How to Bring Lapsed Patients Back With AI

Every practice is sitting on a list it rarely opens. It lives in the practice management system, under the tab the front desk gets to only on a quiet afternoon, which is almost never: the patients overdue for hygiene, the ones who accepted a treatment plan and never booked it, the ones who were regulars and quietly stopped coming. That list is the cheapest production a practice can recover, because those patients already know you, already trust your providers, and are already in your system. The only thing standing between them and a booked visit is someone reaching out. This article is about doing that well, and about the one step in every recall system that reliably breaks. Pesta is one way practices are closing that gap, and we will get to where it fits, but the problem comes first.

The recall list is the highest-return outreach a practice ignores

Reactivating a lapsed patient costs a fraction of acquiring a new one through marketing, and the returning patient is worth more over the following years than a cold lead. The economics are not subtle: practices with strong recall keep retention above 85 percent, while manual or neglected recall drops it toward 60 percent, forcing a constant, expensive dependence on new-patient marketing just to stay level. Every point of recall improvement compounds.

So the recall list is not a chore to fit in when there is time. It is arguably the highest-return outreach a practice has. And it is the most consistently neglected, for one simple reason: working it well takes conversations nobody has time to have.

Recall and reactivation are different work on the same list

Before the how, one distinction the best practices insist on, because getting it wrong wastes the whole effort.

• Recall keeps active patients on schedule: the patient who is a few weeks past their normal hygiene date and just needs a nudge to rebook.
• Reactivation recovers patients who lapsed months or years ago and drifted away, often with open treatment they never completed.

They live in the same database but need different messages. A patient 30 days overdue and one three years gone should not get the same outreach, and treating the whole list identically is the first thing that makes a recall system underperform. Segment by how overdue the patient is and by the value of any unscheduled treatment, and the outreach starts converting.

Text first, voice second: the cadence that actually works

The recall pattern that reactivates patients is multi-touch and cadenced by patient type, and the published playbooks agree on its shape.

Text leads because open rates are high and most patients prefer it for routine scheduling. But text alone leaves money on the table, because the patients who ignore the text are exactly the ones worth a call. The voice touch is where the harder, higher-value reactivations happen. And it is where every recall system quietly breaks.

Everyone agrees the recall call has to book on the spot. Nobody says who makes it

Read any serious recall playbook and you will find the same instruction: the second touch should be a phone call that can actually book the appointment on the line, not one that leaves a callback message. The phone call is described, correctly, as the friction point where reactivation is won or lost.

Then those same playbooks fall silent on the obvious question: who makes that call? The honest answer, in most practices, is nobody. The front desk is busy with the patients in the building. The recall voice touch is the step that requires a real back-and-forth, a patient who wants to explain why they lapsed, ask what the visit will cost, or find a time that fits, and it is exactly the step you cannot template or blast. So it does not happen. The text goes out, the non-responders sit there, and the highest-value reactivations never get the call the playbook said was essential.

That is the gap. Not the strategy, which is well understood, but the labor. The recall system breaks at the one touch that needs a conversation and the one resource no practice has spare: someone to have it.

How an AI voice agent makes the recall call that never gets made

This is the step an AI voice agent is genuinely suited to, and it is worth being precise about where it fits: as the voice second touch the playbooks prescribe, after the text, to the patients who did not respond. It calls the overdue patient, has a real conversation rather than reading a recorded message, answers what the visit involves, and books the appointment on the call, which is the whole point of the voice touch. Done this way it turns the recall list from a tab nobody opens into outreach that actually runs.

Two capabilities matter for a recall call specifically. First, the conversation does not fit a script: a lapsed patient asks "how much will this run me," "do you still take my insurance," "can I come on a Saturday," and a rigid system stalls. Because Pesta answers from a knowledge base rather than a fixed script, it can handle that back-and-forth and still get to a booking, the same reason it resolves the off-script call a scripted tool would fumble. Second, Pesta's call analysis reads every recall call, so you can see which patients are worth a second attempt, which objections keep coming up, and where reactivations stall, which turns a blind list into a workable pipeline. Powered by Deepdub, it also handles callers in a wide range of languages, which a diverse patient base needs more than most software admits. Reactivating patients this way also protects the schedule you rebuild, since the same system can move recovered patients into the reminder workflow that keeps no-shows from draining the day.

None of this automates clinical judgment or replaces your team on the calls that need a person. It handles the high-volume, consented second touch so the reactivations that were slipping away actually get the call.