Timely Filing: The Only Denial You Cannot Work

I have spent most of this year looking at revenue cycle data across specialties, and one property of timely filing keeps standing out. Every other failure I have documented can be fixed after the fact.

A wrong modifier gets corrected and resubmitted. A missing attestation gets added. A claim sent to the wrong payer because of a coordination-of-benefits error gets refiled to the right one. A miscalculated unit count gets adjusted. All of those are recoverable, and the work to recover them is measurable and staffable.

A timely filing denial is not recoverable. The claim is dead. There is generally no appeal right, and the only defense is evidence that you actually filed on time and something else went wrong.

Which makes it strange that the metric most organizations manage this with is an average. If you want the broader measurement argument first, how to improve revenue cycle management covers it, and the 13 steps of the revenue cycle is the plainer orientation.

What Days in AR Actually Measures

Days in accounts receivable is total accounts receivable divided by average daily revenue. It answers one question: on average, how long does money take to arrive after you bill for it.

Common practice puts the target under 40 days, with 30 to 35 treated as strong. Aging is conventionally reported in buckets of 0 to 30, 31 to 60, 61 to 90, 91 to 120 and 120 plus, and the share sitting beyond 90 days is the bucket worth watching, because it is where recoverability drops fastest.

None of that is wrong and all of it is useful. Days in AR is a legitimate operational metric and I am not arguing against tracking it.

I am arguing that it cannot see the thing that permanently destroys claims.

What a Timely Filing Limit is

A timely filing limit is the deadline a payer sets for receiving a claim after the date of service. Miss it and the claim is denied as untimely.

Two properties make it different from everything else in this cluster.

It is absolute: A claim arriving after the limit is denied for that reason alone, regardless of whether the coding was perfect, the patient was eligible and the service was medically necessary. Being a day late is the same as being a year late.

It generally carries no appeal rights: Most other denials come with a defined appeal process. An untimely denial usually does not, because there is nothing to argue about. The one route back is evidentiary rather than substantive: proving the claim actually was submitted on time and something in transmission or processing failed. That defense depends entirely on whether you retained electronic submission reports or transmission confirmations, which is a records-keeping decision made long before the denial.

And the limits vary enormously. Across payers they range from roughly 30 days to two years, with government payers generally at the longer end and commercial payers frequently much shorter.

I am deliberately not publishing a table of specific payer deadlines, and I would treat any article that does with caution. Those limits are contract-specific, they change, and the cost of relying on a stale number is an unrecoverable claim. Your own payer contracts and provider manuals are the only authority for your own deadlines. Building that table for yourself, from your own contracts, is the single most useful hour in this article.

An Average Cannot See a Deadline

Here is the structural problem, and it is the reason a well-run practice can lose money this way without noticing.

Days in AR is computed across thousands of claims. A handful of claims crossing an absolute cutoff barely moves it. You can hold a perfectly respectable 38-day average while a small population of claims sits in a queue past a 90-day commercial limit, dying quietly.

The average is not lying. It is answering a different question. It tells you the central tendency of how fast money arrives. Timely filing is not about central tendency. It is about the tail, and about a threshold that differs per payer.

Two claims sitting at 100 days are not equivalent. One is with a payer allowing a year and is merely slow. The other is with a payer allowing 90 days and is already dead. Days in AR treats them identically, and so does an aging bucket, because both measure age against a calendar rather than against a payer’s contract.

That is the whole argument, and once you see it the fix is fairly obvious: you need days remaining, not days elapsed.

The Metric Improves as the Loss Occurs

This is the part I find genuinely uncomfortable, and it is worth stating plainly because it changes how you should read your own reporting.

When a claim is written off as untimely, it leaves accounts receivable. Total AR falls. Days in AR improves.

So the moment the unrecoverable loss is realized is the moment the headline metric gets better. A quarter in which you wrote off an unusual number of untimely claims can look, in the summary report, like a quarter in which collections accelerated.

This is a specific instance of a general problem covered elsewhere in this cluster: every headline revenue cycle metric can be moved in the right direction by doing something harmful. Days in AR is the clearest example, and timely-filing write-offs are the clearest mechanism.

The practical implication is a reporting one. Write-offs by reason code should sit next to days in AR, on the same report, always. Not in a separate report someone runs occasionally. If your AR improved and your untimely write-offs rose in the same period, those two facts belong in the same eyeline.

Where Claims Actually Get Stuck Long Enough to Die

Timely filing denials are rarely caused by someone forgetting to submit a claim. They are caused by claims entering a state where nobody owns them.

1. The rework loop: A claim denies, goes to a work queue, gets corrected, resubmits, denies again for a different reason, goes back. Each cycle consumes calendar time against a deadline that keeps running. Two or three loops on a 90-day limit is enough.
2. Missing information from a patient or a referring provider: The claim is parked pending something outside the practice’s control, and parked claims are the ones that stop appearing in anyone’s daily view.
3. Secondary claims: Waiting for the primary payer’s remittance before billing secondary is normal. If the primary is slow, the secondary’s clock has been running the whole time.
4. Credentialing and enrollment gaps: Claims held because a provider is not yet enrolled with a payer. These accumulate in bulk and are frequently discovered late, and the volume makes them the most expensive version of this failure.
5. Worklist prioritization by value: Sorting a denial worklist by dollar amount is rational and it systematically starves small claims of attention. Small claims still die on the same deadline, and enough of them add up.

Notice that four of the five are queue-management problems rather than billing errors. That is why denial management built purely around correcting claims does not solve this, and why claims processing configuration matters mainly for reducing the number of rework loops in the first place.

What Your Systems have to Know

Four requirements, and the first one is the whole article.

1. A per-claim deadline, not a per-claim age. Every claim should carry the date its timely filing window closes, derived from the date of service and that payer’s contractual limit. This is a join between a claim and a payer contract table, and it is the single highest-value thing on this list.
2. Sorting and alerting by days remaining. Once the deadline exists as data, the work queue sorts by urgency rather than by age or by dollar value. A small claim with nine days left outranks a large one with six months.
3. A hard stop before the cliff. An escalation at a fixed interval before expiry, sized to your rework cycle time. If a typical rework loop takes eleven days, an alert at seven days remaining is decoration.
4. Retained proof of submission. Electronic submission reports and transmission confirmations are the only defense against an untimely denial you believe is wrong. That is a retention policy decision, and it needs to be made before you need it rather than after.

The integration reality, since I would rather say it: the deadline calculation is trivial arithmetic, and the hard part is that payer contract terms usually do not live anywhere a claim system can read. They live in contract documents, in a spreadsheet someone maintains, or in an experienced biller’s memory. Getting those limits into a table, per payer and per contract, is the actual project. The software is the easy half.

Reading an Aging Report Properly

Given the above, a few adjustments to how the standard report gets used.

Segment by payer before drawing conclusions: A blended aging report mixes populations with different deadlines, which makes the 90+ bucket mean different things for different rows within it.

Add a days-remaining view alongside the days-elapsed view: Not instead of. Both answer real questions.

Report untimely write-offs by reason code every period, next to days in AR, for the reason described above.

Track the rework loop count per claim: Claims on their third loop are the population most likely to die, and the count is a better early warning than age.

Watch credentialing holds as a separate line: They behave differently from every other category, arriving in batches and often invisible until someone asks.

A revenue cycle analytics view that can do the days-remaining calculation is the enabling piece for most of this, and the segmentation argument is the same one made in how to improve revenue cycle management: a blended number is reportable and largely useless for deciding where to act.

[IMAGE: Two columns and a different sort order convert an aging report from a description into a work list.]

Appeal Deadlines are a Second Clock

Worth separating, because the two get conflated and they are different obligations.

The timely filing limit governs submitting the claim. The appeal deadline governs contesting a denial, and it starts from the denial rather than the date of service. For Medicare, redetermination, the first level of appeal, must be requested within 120 days of receiving the initial claim determination, per CMS. Note the clock starts from receipt of the determination, not from the date of service, and not from when someone in your office noticed it.

So a denied claim carries two live clocks: whatever remains of the filing window if resubmission is the route, and the appeal window if contesting is. They can point at different actions with different deadlines, and a work queue that tracks neither will pick whichever the biller happens to reach first.

In the order I would run them.

1. Build the payer deadline table: For your top payers by volume, extract the timely filing limit from the contract or provider manual. If this does not exist as data anywhere, that is the finding, and it is the highest-value hour in this list.
2. Calculate days remaining for your current open AR and sort ascending. Expect the top of that list to look nothing like the top of your aging report.
3. Pull last year’s untimely write-offs by payer: This number is usually available and rarely looked at. It is the size of the problem.
4. Check whether your write-off reason codes distinguish untimely from other adjustments. If they do not, you cannot do check 3, and fixing the coding is prerequisite.
5. Measure your rework loop cycle time, then compare it to your shortest payer deadline. If two loops exceed the limit, your escalation threshold is wrong.
6. Confirm you retain proof of submission and know how to produce it. This is the only defense that works, and it is worthless if the retention policy did not anticipate needing it.