A practice owner showed us his monthly dashboard: days in AR, 41. Total AR over 90 days, 19%. Both numbers had been roughly flat for a year, and he took that as a good sign. When we split the same AR by payer, one Medicaid managed care plan had 58% of its balance over 90 days and a commercial payer had almost nothing past 60. The blended number was an average of a payer that was fine and a payer that was broken, and the average looked fine.
Total AR is a temperature reading. AR aging by payer is the diagnosis. Almost every practice management system can produce the second report, and almost nobody looks at it monthly. This is how to read it.
Key takeaways
- Blended AR metrics average a healthy payer with a broken one. Read aging by payer class and then by payer, with claim counts alongside dollars.
- Age by date of service, insurance AR only, in the standard five buckets. Compare each payer to its own norm, not to the practice average.
- Sample 20 aged claims per troubled payer and classify each as payer-side or practice-side. Most aged AR is practice-side and fixable this week.
- The report earns its place only if it produces a Monday worklist sorted by payer, dollar value and age, with a documented decision on every claim over 120 days.
Build the report correctly
Set the report to insurance AR only, excluding patient responsibility, which ages for entirely different reasons and belongs in its own report. Age by date of service, not by last billed date; re-billing a claim should not make it look young again. Group by payer class first (Medicare, Medicare Advantage, Medicaid, Medicaid managed care, commercial, workers compensation, self-pay), then by individual payer within class. Use the standard buckets: 0 to 30, 31 to 60, 61 to 90, 91 to 120, over 120 days.
Two columns most practices leave off: claim count per bucket, and average balance per claim. Fifty claims at $80 in the over-120 bucket is a different problem from four claims at $1,000. The first is usually a systematic denial (a taxonomy code, an enrollment record, a modifier rule) that one correction fixes in a batch. The second is four individual appeals, each needing records and a letter.
Include credit balances as a separate line rather than netting them against debits. A payer with $40,000 in open claims and $9,000 in credits shows $31,000 net, which understates the follow-up work and hides refund obligations that carry their own deadlines.
What each bucket should look like
| Bucket | What belongs here | A reasonable share of a payer's AR |
|---|---|---|
| 0 to 30 days | Claims in normal adjudication | The majority; 55% to 70% for a clean payer |
| 31 to 60 days | Slow adjudication, pended claims, first rejections being reworked | 15% to 25% |
| 61 to 90 days | Denials being appealed, claims needing records, payer delays | Under 10% |
| 91 to 120 days | Second-level appeals, unresolved denials, unworked follow-up | Under 5% |
| Over 120 days | Mostly unworked or unresolvable claims; approaching or past timely filing | Under 5%, and shrinking |
These ranges shift by payer class. Medicare pays most clean claims within 14 to 30 days, so Medicare AR should be concentrated in the first bucket, and anything over 60 days is worth a look. Medicaid managed care plans and workers compensation carriers are legitimately slower, and the same 61 to 90 bucket may be normal for them. The point of reading by payer is to compare each payer to its own norm, not to a blended one.
Separating payer slowness from practice slowness
An aged balance has one of two causes. Either the payer has the claim and has not acted, or the practice has the claim and has not acted. The report does not tell you which; the claim status does. Take a sample of 20 claims from a payer's 91-plus buckets and classify each one:
- Payer has it, pending or in review: payer slowness. Escalate through provider relations if the payer exceeds its prompt-pay obligation.
- Payer denied it and the denial is unworked: practice slowness. This is the most common finding.
- Payer never received it (rejected at the clearinghouse, never resubmitted): practice slowness, and possibly a timely filing loss.
- Payer paid it and the payment is unposted or misapplied: a posting problem, not an AR problem.
In our experience the sample from a troubled payer is usually 70% or more practice-side. That is good news, because practice-side problems are fixable this week. Payer-side problems need a different tool: most states have prompt-pay laws that require clean claims to be paid within a set period, commonly 30 days for electronic claims and 45 for paper, with interest owed on late payment. A payer whose 61-plus buckets are full of pended clean claims is a letter to provider relations citing the statute, and if that fails, a complaint to the state insurance department. Practices rarely use this, and payers know it.
The numbers to watch monthly
For each payer with meaningful volume, the monthly package should carry four figures. Add one row for the whole practice so the owner still gets the headline, but present the payer rows first.
| Measure | How to read it | What a bad trend usually means |
|---|---|---|
| Percent of AR over 90 days | Compare to the payer's own last six months, not to other payers | Rising three months running: a payer processing change or an enrollment problem. High and flat: follow-up staffing |
| Average days to first payment on clean claims | From submission to first remittance, claims without rejections or denials | Lengthening: the payer changed something, or claims are going out late |
| Count of claims over 120 days | Absolute count, not dollars | Growing: nobody is making decisions on old claims |
| Denial rate on first submission | Denied claims as a share of claims adjudicated in the month | Jumping for one payer: a policy change worth reading about |
A payer whose over-90 share rose three months in a row is a conversation with provider relations. A payer whose over-90 share is high and flat is a follow-up staffing problem. The two look identical in the blended number and need opposite fixes.
From report to worklist
The aging report is only useful if it produces a list someone works. Each Monday, export claims over 45 days with no payer response and claims with a denial and no follow-up action logged, sorted by payer and then by dollar value descending. Assign by payer, so one person learns each payer's portal, phone tree and appeal quirks. Work the highest-value claims first and the oldest second; a $2,400 surgery denial at 62 days beats a $70 visit at 118 days.
Set a rule for the over-120 bucket: every claim gets a documented decision within the month. Appeal, resubmit with proof of timely filing, bill the patient where permitted, or write off with a specific reason code. The bucket should be emptying, not accumulating. A write-off with a reason code is not a failure; it is information. Twenty timely filing write-offs from one payer in a quarter tells you exactly where the process broke, which a bucket of unworked claims never will.
One more rule that saves arguments: every touch on an aged claim gets a note in the system with the date, who was spoken to, the reference number and the next action date. A claim with no note in 30 days is unworked, whatever anyone remembers saying about it.
A worked example
A six-provider multispecialty practice had total AR over 90 days at 22%. By payer: Medicare 6%, the largest commercial payer 11%, a regional Medicaid managed care plan 61%, workers compensation 48%. The Medicaid plan's balance was 340 claims, most under $120, denied for a taxonomy code mismatch after the plan changed its enrollment file in the spring. Nobody had worked the denials because each was small. One corrected enrollment record and a batch resubmission resolved most of them. The workers compensation balance was 19 claims, mostly legitimate delays awaiting adjuster review, and needed calls rather than corrections. The blended 22% had hidden two entirely different problems with two entirely different fixes.
The follow-up was as instructive as the finding. Two months later the Medicaid plan's over-90 share was under 15%, and the practice added a standing rule: any payer with more than 30 claims denied for the same reason code in a month triggers a root-cause review before anyone works the claims individually. That rule has caught two more enrollment-file problems since, each within weeks instead of months.
Questions we hear
What is a good days-in-AR figure?
For most office-based practices, somewhere in the 30s is achievable and under 45 is common. But it depends heavily on payer mix and specialty, and we would rather see a practice with days in AR of 42 and a clean over-90 profile than one at 35 with a growing over-120 bucket it is about to write off.
How often should we review AR aging by payer?
The full report monthly, as part of the close. The worklist weekly. Watching it daily produces noise, not decisions.
Can an outside team do the follow-up?
Yes, and old AR follow-up is often the first thing a practice outsources. The RCM audit includes the payer-level aging analysis described here, and Revelrex billing clients see aging by payer in their dashboard rather than a single blended figure.
What to do this week
- Run insurance AR aging by payer class and payer, aged by date of service, with claim counts and average balance per bucket.
- Find the two payers with the highest share over 90 days and pull 20 claims from each payer's 91-plus buckets.
- Classify each sampled claim as payer pending, denied and unworked, never received, or paid and unposted, and write the tally down.
- Build the Monday worklist from claims over 45 days with no response and denials with no logged action, sorted by payer and dollar value.
- Give every claim over 120 days a documented decision by month end: appeal, resubmit, bill the patient where permitted, or write off with a reason code.
