Order-Driven Scheduling Starts by Translating the Order

Every gastroenterology practice has a queue of open orders, and every one of them knows roughly how big it is. Order-driven scheduling is the work of turning that queue into booked procedures, and the reason it stalls is rarely the booking. It is that the order was written for a clinician and the person who has to say yes to it is a patient who has never seen those words before.

An open order is a decision that has already been made. A clinician determined the patient needs a procedure, wrote it, and moved on. Everything between that moment and the patient arriving prepped is administrative, and all of it depends on someone making contact.

What actually happens is that the order sits. Staff work the queue when the phones allow, which is to say rarely, and the oldest entries age out of relevance while the newest get worked because they are on top. Patients who would have said yes in week one say they will think about it in month four.

The backlog is not an operations annoyance in this specialty. It is the practice’s own scheduled revenue, sitting in a work queue, waiting on a call that nobody has the hours to make.

The order is written in a language the patient does not speak

This is the part that generic outreach gets wrong, and it is why call scripts written from the order text fail.

An order carries abbreviations, procedure codes, and clinical shorthand. A patient hearing that read aloud has no idea whether it is the thing her doctor mentioned last month or something new and alarming. The call either confuses her into declining or produces a yes she does not understand, which becomes a cancellation the day before.

The fix is a mapping the practice owns: each order type paired with the plain-language description the practice already uses when its own staff explain it well. That mapping is written once, reviewed by clinical staff, and then used identically on every call. The AI is reading the practice’s approved wording, not paraphrasing a clinical order on the fly, and that distinction is the whole safety argument.

It also improves the outcome measurably in the direction that matters. Patients agree to procedures they can picture. The description that says what will happen, how long it takes, and what they need to arrange is the one that converts.

Booking the procedure is a sequence, not a slot

In GI the appointment is rarely one appointment, and treating it as one is what produces the day-of cancellations.

A procedure booking typically drags a prep kit, prep instructions specific to the facility and the procedure, an escort requirement for anything with sedation, a pre-procedure medication review handled by clinical staff, and sometimes a preceding office visit or clearance. Miss the escort conversation and the patient arrives alone and goes home unprocedured, which costs the slot, the prep, and the patient’s willingness to reschedule.

So the outreach has to complete a checklist, not a booking. Confirm the patient understands the plain-language description. Confirm the date works against the prep timeline rather than just the calendar. Confirm transportation. Confirm which facility and that the patient knows where it is. Trigger the prep materials. Book any preceding appointment in the same conversation so the sequence holds together.

When a step cannot be completed, the useful behavior is to book the visit and flag the open item rather than abandoning the call. A booked procedure with a transportation flag is a manageable exception. An unbooked order is not.

The order queue has entries that should not be worked

Working every open order is how a program loses credibility in month two.

Some orders are already satisfied elsewhere. The patient had the procedure at a hospital or another group and the result never made it back cleanly, so the order stays open. Calling that patient to schedule something she has already had is the fastest way to make a program look careless.

Some are stale by design. Orders written well in the past may have been superseded, and the practice needs a rule about age past which an order goes back to the ordering provider rather than out to a patient.

Some are blocked. Authorization requirements, an unresolved balance the practice’s policy says must be addressed first, or a provider who has since left. Each of those is a different route, and none of them is a patient call.

So the first pass over the queue is a triage of the queue itself, not the patients: satisfied, stale, blocked, or workable. Only the last group gets outreach. The other three get routed, and the routing is what keeps the workable list short enough to actually finish.

Cadence, channel, and knowing when to stop

Procedure scheduling is a conversation most patients postpone, which makes the contact pattern as important as the script.

Text first is usually right for the initial touch, with a call as the follow-up rather than the opener. The message names the procedure in the practice’s plain language, says the ordering provider requested it, and offers a way to book or to ask a question. Patients who are ready book without a conversation, which is a meaningful share and the cheapest yes available.

For the rest, attempts should vary by time of day rather than repeating the same 10am call four times. And every attempt has to be logged against the order itself, not just the chart, so the next person to look knows what has been tried.

Stopping rules matter as much as starting ones. After a defined number of attempts without contact, the order goes back to the ordering provider with the attempt history attached. That is a real outcome, it closes the loop, and it prevents the queue from filling with entries nobody can move.

What good looks like in the numbers

Conversion rate on workable orders is the headline: of the orders eligible for outreach, how many became booked procedures, and how long it took.

Age of the open queue is the second number, and it is the one that shows whether the program is keeping up with new volume or only clearing a one-time backlog. A queue that drops sharply and then climbs again means the ongoing flow was never automated.

Then the quality checks. Day-of cancellations and unprepped arrivals, split by whether the escort and prep steps were confirmed during the booking call. That split tells you whether the checklist is working or is being skipped for speed.

And finally, orders returned to the ordering provider. This should be a real and stable number. If it is near zero, the queue is quietly accumulating unreachable patients, and the conversion rate you are reporting is measured against a denominator that keeps growing.

Key Takeaways

Pair every order type with the plain-language description the practice already uses, reviewed by clinical staff, and read it identically on every call.
Triage the queue before the patients: satisfied elsewhere, stale, blocked, or workable. Only the workable group gets outreach.
Treat a procedure booking as a checklist, not a slot: prep timeline, prep materials, escort for sedation, facility, and any preceding visit.
Book with a flagged open item rather than abandoning the call. A booked procedure with a transportation flag beats an unbooked order.
Lead with text and follow with a call, varying the time of day rather than repeating the same attempt.
Log every attempt against the order itself so the next person can see what has been tried.
Define a stopping rule that returns unreachable orders to the ordering provider with the attempt history attached.
Report conversion on workable orders, queue age, day-of cancellations split by checklist completion, and orders returned.

The open order queue is the clearest example in gastroenterology of work that has already been earned and is simply waiting for someone to make contact. It stays undone because doing it properly means translating clinical shorthand into words a patient will act on, holding a five-item checklist together on the phone, and doing that hundreds of times a month. An AI team working the order queue inside athenaOne can run that checklist on every order, book the sequence rather than the slot, and hand back the ones that need a person with the reason attached.

Sources