Designing an A/R Follow-Up Process

Almost every follow-up process is triggered by age, and age is the wrong trigger. It is the wrong trigger in both directions at once: a claim can be thirty days old and progressing exactly as that payer always processes, and another can be twenty days old with no record of it at the plan at all. Working the first wastes the touch. Not working the second loses the claim.

An account is workable when it is in a state that a person can act on. That is a compound condition, and it is worth writing down explicitly, because the parts that get omitted are the ones that generate wasted work.

• It was billed, and the payer acknowledged receiving it — an unacknowledged claim is a submission problem, not a follow-up one.
• No decision has arrived. A decided claim belongs to posting or to denial work, not here.
• Nothing is outstanding from the practice's side — no unsent attachment, no correction in progress, no held claim waiting on registration data.
• The time that has passed exceeds what this payer normally takes, rather than a number applied to every payer alike.

The practice's history with a payer is one source for the “too early” line. There is a second, and it is documentary rather than empirical: many payers operate under a defined obligation to act on a claim within a period. A follow-up before that period has run is a follow-up asking a question the payer is not yet obliged to answer — which is a reliable way to spend a touch and learn nothing.

Where the obligation lives depends on the arrangement, and the distinction between the two cases below is the useful part.

The clean-claim definition matters here for a reason that is easy to miss. Where a window applies only to claims that can be processed without further information, a claim the payer says it is missing something on may be outside the window entirely — so the follow-up that matters is the one that establishes which of those two situations the claim is in. That is an argument for asking early enough to find out, and it is not an argument for asking repeatedly.

The mechanism a follow-up process is built on decides how much of the book it can cover. A process that asks by telephone or by portal covers what a person can get through in a day; a process built on the standard electronic inquiry covers the book.

That is available by right rather than by arrangement. Under 45 CFR 162.1401 (opens in a new tab), the health care claim status transaction is defined as an inquiry from a provider to a health plan to determine the status of a claim, together with the plan's response; 45 CFR 162.1402 (opens in a new tab) adopts the ASC X12 technical standard for it; and 45 CFR 162.1403 (opens in a new tab) adopts operating rules governing how the transaction runs, not merely how it is formatted.

The order the work is done in is a solved problem and this article does not re-solve it. Prioritizing denial work sets out the sequencing method — deadline first, then recoverable value, then expected yield, with root-cause batching — and the method is the same whether the backlog is denials or unpaid claims. What is specific here is ownership and evidence.

1. 01

   An account has one owner at a time

   Not a team and not a queue. A shared queue with no assignment produces the pattern where the easiest accounts are touched repeatedly and the difficult ones are touched by nobody, and it is invisible because the activity counts look healthy.
2. 02

   A touch produces a next action and a date

   Every contact ends by writing what happens next and when. An account returned to the queue without those two fields will be picked up by someone who repeats the work that was just done — which is the single largest source of wasted follow-up effort, and it does not show up anywhere as waste.
3. 03

   What was said is recorded, including who said it

   A payer statement that a claim is being reprocessed is worth exactly as much as the record of who said it and when. Nothing on the remittance will confirm the conversation happened.
4. 04

   An account that cannot move escalates rather than recycles

   Two touches with the same answer is a signal, not a reason for a third. Where a process has no escalation path, unmovable accounts age in the queue until a deadline resolves them by default.

Activity measures — touches, calls, accounts worked — tell you the team is busy and nothing about whether the receivable is moving. They also rise when the process is doing the wrong work, because the wrong work is usually faster. The measures worth watching describe the book, not the effort.

• How the book is distributed by age and, more usefully, how that distribution is changing — a shape flattening toward the old end is the signal, whatever any single figure says. The A/R aging distribution calculator works it from your own figures.
• How long money takes to arrive across the whole book — the days in A/R measure, read as a trend against the practice's own past rather than against anyone's published number.
• How accounts leave — the one that actually judges follow-up. An account resolved by payment, by a corrected claim, or by a documented patient balance is the process working. An account that leaves by write-off is the process having run out of options, or out of time.

It is not claim tracking. Following one claim through statuses, asking a payer where it is, and deciding what to do about the answer is covered in Tracking a Claim, and a practice that has read only that has the more important half. This article is the layer above: how the set of claims to be tracked is chosen, assigned, sequenced and measured, so that the tracking happens on the accounts where it will matter.

And it deliberately contains no numbers. There is no correct number of touches, no interval between them, no target days in A/R and no staffing ratio here, because those depend on payer mix, specialty, balance sizes and contract terms, and a published average is a description of somebody else's practice. Every threshold this article asks for is one the practice derives from its own history.