Unscheduled Treatment Follow Up That Recovers Real Revenue

There is a report in your practice management system that almost nobody runs. It lists every patient who was diagnosed, had a plan presented, said yes, and then never got put on the schedule. For most practices it is the largest single pile of recoverable revenue in the building, and it grows every week.

Published benchmarks put industry case acceptance around 45%, with the top 10% of practices reaching 75%. That 30-point spread is not usually a diagnosis problem or a persuasion problem. It is a follow-up problem, and follow-up is the first thing that dies when the phone is ringing.

The catch is that chasing this list badly creates a different problem. A text message that names a procedure has just put clinical information on a channel you do not control, sent to a phone that may not belong to the patient any more.

What counts as unscheduled treatment?

Any treatment a patient accepted but never booked. The diagnosis happened, the plan was presented, the patient said yes, and then nothing was put on the calendar. It sits in your practice management system as a completed conversation and an empty schedule slot, earning nothing.

It is worth separating this from 2 things it gets confused with. A patient who declined is not unscheduled treatment, they are a closed case. A patient who has not been in for 2 years is a recall problem, covered in patient recall and reactivation. Unscheduled treatment is the narrow, valuable middle: people who already agreed.

The expensive leak is between accepted and scheduled, after all the persuasion is done.

That middle is valuable precisely because the hard part is done. There is no need to build trust, explain the problem, or overcome an objection. Somebody already did all of that, in person, and then the appointment did not get made because the front desk was busy or the patient wanted to check a schedule and never called back.

How much is sitting in your chart right now?

More than most owners guess, and it is measurable in an afternoon. Published benchmarks put industry case acceptance around 45%, with the top 10% of practices reaching 75%. The gap between those 2 numbers is treatment that was diagnosed, presented, accepted in principle, and then quietly abandoned.

Be careful with the recovery numbers vendors attach to this. A commonly repeated claim is that practices with a tracking system recover 30 to 40% of unscheduled dollars within 90 days. That figure comes from companies selling tracking systems, not from an independent study, and no honest version of this post can promise you a percentage.

Two benchmarks and two vendor claims. Only the first pair belongs in a business plan.

What is defensible is the mechanism. Accepted treatment that nobody follows up on has a recovery rate near zero, because the patient's own next step was to call you and they did not. Anything you do is measured against zero, which is why even a clumsy first attempt usually produces a result.

A related figure worth treating carefully: a widely circulated claim attributes to the ADA that 20 to 30% of patients go inactive within 18 months without follow-up. It appears on vendor pages rather than in a document we could fetch from the ADA directly, so treat it as directional rather than as a citation you would repeat to a banker.

Why does manual follow-up always stop?

Because it is nobody's actual job. Front desk work is interrupt-driven, and a list of people to call back loses every time the phone rings. The list is also demoralising: it grows faster than anyone can work it, so it gets abandoned rather than finished.

The pattern is predictable enough to plan around. Somebody decides to fix this, works the list hard for 2 weeks, books a few thousand dollars of treatment, and then a staff absence or a busy month arrives and the list goes untouched for a quarter. When they come back to it, the oldest entries are stale, the phone numbers have changed, and the whole thing feels hopeless.

There is a second reason, and it is about who is comfortable making the call. Asking a patient to spend $2,400 feels like selling, and most front-desk staff did not take the job to sell. They will happily book someone who calls in, and quietly avoid a list that requires them to raise money with people who already hesitated once. That discomfort is real and it is not fixed by a better spreadsheet.

Automation helps here for an unglamorous reason: it does not get busy. A sequence that goes out on day 3, day 10 and day 30 runs on day 3, day 10 and day 30 whether or not the practice had a hard week. That consistency, rather than any cleverness in the message, is most of the value.

What should the follow-up actually say?

Less than you think, and never the diagnosis. A message that names a procedure in a text is a disclosure risk on a channel you do not control. The safe pattern references an unfinished plan without describing it, and moves the conversation to a booked call.

Compare 2 messages. "Hi Susan, you still need that crown on tooth 14, want to book?" names a body part, a procedure and a clinical finding, sitting in a text thread on a phone that might be shared, lost, or no longer hers. "Hi Susan, this is Bridge Street Dental. We have some unfinished treatment on your chart. Can we get you booked? Reply BOOK or call us at 555-0139" says the same operational thing with none of the clinical content.

Write every message as though it will reach a stranger, because eventually one does.

The same restraint applies to voicemail, which practices treat as private and is not. A voicemail plays out loud in a kitchen, a car, or an office. Anything you would not text, you should not leave on a machine either, and the safe voicemail is the same 2 sentences: who is calling, and please call back.

The second version also survives the situation everyone forgets: the number is wrong. Patients change numbers, and practices carry old ones for years. A message with no clinical detail sent to a stranger is an awkward wrong number. The first version sent to a stranger is a disclosure.

This is the same boundary an AI receptionist works within on the phone: it can confirm that something is outstanding and book time to discuss it, but it does not describe treatment or answer clinical questions. That division is covered in what an AI receptionist does and where it stops.

How do you decide who to chase first?

By value, not by age. Reported analysis suggests the top 20% of treatment plans by value account for about 80% of unscheduled revenue, which means a list sorted by date wastes most of its effort on the cheapest cases sitting at the top.

That Pareto claim comes from a vendor and should be treated as a rule of thumb rather than a measurement, but the underlying logic holds in any practice with a mix of small and large cases. A single accepted implant case is worth dozens of accepted fluoride treatments, and both appear on the same undifferentiated list.

High value gets a person. Everything else gets a sequence with no clinical detail in it.

Value is not the only sort worth applying, though it is the first. A plan accepted 3 weeks ago is warmer than one accepted 9 months ago, and a patient with an active insurance benefit that resets in December has a deadline the practice can honestly mention. Layering recency and benefit expiry on top of value turns a flat list into something closer to a priority queue.

The practical split is 2 tracks. High-value plans get a human call from someone who can actually discuss scheduling and cost, because a $4,000 case that was accepted 6 weeks ago deserves a conversation. Everything else gets the automated sequence, which is cheap enough to run against hundreds of names and effective enough to book the straightforward ones.

Where should you start?

With one report and one sequence. Pull every accepted plan with no future appointment, sort it by value, and take the top 20 names. Those 20 calls are worth more than the next 200, and they tell you whether the sequence is worth automating.

Do those 20 by hand first. You will learn things a dashboard will not tell you: how many numbers are dead, how many patients forgot entirely, how many were waiting on an insurance answer nobody followed up on, and how many say yes immediately. That information shapes the automated sequence far better than a template does.

One more thing worth building in from the start: a stop rule. A patient who has said no twice should come off the sequence, not cycle forever. Practices that skip this end up with an automation that irritates the same 40 people every quarter, which costs goodwill for no revenue and eventually produces a complaint. Decide in advance how many attempts is enough, and what happens to a name after that.

Then automate the long tail, keep the top of the list human, and put the report on a weekly schedule so the pile stops rebuilding silently. If you want help finding where else revenue is leaking out of the front desk before it ever reaches the chart, our free Growth Leak Audit looks at the whole path.