Waitlist Backfill Is a Data Structure, Not a Sticky Note

Almost every multi-specialty group has a waitlist, and almost none of them have waitlist backfill. The list exists as names in a note, a spreadsheet tab, or the memory of one scheduler per department, and it gets consulted when somebody has time. Cancellations do not wait for somebody to have time.

The economics are unforgiving in a way that a no-show is not. A no-show at least gets billed as a missed appointment at some practices. A cancellation two days out is a slot the group is holding open, staffed, and heated, and if it does not refill it produces exactly nothing. The window to recover it is measured in hours, and a list that requires a human to work it will miss most of them.

The industry reached for penalties instead of recovery

Look at how practices have responded to missed appointments and the gap becomes obvious.

A Jan. 7, 2025 MGMA Stat poll found that 42% of medical group leaders report their practices using a no-show fee, against 58% that do not. The poll had 622 applicable responses.

A fee is a deterrent and a small recovery of administrative cost. It does not put a patient in the chair. For a cancellation with notice, which is the majority of the recoverable volume, a fee usually does not apply at all, because the patient did the thing you asked them to do by calling ahead.

That leaves backfill as the only mechanism that converts a cancellation back into a completed visit, and it is the one most groups have never operationalized. The asymmetry is worth sitting with: significant attention on penalizing the patients who fail to show, close to none on the machinery that would fill the hole either way.

What the list has to know to be useful

A waitlist that stores a name and a phone number cannot be worked automatically, and that is the actual reason backfill stays manual.

To match a person to a freed slot without a human adjudicating, an entry needs to carry the appointment type they need, the provider or provider group they can see, the department and location they can reach, how much notice they require, and the windows they are genuinely available. Miss any one of those and every candidate has to be verified by a person, which puts you back where you started.

When those fields exist, backfill becomes a query rather than a project. A cancellation frees a slot with known attributes, the list is filtered to entries compatible with those attributes, and the offer goes out in priority order within minutes.

This is what treating the waitlist as a structured record rather than a note buys. Storing entries against the schedule itself, with the attributes attached, is the difference between a list that can be worked in software and one that can only be worked by whoever remembers who has been waiting longest.

The complication: a freed slot is not fungible

In a multi-specialty group this is where naive backfill produces a mess that costs more than the empty slot.

The obvious constraint is that a cancelled gastroenterology slot cannot absorb an orthopedics patient. The less obvious ones are what break automation in practice. A generic thirty-minute template opening may be eligible for several appointment types but not for a new patient who needs an hour, and the schedule will happily accept a booking that cannot actually be delivered.

Then there is credentialing, which is the constraint nobody sees until a claim dies. Whether a provider is enrolled with a patient’s plan resolves per provider, per payer, and often per location, and that grid usually lives in a spreadsheet outside the scheduling system entirely. Filling an open slot with a patient whose plan that provider is not enrolled in produces a visit that happens and a claim that fails weeks later. The group recovered the slot and lost the money, which is worse than leaving it empty.

Authorization adds the third. An appointment attached to an authorization does not move cleanly. Rescheduling can break the link, and a visit whose authorization has not come back can be pushed later but not pulled earlier, which rules out exactly the patients a backfill wants to move up.

So the rule the automation enforces is narrow and strict. Offer the slot only to entries that match on appointment type and duration, where the provider is enrolled with that patient’s plan, and where no authorization constraint prevents moving the visit earlier. Everything that fails one of those tests goes to a scheduler with the reason attached rather than being offered and then unwound.

Speed is the feature, and it has to run without staff

The recovery rate on a cancelled slot falls sharply with every hour it stays open, which makes response time the whole design constraint.

Cancellations do not arrive during a convenient window. They come in the evening, over the weekend, and in the twenty minutes when the front desk is checking in a line of patients. A backfill process that depends on staff attention is structurally unable to catch the ones that matter most, which are the late ones.

The working pattern is to offer immediately and to several compatible candidates at once rather than sequentially, with the slot going to the first confirmed acceptance and the others told promptly that it is gone. Sequential offers with a two-hour response window burn the entire recovery period on three people who were never going to answer.

The patient experience detail that matters here: an offer for tomorrow morning made at 7pm needs to be easy to accept and equally easy to decline. Declines are useful data. A patient who declines three short-notice offers should stop receiving them, which is a rule the list can hold and a human cannot.

The numbers that show whether it is working

Most groups cannot say what share of cancellations they refill, which means they cannot say what backfill is worth to them.

Start there. Cancellations by department, and of those, how many were refilled before the appointment time. That single ratio is the program, and the baseline is usually low enough to make the business case without any further analysis.

Then watch time to refill, because it tells you whether the mechanism is fast enough to catch late cancellations or is only recovering the easy ones with days of notice. And watch offers per fill, which is your early warning that the list has gone stale. A rising number there means the entries no longer reflect who actually wants an earlier appointment.

The last one is the honest check on quality: refilled visits that were later cancelled, rescheduled, or denied at billing. A backfill program that books aggressively and ignores the enrollment and authorization constraints will look excellent on refill rate and show up as denials a month later.

Key Takeaways

Store the waitlist as structured entries against the schedule, carrying appointment type, provider group, department, notice required, and real availability windows.
Check payer enrollment for that provider before offering the slot. A refilled visit that dies at billing is worse than the empty slot you started with.
Respect authorization constraints. A visit whose authorization has not returned can be pushed later but not pulled earlier, which excludes it from backfill.
Offer to several compatible candidates at once rather than sequentially. Sequential offers spend the entire recovery window on people who never answer.
Record declines and stop offering short-notice slots to patients who have declined repeatedly. That is a rule a list can hold and a person cannot.
Measure refill rate by department and time to refill. A high refill rate on days-ahead cancellations only means the late ones are still being lost.
Track refilled visits that were later cancelled or denied, so aggressive booking does not get counted as recovery.

Waitlist backfill is the highest-return scheduling work available to a multi-specialty group, because the demand and the capacity both already exist and are simply failing to meet. It stays undone because doing it properly requires matching on constraints that live in four different places and acting within hours. An AI team working waitlist entries against freed slots inside athenaOne can apply those constraints every time and move in minutes, and hand a scheduler the exceptions with the reason already attached.

Sources