are netted in it, because the answer changes both headline metrics in §31.3 and §31.4.

This is the book's sixth theme working at full scale: every day a claim sits, it is worth less. Chapter 1 promised this chapter would quantify that. Honestly, in ranges (Tier 2 — the exact figures vary by specialty, payer mix, and study, and you should distrust anyone who quotes them to a decimal place): industry experience is that a balance pursued in the first 30 to 60 days collects at a rate several times higher than the same balance pursued after 120 — patients move, coverage terminates retroactively, filing windows close, memories of the visit fade, and the paperwork that would win an appeal gets harder to assemble. A patient balance that reaches a collection agency typically returns only a minority of its face value, and the agency's contingency fee comes out of that. The decay is not linear and it is not gentle. Speed is not enthusiasm in this business; it is a line item.

31.2 The aging report and its buckets

The aging report is AR's org chart: every open balance, sorted into columns by how long it has been open. The columns are the aging buckets, and the near-universal convention is thirty-day steps:

   THE AGING REPORT — the shape of the standing pile

   0–30        31–60       61–90       91–120      121+
   ┌─────────┬─────────┬─────────┬─────────┬─────────┐
   │ mostly  │ pending │ problem │ chased  │ dying   │
   │ normal: │ + slow  │ claims: │ or      │ or      │
   │ claims  │ payers +│ denials,│ forgot- │ dead:   │
   │ in      │ first   │ appeals,│ ten —   │ decide  │
   │ process │ patient │ no-     │ which?  │ some-   │
   │         │ state-  │ response│         │ thing   │
   │         │ ments   │ claims  │         │         │
   └─────────┴─────────┴─────────┴─────────┴─────────┘
    a claim should      ◄── each bucket is a question,
    LEAVE 0–30 by           not a verdict
    paying, not by aging

A healthy report is front-loaded. Claims adjudicate in days to a few weeks (Account 10-4471's first remittance arrived on day 17); patient statements cycle monthly. So most of a well-run AR sits in the first bucket, the second bucket holds the slow and the pending, and everything past ninety is small — and, more important than small, explained. §31.4 is about what "explained" means.

Before you read any aging report, ask three questions about how it was built. The answers change everything on the page, and no report prints them.

Aged from what date? The honest default is date of service — the clock the patient and the payer both experience. But systems can age from date of claim submission or date of last activity, and the last one is where the trouble lives. If rebilling a claim resets its age — a practice called re-aging — then every corrected claim, every resubmission, every statement cycle moves an account back to the young end of the report. An AR that is being worked badly but touched often will look young forever. Re-aging is occasionally defensible for specific dates (a secondary claim genuinely could not exist before the primary paid — Chapter 28 §28.10), but as a default it converts the aging report from a measurement into a diary of how recently somebody clicked.

Split by what? At minimum: insurance versus patient responsibility (§31.1's two workflows), and by payer within insurance. An aggregate aging is a landscape photograph; the payer split is where you see that one payer's column looks nothing like the others — which is a finding about that payer, not about your office (Chapter 29's Case Study 1 is the cost of having no category for that sentence).

And are credits netted? A \$5,000 credit balance sitting in the 121+ column reduces that column's total. Old credits camouflage exactly the old debits you most need to see — and the credits themselves are the subject of §31.8 and §31.9, where they turn out to be the compliance problem. Run the aging gross, with credits on their own report.

📋 Read the Chart

Source: month-end insurance-and-patient aging summary, a five-provider primary care practice (constructed teaching example) What it says:

```text AGING SUMMARY — all financial classes, aged by date of service $ % of AR 0–30 ...................... 198,400 48.1% 31–60 ..................... 94,850 23.0% 61–90 ..................... 48,200 11.7% 91–120 .................... 31,150 7.6% 121+ ...................... 39,700 9.6% ───────── ────── TOTAL AR .................. 412,300 100.0%

AR OVER 90 ................ 70,850 17.2% CREDIT BALANCES (memo) .... (28,600) not netted above ```

Checks: 198,400 + 94,850 + 48,200 + 31,150 + 39,700 = 412,300 ✓ · 31,150 + 39,700 = 70,850 ✓ · 70,850 ÷ 412,300 = 17.2% ✓

What it means: roughly half the AR is young and presumably normal; about a sixth is over ninety days old. Whether 17.2% is a problem is not answerable from this page — §31.4 decomposes it — but three things on this page are already worth something.

The report is aged by date of service and says so. That one line makes every other number on the page more trustworthy than the same numbers on a report that does not say.

The credits are a memo line, not a deduction. Whoever built this report understood §31.1's warning. The \$412,300 is gross; a netted report would have shown \$383,700 and looked better for the worst possible reason.

And the middle buckets are where the money is moving. \$94,850 in 31–60 is next month's 61–90. The cheapest place to fix the over-90 number is two buckets to its left — which is the entire argument of §31.5.

What it doesn't say: whose money any of this is, what any account is worth net of contractuals, or what anyone has done about any of it. An aging report locates money in time. It says nothing about money in space — which payer, which root cause, which action. That is the queue's job.

One habit worth building now: read the aging report as flow, not stock. The buckets are a snapshot, but the interesting question is always the conveyor — what share of each bucket resolves before it ages into the next one. A 0–30 bucket that is 48% of AR is fine if it drains; the same bucket is a disaster in formation if it does not. Comparing this month's 31–60 with last month's 0–30 takes thirty seconds and tells you whether the machine is keeping up — which no single month's snapshot can.

31.3 Days in AR and how to compute it honestly

Days in AR is the industry's headline metric, and the idea is genuinely good: convert the standing pile into time. If all charging stopped today, how many days of typical charges does the uncollected pile represent?

Days in AR = total accounts receivable ÷ average daily charges

That is the whole formula, and every word of it is a decision.

🧮 Run the Numbers

The same practice, the same afternoon, four defensible numbers. (Constructed teaching example — the practice from §31.2's aging report.)

```text THE INPUTS AR per the aging report (credits not netted) .... 412,300 credit balances ................................. 28,600 trailing-12-month gross charges ............... 2,920,000 → average daily charges: 2,920,000 ÷ 365 ...... 8,000 trailing-90-day gross charges (a strong quarter) 766,500 → average daily charges: 766,500 ÷ 90 ......... 8,516.67

FOUR COMPUTATIONS 1 honest: 412,300 ÷ 8,000 ............ 51.5 days 2 credits netted: (412,300 − 28,600) = 383,700 ÷ 8,000 .......... 48.0 days 3 90-day denominator, after a busy quarter: 412,300 ÷ 8,516.67 ......... 48.4 days 4 after writing off 67,900 of aged denials: 344,400 ÷ 8,000 ............ 43.1 days ```

Checks: 2,920,000 ÷ 365 = 8,000.00 ✓ · 412,300 ÷ 8,000 = 51.54 ✓ · 383,700 ÷ 8,000 = 47.96 ✓ · 766,500 ÷ 90 = 8,516.67; 412,300 ÷ 8,516.67 = 48.41 ✓ · 412,300 − 67,900 = 344,400; 344,400 ÷ 8,000 = 43.05 ✓

Same practice. Same accounts. 51.5 or 48.0 or 48.4 or 43.1 — an eight-day spread produced entirely by definitional choices, without collecting one additional dollar. A manager comparing this practice's "48" to a benchmark, or to its own "51" from last year under a different definition, is comparing nothing to nothing.

Walk through the four, because each one is a lesson.

Netting credits (computation 2) understates AR with money you owe other people. The \$28,600 of credits is not collection performance; much of it is refund obligation, and some of it may be on a legal clock (§31.9). Netting it into the numerator makes the practice look faster because it is holding money it should have returned. Compute days in AR gross, and report the credit total separately, where it can frighten someone appropriately.

The denominator window (computation 3) imports whatever just happened to charges. A 90-day window after a strong quarter deflates the metric; the same window after a slow month or a provider's leave inflates it — in both cases saying something about charge volume, while everyone reads it as collection speed. A 365-day window is more stable; a 90-day window is more current; either is defensible, but only one at a time, forever. (And note what both windows inherit: the numerator and denominator are both stated at charges — so a fee schedule increase raises both and roughly cancels, but a charge-capture failure shrinks the denominator and makes collections look slower. No version of this metric is independent of Chapter 23.)

And the write-off (computation 4) is the important one, because it is the one people do on purpose. Writing off \$67,900 of aged, denied, probably-dead claims may be exactly the right decision — §31.7 is about making it deliberately. But it improves days in AR by 8.4 days on the day it happens, and the improvement is indistinguishable, on the dashboard, from getting paid faster. A business office under pressure to hit a days-in-AR target owns a lever that hits the target by surrendering. The metric cannot tell pursuit from abandonment. Only the write-off report can — which is why §31.11's dashboard never shows days in AR without adjustments beside it.

⚠️ Where Claims Die

The failure mode is not computing days in AR wrong. It is changing how you compute it without writing the change down.

A new practice management system nets credits where the old one did not. A new manager prefers the 90-day denominator. A conversion drops a legacy AR bucket. Each produces a step-change in the metric that looks like performance — in either direction — and a trend line across the change is meaningless and looks real. Chapter 29 §29.7 made this exact argument about the denial rate, and it generalizes: a metric whose definition moved is worse than no metric, because you can pick.

The fix costs one paragraph: write the definition on the dashboard itself — numerator, credits gross or net, denominator window, and the write-off total for the period. §31.11 does. And when you must change a definition, run both versions side by side for a few periods, so the trend has a bridge.

Where should the number be? Tier 2, ranges, verify against current benchmarking sources: practices commonly target days in AR under 40, and under 35 is considered strong; hospital figures run higher and are computed differently. The Healthcare Financial Management Association (HFMA) maintains standardized definitions for revenue cycle metrics — its MAP Keys — precisely because the definitional games above made cross-organization comparison meaningless; Medical Group Management Association (MGMA) survey data is the common practice-side benchmark. Use external benchmarks to ask questions, never to declare victory — Chapter 29 §29.7's warning that a published rate was computed under somebody else's definitions applies here with at least as much force. The comparison that actually manages a practice is you against you, under one written definition.

🎓 Exam Watch

Certification exams — the Certified Professional Biller (CPB) exam in particular — test days in AR as formula recall and arithmetic: total AR divided by average daily charges, with average daily charges built from a stated period. Read the question stem for the period — annual charges must be divided by 365 before anything else happens, and the distractor answers are what you get by skipping that step or dividing by 12.

The exams also like gross versus net collection rate (Chapter 23 §23.10): gross = payments ÷ charges; net = payments ÷ (charges − contractual adjustments). The trap is that "collection rate" unqualified usually means net on the exam and means whatever the report writer wanted in real life. And know the aging buckets as conventions: 0–30, 31–60, 61–90, 91–120, 120+. The exam treats these as facts; this chapter's point — that the honest content of the buckets depends on dating and netting choices — is a working-life lesson layered on top of the exam answer, not a replacement for it.

31.4 AR over ninety, and the number that hides in it

The second headline metric is the share of AR older than ninety days — "AR over 90" — and it exists because days in AR is an average, and averages hide tails. Two practices can share a days in AR of 45 where one has a uniform, slightly slow pipeline and the other collects beautifully except for a fifth of its AR that is quietly fossilizing. AR over 90 is the fossil detector.

Tier 2, as always in ranges: common practice targets put AR over 90 under roughly 15–20% of total AR, with the usual variation by specialty and payer mix — a practice with heavy litigation-driven coverage (workers' compensation, liability) will run structurally older. Verify current benchmarks; compare against yourself.

But the metric's real use is not the percentage. It is the decomposition — because "over ninety" is not a category of money, it is an age, and completely different kinds of money reach that age for completely different reasons.

The practice from §31.2 has \$70,850 over ninety days — 17.2%. Here is what it actually is (constructed teaching example):

   AR OVER 90 — 70,850 — DECOMPOSED

   patient balances on ACTIVE payment plans .......... 16,400
     paying as agreed; old only because the plan
     is longer than 90 days
   one payer's pended-claims project ................. 21,300
     a known enrollment-file problem, documented,
     escalated, with a payer contact and a date
   denied lines with appeals PENDING DECISION .........  9,850
     worked; waiting on the payer's clock, tracked
     per Chapter 30 §30.10
   patient balances, statements exhausted,
     pre-collection review .......................... 10,600
     decision required: §31.7 and §31.10
   insurance balances, NO ACTIVITY ON RECORD ......... 12,700
     ◄── THE NUMBER THAT HIDES IN IT
                                                     ───────
                                                      70,850

Check: 16,400 + 21,300 + 9,850 + 10,600 + 12,700 = 70,850 ✓

Read it from the bottom. Of \$70,850 that the headline metric presents as one problem, **\$12,700 is the actual problem — insurance money nobody has touched, where every passing week burns appeal windows and filing deadlines. Another \$10,600 needs a decision, not a chase. The remaining \$47,550 is old but known: managed payment plans, a documented payer project, appeals awaiting decisions. Old-and-known is a fine thing for AR to be. Old-and-unexplained is the only bad kind of old**, and the headline percentage cannot tell them apart.

Two corollaries, both of which run against instinct.

A low AR over 90 can be a symptom. The fastest way to a beautiful aging report is §31.3's fourth computation: write off everything that gets old. A practice that abandons aggressively will show gorgeous AR metrics and a quietly eroding net collection rate — which is why neither number means anything alone, and §31.11 chains them to each other on purpose.

And a rising AR over 90 is sometimes the sound of doing the right thing. A practice that starts appealing denials it used to write off will age — appeals take Account 10-4471's forty-plus days to resolve — and the aging is the cost of recovering money it used to abandon. The metric punishes the improvement for a quarter or two. A manager who cannot decompose the number will order the improvement reversed.

🔍 Check Your Understanding

The practice above hires you, hands you the \$70,850 decomposition, and gives you one afternoon.

1. Which component do you work first, and why?
2. Which component should probably leave AR entirely, and for which two different destinations?
3. Which component is evidence about a payer rather than about the practice, and what does Chapter 29 §29.4 say a practice needs before that sentence can appear in its data?
4. The managing partner wants "over-90 under 12% by year end." Name one honest way and one dishonest way to get there, and how §31.11's dashboard would tell them apart.

Answers: (1) The \$12,700 with no activity — it is the only component actively losing rights, and §31.5's deadline sort exists for exactly this money. (2) The \$10,600 pre-collection patient pool: some of it to a financial-assistance determination (charity care, §31.10), the remainder to a deliberate bad-debt disposition (§31.10) — both of which are decisions, and neither of which is served by leaving the balances to age further. (3) The \$21,300 pended-claims project — and a root-cause category for payer error, without which it would have been recorded as the practice's own failure. (4) Honest: drain the no-activity pool and resolve the pre-collection pool through real determinations. Dishonest: a bulk write-off timed to the report. The dashboard's adjustments line — write-offs by category, in dollars, beside the aging metric — is the tell.

31.5 Building a work queue that finds money

Several chapters have promised this section. Chapter 6 §6.8 said a billing queue should be sorted by whatever destroys value fastest and deferred the build; Chapter 1 §1.3 warned that an unsorted work list eventually writes off a collectible account for timely filing. Chapter 29 §29.5 built the denial queue — deadline, then category, then dollar. This section builds the general AR follow-up queue around the same spine, because the logic that ordered denials orders everything: the queue's job is not to process accounts. The queue's job is to find money — and to find it before a date kills it.

Principle one: subtract what is not money before sorting what is. The single most common queue failure is a queue full of items that cannot produce revenue no matter how hard they are worked:

• Credit balances — route them to §31.8's dedicated queue. They are obligations. Working them "for money" is exactly backwards, and burying them in a collection queue is how sixty-day clocks get missed.
• Dead items — timely filing expired with no proof (Chapter 27 §27.7), appeal rights exhausted, payer insolvent, balance below the §31.7 threshold. Close them, categorize them, count them (Chapter 29 §29.5's "no action" discipline). A queue carrying corpses overstates the recoverable work and demoralizes the people working it.
• Items nobody in the queue can fix — credentialing, chargemaster, contract loading. Route them to their owners. Chapter 29's work-queue export showed 58 items of guaranteed-futile labor; the AR queue grows the same tumors.

Principle two: sort by what destroys value fastest. For AR, in order:

   THE AR FOLLOW-UP QUEUE — SORT ORDER

   1  DEADLINE       timely filing on unbilled/corrected claims ·
                     appeal windows on denied lines (Ch. 30) ·
                     secondary filing windows after a primary pays
                     ► a date is the only thing that converts
                       recoverable to unrecoverable all by itself
   2  EVENT DUE      the follow-up date a previous touch set:
                     "payer committed to adjudicate by the 22nd" ·
                     "appeal decision due day 59" · "statement 2
                     dropped, review in 30"
   3  CATEGORY       like with like: one payer's no-response claims
                     are ONE investigation and often ONE phone call
   4  EXPECTED VALUE within the above — and expected value means
                     the ALLOWED amount, never the charge

Why expected value and not charge, and why it is fourth and not first. A queue sorted by charge chases the biggest sticker prices — and Chapter 23 established that the sticker is arbitrary. A \$1,400 charge that adjudicates to a \$212 allowed amount is a \$212 item. Sorting by allowed keeps the queue honest about what it is actually retrieving; sorting it fourth keeps a large old claim from outranking a small claim that dies Friday — Chapter 29 §29.5's argument, unchanged. (And when claims are held or corrected across a code-set boundary, code from the files in force on the date of service: ICD-10-CM changes every October 1, CPT every January 1, HCPCS Level II and the NCCI edits quarterly. Verify in the current book or encoder — never from a textbook, including this one.)

Principle three: feed the queue with events, not just age. This is the difference between a queue that finds money and a queue that processes the aging report. An aging-fed queue meets a no-response claim when it turns 45 or 60 days old. An event-fed queue met it weeks earlier, because the events were already on file:

• The acknowledgment trail (Chapter 27 §27.6): a claim with no 277CA a few days after submission is findable at day 5, not day 50.
• Expected adjudication windows: this payer normally adjudicates in 14 days (Northfield Mutual returned Account 10-4471's remittance on day 17); a claim at day 25 with no remittance is an exception now.
• Commitments from prior touches: every §31.6 call ends with a date; the date enters the queue.
• Remittance events: a primary payment starts the secondary clock (Chapter 28 §28.10); a denial starts Chapter 29's process; an appeal submission starts Chapter 30 §30.10's tracker.

An aging report finds money at ninety days. A status event finds the same money at twenty — and Chapter 1's decay curve says those are different amounts of money.

Principle four: the queue is also a measurement, so instrument it honestly. Count dollars resolved per hour worked and touches per resolution — not accounts touched. And read Chapter 29 §29.5's four questions against it monthly: how many items, what is the oldest and why, how many closed as no-action, and what is in here that nobody in this room can fix.

⚠️ Where Claims Die

The status note that says nothing.

Open any aged account in any practice and read its history. In a queue managed by touches, you will find this, repeated monthly, sometimes for a year:

text 04/12 called payer — claim in process. F/U 30 days. 05/14 called payer — claim in process. F/U 30 days. 06/15 called payer — still in process. F/U 30 days.

Every touch was counted as work. No touch moved the claim, because no touch asked for anything a payer could commit to, obtained anything a person could act on, or escalated anything when the pattern became visible — which it did, in this example, on the second call.

A worked account and an advanced account are different things, exactly as Chapter 29 §29.7's denials-resolved-not-worked rule says — and a queue measured on touches will produce touches. The §31.6 discipline exists so that every touch either advances the account, sets a real event, or triggers the next escalation step. Three "in process" notes in a row is not follow-up. It is a metronome.

31.6 Follow-up: the call, the portal, the escalation

Follow-up is what a queue entry turns into when a person picks it up. The tools come in an escalation ladder, and using the cheap ones first is most of the efficiency.

The portal and the status transaction come first because they are nearly free. The 276/277 claim status inquiry (Chapter 27 §27.9) and the payer portal answer the first question — does the payer have this claim, and where is it — in under a minute, without hold music. Batch them: fifty status checks before lunch is a normal number. What the portal cannot do is negotiate, explain, or commit — its vocabulary is the claim status codes, and "in process" is most of what it says.

The call is expensive and therefore purposeful. A payer call costs real minutes (hold time is most of them — this is why §31.5 batches by payer, so one call carries many claims). The discipline is to arrive knowing what the portal already told you, and to leave with things only a human can give:

📞 On the Phone

A follow-up call that produces a commitment, and one that produces a metronome note.

The weak version: "I'm calling to check status on a claim." — "It's in process." — "Okay, I'll check back." That call cost eleven minutes and produced the ⚠️ note in §31.5.

The strong version, on the same claim:

"I'm calling about claim 44-31207 (a different account — constructed dialogue), date of service April 2. Your portal shows in-process since it acknowledged on April 5 — that's past your usual adjudication window. Is anything pended, and is anything needed from us?"

"It's pended for review."

"Pended for what specifically — records, coordination of benefits, an internal edit? If anything is needed from us, I can send it today."

"Internal review. Nothing is needed from you."

"Understood. What date will this adjudicate by, and what happens if it doesn't?"

"Should be within two weeks."

"So by June 2. Can I have a call reference number and your first name? … Thank you. If it hasn't adjudicated by June 2 I'll call back and ask for a supervisor with this reference."

What just happened: the claim's queue entry now carries an event (June 2), a commitment, a reference number, and a name — and the next touch is pre-authorized to escalate instead of starting over. The failure modes are all versions of accepting the first answer: "in process" is a screen, not a status; "pended" is a category, not a reason; "soon" is not a date. And write it all down — noting that Chapter 27 §27.7 ranked "a note about a call" as the weakest form of timely-filing proof. The note is not evidence for a payer dispute; it is memory for your own next touch. For evidence, get the document: ask the representative to reprocess, to send the request in writing, or to confirm on the portal where it leaves a record.

Escalation is the ladder's top, and it has real rungs. In rough order: the supervisor on the provider-services line, with your reference numbers · the provider relations representative — the payer employee whose job is the relationship with your practice, reachable outside the call center, and the right owner for a pattern ("here are 41 claims pended for internal review past your adjudication window," delivered as a spreadsheet) · project claims — most payers will take a batch file of a systemic problem rather than 41 phone calls · the contract's remedies, because Chapter 2's participation agreement usually specifies clean-claim payment timelines · and, for the genuinely stuck, state remedies: most states have prompt-pay statutes requiring insurers to pay clean claims within a defined window with interest, enforced by the state insurance department, and several accept provider complaints online. State-specific, plan-type-specific (self-funded plans under the Employee Retirement Income Security Act (ERISA) generally sit outside state insurance regulation — Chapter 2 §2.5), and worth verifying before invoked — but the existence of the ladder changes the calls below it. A practice that never escalates teaches its payers what it will tolerate.

Patient-side follow-up runs on a different ladder — statements, a call, a plan, and the collection decision — and it belongs to the next two sections and to Chapter 32, which owns the statement itself (§32.6), payment plans (§32.7), and the conversation with a frightened person (§32.10). What this section contributes is the same discipline: every patient touch should end with a date or a decision, not with "sent another statement."

31.7 Small balances, write-off thresholds, and the arithmetic of giving up

Everything in this chapter so far assumes the balance is worth pursuing. This section is about the ones that are not — and about making that call with arithmetic instead of fatigue.

Start with the number the arithmetic needs. Labor is the business office's dominant cost, and the cost of labor is not the wage. A denial specialist's wage, plus payroll taxes, plus benefits, plus the workstation, software licenses, space, and supervision that make the work possible, is the fully loaded cost — and it is the only honest rate for costing a task.

**This book's fully loaded cost of denial-management staff time — used from here through

Chapter 40 — is \$36.00 per hour, which is \$0.60 per minute.**

Constructed teaching figure: representative of a mid-range billing wage carried to a loaded rate; your practice's actual figure comes from your own payroll and overhead, and computing it takes an hour with a bookkeeper. The convenient property of \$36.00/hour is that it makes minutes legible — every minute of business-office attention costs sixty cents — and minutes are the unit this book has been recording since Chapter 24 §24.1 priced the error ladder in minutes and declined, deliberately, to convert. Chapter 29 §29.10 recorded Account 10-4471's three touches in minutes the same way. This section supplies the rate. Chapter 40 owns the assembly.

Cost to collect is the same idea at department scale: everything the collection function costs — loaded labor, statements, vendors, software, agency fees — divided by what it collects. Tier 2, ranges only: physician practices commonly measure cost to collect in the low single digits as a percent of collections; the figure varies with specialty, payer mix, and what gets counted, and HFMA's MAP Keys exist partly to standardize that last item. Verify current benchmarks before quoting one. §31.11 puts it on the dashboard — with a warning attached, because a practice can make cost to collect look better by not doing the work, which is §31.3's write-off game wearing a cost hat.

Now the arithmetic of giving up. A statement costs real money to send: production and postage (or the vendor's per-statement fee), plus the minutes of handling around it. A rebill costs minutes of research plus the transaction. A call costs minutes. Whenever the cost of the next touch exceeds what the touch can recover, the touch is a donation — and the small balance write-off exists so the donation is declined deliberately, by policy, instead of made accidentally, forever.

🧮 Run the Numbers

Three balances against the rate. (Constructed teaching figures — build your own with your own costs.) Assume a statement cycle costs **\$2.95** — \$1.75 production and postage plus 2 minutes of handling at \$0.60 = \$1.20 — and that a small patient balance typically takes about two cycles to pay when it pays at all.

```text BALANCE 1 — patient owes 4.15 (coinsurance rounding residue) expected cost to pursue: 2 statement cycles × 2.95 = 5.90 5.90 > 4.15 ►► WRITE OFF AT POSTING, under policy

BALANCE 2 — patient owes 14.85 2 cycles × 2.95 = 5.90 · 14.85 − 5.90 = 8.95 margin ►► PURSUE — two cycles. THEN STOP AND RE-DECIDE: each FURTHER touch is a fresh 2.95-or-more against whatever remains, and sunk cost is not an argument

BALANCE 3 — payer paid 2.40 under the expected allowed research: 4 min × 0.60 = 2.40 — a wash BEFORE the rebill transaction costs anything ►► DO NOT REBILL THE INSTANCE. CLASSIFY IT FIRST: the same 2.40 variance × 1,150 occurrences a year = 2,760.00 — a CONTRACT-LOADING QUESTION (Ch. 28 §28.8), pursued once, at the pattern level ```

Checks: 1.75 + 1.20 = 2.95 ✓ · 2 × 2.95 = 5.90 ✓ · 14.85 − 5.90 = 8.95 ✓ · 4 × 0.60 = 2.40 ✓ · 2.40 × 1,150 = 2,760.00 ✓

The threshold falls out of the arithmetic: if two cycles cost \$5.90, a patient-balance write-off threshold around \$5.00 declines the guaranteed-loss pursuits automatically. Set it from your own costs, round it, write it into policy, apply it uniformly, and revisit it when postage or wages move. The exact number matters much less than that it exists, is written, and is applied without discretion — discretion is how thresholds become favors.

And Balance 3 carries the section's most important rule: give up on instances, never on patterns. The write-off code must say small balance (Chapter 28 §28.6 built the category), and somebody must read the category's total by payer and by reason — because a small balance repeated a thousand times is not a small balance, and the only measurement that can see it is the one §28.8 built. The arithmetic of giving up applies to the account in front of you. It never applies to the category.

⚖️ Compliance Check

A small-balance policy touches federal law in two places, and both are about uniformity.

Patient cost sharing on federal program claims is not yours to waive as a habit. Routinely writing off Medicare copayments and coinsurance — or advertising that you will — can constitute an inducement to the beneficiary and misstate your actual charge, implicating the Anti-Kickback Statute and the Civil Monetary Penalties Law's beneficiary-inducement provisions (Chapter 5 §5.4, §5.5). What keeps a threshold policy defensible is what makes it good practice anyway: it is uniform (all payers, all patients), cost-justified (the arithmetic above, on file), modest, unadvertised, and it coexists with genuine collection effort above the threshold and documented financial-hardship handling (§31.10) below none.

And a "small balance" that is a small CREDIT is not yours to keep at any threshold. Writing off a \$3.40 balance you cannot collect is a business decision. Writing off a \$3.40 balance you OWE — absorbing a small overpayment because refunding it costs more than \$3.40 — is keeping other people's money, and on federal program accounts §31.9's sixty-day rule does not have a de minimis exception you get to invent. Payer contracts and state unclaimed-property law constrain the commercial and patient sides too. Route small credits through §31.8's process like large ones; the refund transaction's cost is not a defense.

Requirements change and vary by program, state, and contract. Have the compliance officer review the written threshold policy — the policy's existence in writing is itself part of the defense.

31.8 Credit balances and unapplied cash

Run the aging report from §31.2 again and look at the memo line: \$28,600 in credit balances. This section is about what that number is, why it is dangerous out of all proportion to its size, and the two housekeeping failures — credits and unapplied cash — that corrupt every metric upstream of them.

A credit balance is an account balance below zero: the account has received more in payments and adjustments than it was ever charged. How does that happen to a system that starts every account at a positive charge? Almost always one of five ways:

1. Two payers both paid as primary — a coordination-of-benefits failure (Chapter 2 §2.8), or a crossover claim that also went out as a paper secondary (Chapter 28 §28.10).
2. The patient paid and then the plan did too — a point-of-service collection (Chapter 24 §24.8) followed by adjudication that assigned less patient responsibility than was collected; Chapter 1's ⚠️ about collecting against the charge instead of the allowed amount manufactures these at scale.
3. A payment posted to the wrong account or line — in which case the credit is a phantom, and somewhere else a real balance is being chased that does not exist.
4. An adjustment posted twice — autoposting plus a manual touch (Chapter 28 §28.7).
5. A payer reprocessed a claim and the reversal half of the pair never posted (Chapter 28 §28.9).

Note what that list means: a credit balance is a symptom, not a diagnosis. Numbers 3 and 4 are bookkeeping errors — nobody is owed anything; the fix is a corrected posting. Numbers 1, 2, and 5 are real money that belongs to someone else — a payer or a patient — and the moment you know which, you are holding an identified obligation. The work of a credit-balance queue is exactly that triage: posting error, patient refund, or payer refund — decided from the documents, before any money moves. Refunding first and researching second sends checks to the wrong parties and turns one error into two.

📋 Read the Chart

Source: month-end credit balance report, the same practice (constructed teaching example) What it says:

```text CREDIT BALANCE REPORT — 214 accounts, total (28,600)

BY APPARENT CAUSE (after first-pass review) posting error / misapplied ............ ( 9,300) patient overcollection ................ ( 7,850) payer duplicate / COB ................. ( 8,900) unresolved — needs research ........... ( 2,550) ──────── (28,600)

AGE OF OLDEST CREDIT ..................... 311 days FEDERAL-PROGRAM ACCOUNTS IN TOTAL ........ ( 6,140) OLDEST FEDERAL-PROGRAM CREDIT ............ 194 days ```

Check: 9,300 + 7,850 + 8,900 + 2,550 = 28,600 ✓

What it means: about a third of this is not money owed to anyone — it is posting hygiene, and clearing it also clears the phantom debits it created elsewhere. About \$16,750 is real refund obligation, roughly half to patients and half to payers. And two lines on this page are compliance findings, not housekeeping findings: \$6,140 of it sits on federal-program accounts, and the oldest of those credits is 194 days old. If any part of that \$6,140 is an identified overpayment, §31.9's sixty-day clock did not wait for this report to be run. A credit-balance report is the only place in the practice where an aging number can be a legal exposure — which is why §31.11 puts the oldest federal credit's age on the dashboard itself.

What it doesn't say: whether anyone is working it, which is the point of routing credits to a dedicated queue with its own cadence — weekly, not "when we get to it" — worked by someone with posting-correction rights and a direct line to the compliance officer for anything federal.

Unapplied cash is the mirror problem: money that arrived and never found its account. A patient pays online with a miskeyed account number; a payer's electronic funds transfer (EFT) lands without its electronic remittance advice (ERA) because the practice enrolled for one and not the other — Chapter 27's three separate enrollments, arriving as a bank deposit nobody can post; a paper check references a patient the system spells differently. The money sits in a suspense account — real cash, banked, and invisible to every account it belongs to.

The damage is not the cash — the cash is safe. The damage is everything the system believes while the cash sits unapplied. The accounts it belongs to still show open balances: statements go to patients who have already paid — the single most goodwill-destroying document a practice can mail — follow-up staff chase payers who have already paid, and every AR metric in §31.3 and §31.4 reads high. Unapplied cash is a small number that lies through every other number. Dashboard treatment (§31.11): total and oldest item, target near zero, reviewed weekly. The fix is almost always upstream — complete the ERA/EFT enrollments so remittances arrive with their money, and put an account number on every patient payment channel.

31.9 Overpayments, recoupments, and the sixty-day rule

Chapter 5 §5.1 stated the legal frame in one sentence: retaining a known overpayment is itself a False Claims Act violation. This section is the operational side — how overpayments surface, what "identified" means, what the clock requires, and who moves first.

An overpayment is money received from a payer or patient to which the practice is not entitled — after all the ordinary reconciliation of contractual adjustments and cost sharing. Duplicate payments, payment for another provider's patient, payment above the contract, payment for a service later found not to have been documented or covered, the credit-balance causes in §31.8 that turned out to be real. The category is legal, not clerical: a credit balance is what your ledger shows; an overpayment is what the law sees once you know why.

The sixty-day rule, operationally. Under the Affordable Care Act's overpayment provision, a Medicare or Medicaid overpayment must be reported and returned within sixty days of being identified (or by the date any corresponding cost report is due, where that applies). A retained overpayment past that deadline becomes an "obligation" whose knowing concealment or avoidance is actionable under the False Claims Act's reverse-false-claim provision — with everything Chapter 5 attached to that statute. The operational content is in three words:

"Identified" is not "a letter arrived." Under the Centers for Medicare & Medicaid Services (CMS) rule implementing the provision, a provider has identified an overpayment when it has determined — or should have determined, exercising the diligence the current rule requires — that it received one, and quantification is part of the exercise, not an excuse to postpone it indefinitely. The practical consequence: credible information that an overpayment probably exists starts an obligation to investigate promptly, and the investigation's reasonable duration is bounded, not open-ended. The regulation's exact standard has been revised since first issued and this book will not freeze it — verify the current text of 42 CFR 401.305 with your compliance officer — but no version of it has ever rewarded the practice that declined to look. A lookback period measured in years attaches to identified overpayments; verify the current span the same way.

Report and return has real mechanics. The return half is a refund with documentation — to the Medicare Administrative Contractor (MAC) on its voluntary refund/overpayment process, or per the state Medicaid program's process. The report half matters too: an overpayment that came from conduct — a coding pattern, a billing configuration, the kind of thing Chapter 21's Account 31-2245 became — may belong in a self-disclosure rather than a quiet refund, which changes the protocol, the protections, and who signs. Chapter 37 §37.9 owns self-disclosure; the operational rule here is only this: the moment an overpayment looks systemic rather than clerical, stop, preserve the analysis, and involve compliance and counsel before money or paper moves.

Recoupment is the payer-initiated version, and Chapter 28 §28.9 showed its mechanism: a demand letter, then an offset — the payer deducting the overpayment from unrelated future remittances, with a provider-level adjustment line referencing the original claim. Operationally:

   A RECOUPMENT DEMAND, WORKED
   [constructed teaching sequence — Medicare shape; timelines
    are illustrative, verify current rules and your contract]

   1  READ IT.  Which claims, which reason, how much, what
      deadline, what appeal rights. A demand letter is a
      DETERMINATION — it can be right, wrong, or partly both.
   2  VERIFY IT.  Pull the claims and the remittances. Payers
      recoup in error too (Ch. 28 §28.8's stale-fee-schedule
      contract was found by exactly this reading).
   3  DECIDE, ON THE MERITS AND THE CLOCK.
        agree ......... repay or let the offset run; POST IT
                        RIGHT (to the original claims — never
                        as a contractual haircut on the
                        unlucky remittance it rode in on)
        disagree ...... APPEAL (Ch. 30). On Medicare, filing
                        the first-level appeal within the
                        short window stated in the demand
                        pauses recoupment at the first two
                        levels — a deadline measured in days,
                        not the appeal's own longer limit
   4  READ THE PATTERN.  One demand is an account event.
      Forty demands with one reason are Chapter 37 arriving
      early — and possibly §31.9's own reporting obligation
      for the claims the payer has NOT found yet.

Two postings rules keep the books honest through all of this. Post recoupments and refunds to the claims they belong to, so the account history stays true (Chapter 28 §28.9's warning about posting the net). And date-stamp identification: the day the practice concluded an overpayment exists is a fact the sixty-day rule turns on, and a contemporaneous record of it — what was found, when, how quantified, when repaid — is both the compliance file and the proof of good faith.

⚖️ Compliance Check

The sixty-day rule inverts the business office's instincts, and the inversion must be taught explicitly.

Everywhere else in this book, money in the door is the goal and speed serves collection. Here, money in the door is the hazard, and speed serves return. A credit-balance backlog that would be mere sloppiness in any other industry is, on federal-program accounts, a queue of potential False Claims Act obligations aging toward a deadline — at treble damages and per-claim penalties, with qui tam relators (Chapter 5 §5.3) among the people best positioned to notice. The first False Claims Act case built on the sixty-day rule involved exactly this shape: overpayments flagged internally, and a repayment effort a court found too slow and too passive after the flag — this chapter's Case Study 1 walks through it.

The defensive posture is boring and completely effective: a credit-balance queue with a weekly cadence and federal accounts worked first (§31.8) · a written identification-to-refund procedure with the sixty-day clock stated in it · date-stamped identification records · a standing rule that systemic findings go to compliance before anything else happens · and a refund process that is easy, because a refund process with friction is a backlog generator. Requirements change — the identification standard itself has been revised — and Medicaid specifics vary by state. Verify with the compliance officer and current CMS guidance; do not rely on this or any summary.

31.10 Bad debt, charity care, and the difference

At the end of every patient-side pursuit — statements sent, calls made, plans offered — a balance remains, and the practice must write it off as one of two things. The two things are not interchangeable, and the difference is not bookkeeping.

BAD DEBT is a balance the patient could have paid, and after genuine collection effort, did not. It is a cost of doing business, written off as uncollectible, possibly placed with an agency.

CHARITY CARE is a balance the patient could not pay, forgiven under a financial assistance policy, and it should never have been pursued as if it were the first kind.

The classification is a determination about the patient's circumstances — which means it requires information, which means somebody has to ask. That is the operational heart of this section: screening for financial assistance must sit upstream of collection activity, not downstream of its failure. A practice that pursues first and classifies at the write-off has already sent statements, made calls, and possibly placed accounts against people its own policy says should have been forgiven — and it will have done so disproportionately to the patients least equipped to navigate the paperwork that would have protected them. Chapter 1's commitment stands here or nowhere: most American medical debt is held by people who were insured and did everything right, and a balance is not a character problem.

For tax-exempt hospitals this is federal law, not just decency. Internal Revenue Code §501(r) requires a written financial assistance policy (FAP), publicized plainly, and — the operational teeth — forbids extraordinary collection actions (reporting to credit bureaus, selling debt, lawsuits, liens, garnishment) until the hospital has made reasonable efforts to determine FAP eligibility, with defined notice periods and a months-long window in which a patient may still apply. Physician practices are outside §501(r) but inside its logic; state charity-care laws reach further in some states — several require screening before any collection activity — and are state-specific; verify. Chapter 32 §32.8 owns the assistance policies themselves and presumptive eligibility (approving assistance from data when the patient never manages to apply); this section's contribution is the sequencing rule: determine, then pursue — never the reverse. Case Study 2 is what the reverse looks like at system scale, in the public record.

Why the classification also matters to the numbers. The two write-offs mean opposite things on every report that contains them. Bad debt measures collection performance and payer mix; charity care measures community benefit — hospitals report it publicly, and it figures in what tax exemption is exchanged for. Blend them and both numbers lie: charity buried in bad debt makes collections look weak and understates community benefit; bad debt dressed as charity does the reverse and can misstate cost-report figures where those apply. Separate write-off codes, separate reports, separate meanings — Chapter 28 §28.6's principle that the adjustment code is a decision, applied to the two biggest patient-side categories there are.

Collection agency placement is the bad-debt path's last step, and it is a decision with machinery. What placement actually is: the practice engages an agency — typically on contingency, commonly for a third or so of recoveries; ranges vary, verify — and hands it accounts to pursue. Operational rules that survive contact with reality:

• Placement is delegation, not disposal. The agency acts for the practice; under the Fair Debt Collection Practices Act (FDCPA) and state law its conduct reflects on and can implicate the practice; and the practice sets the rules — whether to credit-report, whether litigation is ever authorized, which accounts come back on hardship. Choose an agency the way you would choose an employee who will talk to your patients about money.
• Screen before placing — every account, against the FAP, every time (this section's rule), and pull back any account where new hardship information surfaces.
• Reconcile monthly. Agency-reported balances drift from practice systems; patients pay the practice directly on placed accounts; recalls happen. An unreconciled agency file eventually bills a paid balance, which is §31.8's worst phone call with a collector's voice.
• And know the current reporting landscape before assuming a credit report is leverage. The national credit bureaus have removed paid medical collections and small-balance medical collections from reports and lengthened the reporting delay; further federal rulemaking on medical debt has been in motion and litigated. The details move — verify — but the direction has been consistent: medical debt is a weakening threat and was always a poor collection tool. Chapter 32 §32.9 does the reputational arithmetic in full.

📞 On the Phone

"I just can't pay this."

Four words, on a follow-up call about a \$460.00 balance, and what happens next depends entirely on what the person on the practice's end has been trained to hear.

Heard as a refusal, it routes to the bad-debt path: firmer letters, the agency, a write-off that closes the account and the relationship.

Heard as information, it is the opening of a financial assistance conversation:

"Thank you for telling me — that changes what happens next, and there are options. We have a financial assistance policy; a lot of people qualify and don't know it. Can I send you the application, or take the basics now? And separately from that, if any part of the balance stands, we can set up a payment plan that actually fits — Chapter 32 has the details of what we can offer. What I don't want is for this to sit and turn into collection letters nobody wants."

The failure mode is the script that treats hardship disclosure as a stalling tactic — because some fraction of the time it is, and a policy tuned to that fraction pursues the many to catch the few. The practice's protection against being gamed is not suspicion on the phone; it is the application — documented criteria, uniformly applied (§31.7's compliance logic again). The person who says "I can't pay" and the person who says nothing and lets statements age into §31.4's pre-collection pool may be in identical circumstances. The first one handed you the information. The screening rule exists for the second one.

31.11 The dashboard a manager should actually look at

Every number this chapter has built can be gamed, and every one has a blind spot the chapter has named. That is not a counsel of despair — it is the design specification for the dashboard. A useful AR dashboard is not many numbers; it is few numbers, each carrying its own definition and each paired with the number that catches its blind spot.

Chapter 29 §29.7 supplied the governing rule and Account 10-4471 supplied the proof: the same account is a 25% denial rate by line and 100% by claim. A rate without its definition is not information. This dashboard writes the definition into the row.

📋 Read the Chart

Source: one-page monthly AR dashboard, the practice from §31.2–§31.8 (constructed teaching example — the format is the teaching point; the targets are illustrative and yours come from your own baseline)

```text AR DASHBOARD — MONTH END defn this last 12-mo note mo mo trend ── SPEED ───────────────────────────────────────────────────── Days in AR ....................... [1] 51.5 53.2 ↘ AR over 90, % of AR .............. [2] 17.2% 18.1% ↘ of which UNEXPLAINED, $ ........ [2] 12,700 16,900 ↘ ── YIELD ───────────────────────────────────────────────────── Net collection rate, 12-mo ....... [3] 96.1% 96.0% → Underpayment variance found, $ ... [3] 2,760 890 ↗ ── DEFENSE ─────────────────────────────────────────────────── Initial denial rate (lines) ...... [4] 6.4% 6.6% ↘ Rejection rate (claims) .......... [4] 1.9% 2.0% → Preventable admin write-offs, $ .. [5] 3,180 4,415 ↘ ── OBLIGATIONS ─────────────────────────────────────────────── Credit balances, total $ ......... [6] 28,600 31,250 ↘ Oldest FEDERAL credit, days ...... [6] 194 171 ↗ ⚠️ Unapplied cash, $ / oldest days .. [7] 4,120/44 6,300/61 ↘ ── COST ────────────────────────────────────────────────────── Cost to collect, % of collections [8] 3.4% 3.3% →

DEFINITIONS ON THIS PAGE (they do not change silently) [1] gross AR, credits NOT netted ÷ trailing-365-day avg daily charges. Write-offs this month, by category, are on page 2 — READ THEM TOGETHER. [2] aged from date of service; decomposed page 2 (§31.4); "unexplained" = no activity or plan on record [3] payments ÷ (charges − contractual adj), trailing 12 mo, per Ch. 23 §23.10 — RISES on silent underpayment, so it is never read without the variance line (Ch. 28 §28.8) [4] denial rate: denied LINES ÷ lines ADJUDICATED, zero-pay, first submission (Ch. 29 §29.7). Rejection rate beside it because rejected claims never deny (Ch. 27 §27.7) [5] Ch. 28 §28.6's category: our own process failures, in $ [6] gross, own report (§31.8); federal age vs the §31.9 clock [7] suspense total and age of oldest item (§31.8) [8] all collection-function cost ÷ collections, 12-mo ```

What it means — and it takes five minutes, which is the point: speed is improving and the improvement is real, because the unexplained over-90 pool is draining while write-offs (page 2) are flat. Yield looks flat at 96.1% — and the variance line beside it says underpayment findings tripled, which means some of that flat 96.1% is §28.8 clawing back money the ratio itself could not see. Defense is improving on both of its paired lines at once, which is the only way an improvement there can be trusted. The one deteriorating number on the page is wearing the warning flag: the oldest federal credit is 194 days and aging — a §31.9 clock, not a metric — and it, not the seventeen-point-two, is this month's management action.

What it doesn't say: why. A dashboard locates problems; it never diagnoses them. The diagnosis lives in page 2's decompositions and, past that, in the rows.

Five disciplines make a page like that work, and all five are inherited from earlier chapters:

Every rate carries its definition, in writing, on the page — Chapter 29 §29.7's denominator problem, solved by typography. When a definition must change, both versions run in parallel across the seam (§31.3's ⚠️).

Every metric is paired with its blind-spot's detector. Days in AR travels with the write-off report (§31.3). Net collection rate travels with underpayment findings (Chapter 23 §23.10, Chapter 28 §28.8 — cited, not rebuilt: a silent underpayment raises the ratio, so the ratio alone can only ever deliver good news). Denial rate travels with rejection rate (Chapter 27 §27.7). Collections travel with preventable write-offs (Chapter 28 §28.6). An unpaired metric is an invitation to optimize the visible half of a tradeoff.

Trend against yourself, not against benchmarks. Chapter 29 §29.9's argument, unchanged: an external figure was computed under somebody else's definitions; your own trailing twelve months, under a frozen definition, is the comparison that means something.

Test one story a month against the rows. Chapter 28's Case Study 1 is the standing warning — a number that moved, was reviewed, was discussed, and was assigned a plausible cause nobody tested. "A number with a story attached stops being a question." The countermeasure is mechanical: each month, pick one metric's explanation and pull forty rows behind it (Chapter 29's Case Study 1 proved the forty-row afternoon works, and who is motivated to run it). Most months the story survives. The month it does not pays for the year.

And a dashboard is not a control. A person who reads the dashboard is a control — the lineage runs from Chapter 27's report to Chapter 29's queue to here, and it is the reason this section's title says look at. A dashboard nobody interrogates is Chapter 27's acknowledgment report delivered faithfully to a mailbox nobody owned — the numbers improving, honestly reported, for reasons nobody would have liked.

31.12 🗂️ The Encounter — where the \$185.00 sat for 49 days

Account 10-4471, through this chapter's lens: not one claim moving forward, but one line standing still.

   LINE 1 (99214-25, charge 185.00) IN ACCOUNTS RECEIVABLE

   day 0–2  charge captured, claim built and submitted —
            185.00 is in AR at charges
   day 17   remittance: CO-45 posts 56.60 contractual;
            CO-97 denies the line ► 128.40 sits, DENIED,
            expected value, appeal pending
   day 24   appeal submitted (Ch. 30) — status, not motion
   day 31   line 1 ages into the 31–60 bucket
   day 45   queue event fires: follow-up call on the
            pending appeal (touch 3 begins its work)
   day 59   payer decision letter: overturned
   day 61   line 1 ages into the 61–90 bucket
   day 66   second remittance posts 98.40 + 30.00 copay
            (collected day 0) ► line 1 leaves AR
                                ────────────────────────
            DENIED ON DAY 17, RESOLVED ON DAY 66:
            49 DAYS IN THE AGING as a denied receivable

Check: 66 − 17 = 49 ✓ · 128.40 = 98.40 + 30.00 ✓

What the aging report saw, and what it did not. On the March month-end report (day 17), line 1 sat in 0–30 — invisible, normal. On the April report (day 47), it sat in 31–60 — one line among dozens, indistinguishable from a slow payer or a forgotten claim. It grazed 61–90 for five days and left. At no point did the aging report flag this line, and at no point should it have — the aging report is a lagging indicator that would first have raised its hand around day 107, long after the money arrived. What actually surfaced the line, twice, was the work queue: the denial event on day 17 routed it into Chapter 29's process (touch 1, day 20), and the appeal-tracking event — follow-up at day 45, decision due — kept it owned while it aged (Chapter 30 §30.10's tracker doing §31.5's job). The bucket told the truth slowly. The queue told it in time. That is this chapter's whole argument in one line item.

What the 49 days cost. Three things, and this book prices each in its own currency:

• Staff attention: 58 minutes across three touches — published, in minutes, in Chapter 29 §29.10. As of this chapter, the rate that converts business-office minutes to dollars is also in print: §31.7's \$36.00 per hour, \$0.60 per minute. This checkpoint states the minutes and the rate and deliberately does not multiply them — that assembly, and the conclusion it feeds, is Chapter 40's.
• Time value: \$128.40 arrived 49 days after it was denied and 64 days after the claim went out. No interest attached — the payer adjudicated and then re-adjudicated within its windows, so no prompt-pay remedy (§31.6) applies. The delay simply transferred float from the practice to the payer, invisibly, the way §31.1's decay curve says delay always does.
• And nothing else — no timely filing risk (the appeal was filed on day 24 against a much longer window), no re-aging games (the line aged honestly from date of service), no write-off pressure (128.40 sits far above any §31.7 threshold). This is what a denial looks like when the machine works. The 49 days were not a failure of follow-up; they were the price of the appeal path itself — which is a sentence Chapter 40 will weigh.

Meanwhile, the patient's side of the account ran this chapter's other lesson. The \$17.58 coinsurance became patient responsibility on day 17 but was not billed until day 70 — the practice holds statements until the account fully resolves, so the patient sees one honest number instead of a moving one (Chapter 28 §28.11's credit-shown statement; Chapter 32 §32.6 takes it from here). Aged from date of service, that \$17.58 was 100 days old when it was paid — deep in §31.4's "over-90" territory — yet the patient paid within 30 days of the first statement. Date-of-service aging measured the practice's sequencing choice and called it patient slowness. Every aging number is an answer to "aged from when?" — and this account closes the chapter by demonstrating the question twice, once on each side of the ledger.

Open questions. Q4 — could the denial have been prevented? — remains open, and remains Chapter 40's. What this chapter changed is that the last of Chapter 40's inputs is now in print: the minutes (Chapter 29), the overturn rate (Chapter 29), the allowed amount (Chapters 2, 23, 28), and now the rate (§31.7). The file is complete except for the arithmetic, and the arithmetic is the capstone's.

Summary

Accounts receivable is a portfolio, not a pile. Gross AR at charges is inflated by contractual air that no one will ever collect; adjudicated AR is real money with a decay rate. The first cut is always insurance AR versus patient AR — two kinds of money, two workflows, two clocks — and Account 10-4471 carried both at once: \$128.40 on appeal and \$17.58 awaiting a statement, inside one \$145.98 balance.

The aging report sorts open balances into thirty-day buckets, and three build-choices decide what it shows: aged from what date (date of service is honest; re-aging on touches makes worked-badly look young), split by what (insurance/patient, then payer), and whether credits are netted (never — old credits camouflage old debits and are themselves the compliance problem). Read the report as flow: this month's 31–60 is last month's undrained 0–30.

Days in AR = total AR ÷ average daily charges, and every term is a choice. One constructed practice produced 51.5, 48.0, 48.4, and 43.1 in one afternoon by netting credits, moving the denominator window, and writing off aged denials — the last being the metric's signature failure: it cannot tell collection from surrender. Write the definition on the dashboard; disclose write-offs beside the metric; trend against yourself. Benchmarks (days in AR under 40 common, under 35 strong — Tier 2, verify) are for questions, not victory laps.

AR over 90 is the fossil detector, and its value is the decomposition. Of the practice's \$70,850 over ninety, only \$12,700 was the real finding — untouched insurance money losing rights — while payment plans, a documented payer project, and pending appeals explained the rest. Old-and-known is fine. Old-and-unexplained is the only bad old. A low over-90 can be the sound of aggressive abandonment; a rising one can be the sound of appeals worth filing.

The work queue finds money by subtracting what is not money (credits out, corpses closed and counted, unownable items routed), then sorting by what destroys value fastest: deadline, event due, category, expected value — allowed, never charge. Feed it events — acknowledgments, adjudication windows, call commitments, remittance triggers — because an aging report finds money at ninety days and a status event finds the same money at twenty. Measure dollars resolved per hour, not touches: three "in process" notes in a row is a metronome, not follow-up.

Follow-up is a ladder: portal and 276/277 first (free, shallow), then the call that leaves with a specific pend reason, a date, a reference number, and a name — then escalation with real rungs (supervisor, provider relations with a spreadsheet, project claims, the contract, state prompt-pay remedies — state- and plan-type-specific, verify). A note about a call is memory, not evidence (Chapter 27 §27.7); for evidence, get the document.

§31.7 published the number this book has been withholding: fully loaded denial-management staff time costs \$36.00 per hour — \$0.60 per minute (constructed; derive your own). Against it, the arithmetic of giving up: a \$2.95 statement cycle makes a \$4.15 balance a guaranteed loss and justifies a written, uniform, cost-derived small-balance threshold near \$5.00 — applied to instances, never patterns: a \$2.40 variance times 1,150 claims is a \$2,760.00 contract question (Chapter 28 §28.8), pursued once at the pattern level. Federal cost sharing is not yours to waive as a habit, and a small credit is not yours to keep at any threshold.

Credit balances are symptoms with three diagnoses — posting error, patient refund, payer refund — triaged from documents before money moves, on a dedicated weekly queue, federal accounts first. Unapplied cash is banked money lying through every other number, generating statements to people who already paid; fix the enrollments upstream and keep the suspense account near zero.

An overpayment is a legal category with a clock: identified federal overpayments must be reported and returned within sixty days, or they become False Claims Act obligations. "Identified" has never rewarded not looking (verify the current 42 CFR 401.305 standard); date-stamp identification; make refunds frictionless; route anything systemic to compliance and Chapter 37 §37.9's self-disclosure analysis before money or paper moves. Recoupment demands are determinations — read, verify, then repay or appeal (the Medicare appeal filed fast pauses the offset), post to the original claims, and read forty demands as one pattern.

Bad debt is could-pay-didn't; charity care is couldn't-pay — and the determination must precede collection activity, not follow its failure. §501(r) makes that sequencing federal law for tax-exempt hospitals; decency and several states' laws extend it further. Separate write-off codes, separate meanings. Placement is delegation, not disposal — screen every account against the FAP first, reconcile monthly, and do not mistake a weakening credit-report threat for a collection strategy (Chapter 32 finishes this).

And the dashboard: few numbers, definitions printed on the page, every metric paired with its blind spot's detector — days in AR with the write-off report, net collection rate with underpayment findings (it rises on silent underpayment — Chapter 28 §28.8, cited not rebuilt), denial rate with rejection rate, collections with preventable write-offs, credits with the age of the oldest federal item. Trend against yourself. Test one story a month against forty rows. A dashboard is not a control; a person who reads it is.

Line 1 was denied on day 17 and resolved on day 66 — 49 days in the aging — surfaced both times by the queue and never by the bucket. The minutes are published (58, Chapter 29). The rate is now published (\$0.60, §31.7). Chapter 40 multiplies.

Key Terms

Accounts receivable aging · aging report · aging bucket · re-aging · insurance AR · patient AR · days in AR · AR over 90 · follow-up · escalation ladder · small balance write-off · write-off threshold · fully loaded cost · cost to collect · credit balance · unapplied cash · overpayment · recoupment · 60-day rule · bad debt · charity care · financial assistance screening · collection agency placement · AR dashboard · metric pairing

Spaced Review

From Chapter 28 §28.6, §28.8, §28.9 — the write-off code is a decision; the underpayment method (expected allowed, actual, variance, threshold, classify) and the fact this chapter's dashboard is built on: a silent underpayment raises the net collection rate. And the offset's reference names a claim — this chapter's §31.9 is where that finding acquired a legal clock.

From Chapter 29 §29.5, §29.7 — the queue sort (deadline first, because time is what converts recoverable to unrecoverable) generalized from denials to all of AR; and the denominator problem: Account 10-4471 is 25% by line, 100% by claim, which is why every rate on §31.11's dashboard carries its definition in print.

From Chapter 27 §27.6, §27.7 — the acknowledgment trail that lets §31.5's queue find a lost claim at day 5 instead of day 50, and the ranked proof of timely filing that decides what a follow-up call's documentation is actually worth.

From Chapter 23 §23.7, §23.10 — charges are arbitrary, which is why the queue sorts on allowed and why gross AR and the gross collection rate flatter nobody honestly; and the two collection ratios this chapter chained to their blind-spot detectors.

From Chapter 24 §24.8 — point-of-service collection done against the allowed amount, not the charge: the cheapest prevention for §31.8's second-largest credit-balance cause.

Coming up: Chapter 32 turns to the patient side this chapter kept deferring — the estimate, the statement (\$47.58, day 70, with the \$30.00 credit shown), payment plans, financial assistance and presumptive eligibility, the collections rules, and how to explain a bill to a person who is frightened.