Every practice owner has the same buried asset and does not see it on any balance sheet: the patients who used to come in and quietly stopped. They did not fire you. They did not switch providers after a bad visit. They meant to reschedule, life happened, and eighteen months later they are a name in your EHR with a last-visit date and no next appointment. Multiply that by a few thousand and you are looking at the single largest pool of recoverable revenue your practice has, sitting untouched because nobody has the hours to work it.
The question is not whether to run recall. Everyone agrees recall matters. The question is how, and the honest answer is that the method you choose determines whether you recover five figures or six. This is a side-by-side ROI comparison of the two realistic options for a small or mid-size practice: manual phone calls made by your front-desk staff, and AI patient recall calls that dial the entire overdue list without adding a single hour to anyone's day. The gap between them, on a book of any real size, is not close.
Why an Overdue Patient Is Worth $600, Not One Visit Fee
The first mistake in recall math is valuing a dormant patient at the price of their next appointment. If you think of a lapsed hygiene patient as a $180 cleaning or a lapsed derm patient as a $150 skin check, you will under-invest in getting them back and you will misread every ROI number that follows.
The right unit is multi-year value. A reactivated patient does not come in once. They resume a cadence: two hygiene visits a year, an annual physical plus the follow-ups it generates, a chronic-care rhythm of quarterly check-ins. Over a three-year window, a routine primary care or dental patient is worth somewhere between $900 and $1,800 in direct visit revenue, and more once you count the referrals, imaging, labs, and procedures that flow from staying engaged. For this comparison I will use a deliberately conservative $600 in three-year recovered value per reactivated patient, well below what most practices actually see.
Now the pool. A practice with 12,000 active and semi-active patient records will typically carry 20-30% of them as overdue by any reasonable recall standard: past their recommended interval, no future appointment, no recent contact. Call it 3,000 overdue names. That is not a marketing list. Those are people who already chose you once and are, statistically, far easier to bring back than a cold prospect is to acquire. At $600 each, the theoretical ceiling on that backlog is $1.8 million in recoverable lifetime revenue. Nobody recovers the ceiling. The method decides how close you get.
The Front-Desk Math: Why Manual Recall Stalls at the Same Names Every Time
Manual recall sounds simple and fails predictably. Ask the person at your front desk to spend part of each day calling overdue patients and you run into a wall of arithmetic.
A staffer juggling check-ins, insurance verification, incoming calls, and a lobby cannot dedicate a full day to outbound dialing. In practice you get maybe 60-90 minutes of real recall time, which is 40-60 dial attempts. Of those, most go to voicemail, a wrong number, or a "can't talk right now." The connect rate on cold outbound recall calls sits around 25-30%, so 40-60 dials yields perhaps 12-18 live conversations, and of those a portion actually book. On a good day one person reactivates a handful of patients.
Do the full-list math. Three thousand overdue names, worked at 40-60 attempts a day with time lost to no-answers and repeat attempts on the same numbers, is a multi-month project even before anyone quits, goes on vacation, or gets pulled onto a busier task. And they will. Recall is always the first thing dropped when the lobby fills up, so the list never gets finished. The names at the bottom age further, become harder to reach, and eventually get written off. Realistic manual reactivation across a 3,000-name backlog lands around 8-12% of the list actually returning, because you simply never reach most of them with a real conversation. That is 240-360 patients, or roughly $144,000-$216,000 in recovered three-year value, spread thin and gated by whoever you can spare.
There is a hidden cost on top of the visible one. Every hour your front desk spends dialing dormant patients is an hour not spent on the patients in front of them, and the loaded cost of that staff time is real money whether or not the calls convert.
The Text-Only Trap and the $525K-$975K Reactivation Gap
The common upgrade from manual dialing is a bulk texting tool, and it is a genuine improvement in reach: you can push a "we miss you, time for your visit" message to all 3,000 names in an afternoon. Reach is not the problem. Response is.
Dormant patients are dormant for a reason, and a one-way broadcast text does not overcome it. They have questions the text cannot answer: Is my insurance still accepted? Do I even need to come in? Can you do a Saturday? The text asks them to solve their own re-entry, and most do not bother. Text-only recall to a genuinely lapsed list reactivates around 8-12%, roughly the same ceiling as exhausted manual dialing, for a different reason. Manual calling fails on reach; texting fails on depth.
A live conversation clears both bars. It reaches everyone the way a text does, and it does what a text cannot: it answers the question that is actually keeping the patient away, then books the appointment before the moment passes. Practices that put a real conversation in front of the whole overdue list reactivate 25-35% of it. Take the midpoint, 30%, against text-only's midpoint of 10%. That is a 20-point swing on 3,000 patients: 600 additional reactivations at $600 each.
That is the number that reframes the whole decision. The gap between texting your overdue book and actually calling it is $360,000 a year at the midpoint, and it stretches to $525,000-$975,000 once you flex reactivation across the realistic 25-35% band and value each patient across their full multi-year cadence rather than a single visit. This is not a marginal optimization. It is the difference between recovering a tenth of your dormant asset and recovering a third of it.
flowchart TD
A[3000 overdue patients<br/>in your EHR] --> B{Recall method}
B -->|Manual dialing| C[Reach capped at<br/>40-60 calls a day]
C --> D[List never finished<br/>names age out]
D --> E[8-12% return<br/>240-360 patients]
B -->|Text only| F[Full reach<br/>but no conversation]
F --> G[Questions unanswered<br/>most ignore it]
G --> H[8-12% return<br/>240-360 patients]
B -->|AI recall calls| I[Whole list dialed<br/>in days]
I --> J[Live answers plus<br/>books on the spot]
J --> K[25-35% return<br/>750-1050 patients]
K --> L[$525K-$975K more<br/>recovered per year]What AI Patient Recall Calls Actually Do Across the Whole Book
The reason AI patient recall calls close the gap is that they combine the reach of texting with the depth of a phone conversation, and then remove the labor ceiling that kills manual recall. The system works the overdue list your EHR already holds, calls each patient in their preferred language, has a natural back-and-forth that handles the real objections, and books directly into your live schedule when the patient says yes. It leaves a warm voicemail and follows up by text on no-answers, so nobody drops off the list because the front desk got busy.
The economics flip because the marginal cost of an additional call is effectively zero. Whether the list is 300 names or 3,000, it gets worked completely, and it gets worked again on a cadence for the patients who did not answer the first time. There is no fatigue, no day where recall gets skipped, no bottom of the list that ages into unreachability. The same engine that recovers dormant patients also runs the ongoing recall that keeps them from lapsing again, plus the reminders and waitlist backfill that protect the visits you already have. You can see the full scope of what the AI front desk and recall engine cover on the /features page.
Run the payback. If AI recall lifts reactivation from a text-only 10% to a conservative 28% on 3,000 overdue names, that is 540 additional patients at $600, or $324,000 in recovered three-year value, with roughly a third of it landing inside the first twelve months as those patients resume care. Against a subscription that costs a small fraction of that, the tool clears its own cost inside the first billing cycle and runs at 10x or better thereafter. Transparent, per-practice pricing is laid out on the /pricing page so you can drop your own patient count and visit value into the model rather than trusting mine.
Running the ROI on Your Own Numbers Before You Commit
You do not have to take the $600 figure or the 30% reactivation rate on faith. The comparison holds at almost any inputs, and it is worth doing the arithmetic with your own to see where your practice lands.
Pull four numbers. First, your overdue patient count: how many active records are past their recommended interval with no future appointment. Most practices are surprised it is a quarter of the book. Second, your realistic multi-year value per reactivated patient, using your average visit revenue times your typical annual visit cadence over three years. Third, your current recall reactivation rate, which for most manual or text-only programs is in the high single digits. Fourth, the reactivation rate a conversation-driven program can reach, which the field data puts at 25-35%.
The formula is short:
- Recoverable gap equals overdue count times the reactivation-rate lift times multi-year value.
- Payback period equals annual subscription divided by the monthly slice of that recovered revenue.
Plug in a mid-size practice: 3,000 overdue, an 18-point lift from 10% to 28%, $600 value. That is 540 patients and $324,000 in recovered lifetime revenue, with well over $100,000 landing in year one. Even if you halve every assumption, cut the value to $300, the lift to 9 points, the list to 1,500 names, you still recover $40,000-plus a year against a cost measured in the low thousands. There is no realistic set of inputs where automating recall on a book of any size loses money. The only variable is how much you win, and that scales directly with how many dormant patients you are currently leaving in the EHR untouched.
The Real Comparison Is Reactivation Rate, Not Cost
It is tempting to frame this as a labor-savings story, and the labor savings are real: you get your front desk's hours back and stop losing recall to whatever is busier that afternoon. But labor is the small number. The decision is not about saving the cost of an hour of dialing. It is about the difference between recovering a tenth of your dormant patient asset and recovering a third of it, which on a 3,000-name book is a swing of half a million dollars a year or more.
Manual calling caps you on reach. Texting caps you on depth. A conversation at scale removes both caps, and it does so without adding a person to payroll. Pull your overdue count this week, put your own numbers into the two-line formula above, and the right method stops being a matter of opinion. The patients are already yours. The only question is whether anyone actually calls them.