A dental recall workflow should begin with the practice’s own approved recall list. If evaluating Zanus AI calling, test how that list becomes an invitation, response and appointment request. The assistant should not decide a patient’s recall interval or clinical need from conversation alone.
Keep recall eligibility with the practice
Have responsible practice staff prepare the eligible list using their normal systems and review process. Keep the recall reason, responsible clinician or team, contact preference and current booking status linked to the record.
Use test data first and complete the practice’s supplier and information-handling review before involving real patients. A general product description does not establish that a proposed configuration is suitable for patient information or practice requirements.
Make the invitation an administrative request
| Patient response | Action to verify |
|---|---|
| Wants to arrange a visit | Offer only practice-approved booking options or an owned callback |
| Already booked | Reconcile the existing appointment and stop duplicate outreach |
| Has a clinical question | Route through the practice’s approved process |
| Declines further calls | Apply the appropriate preference under practice policy |
A request to book remains distinct from a saved, accepted appointment.
Test a booking made outside the campaign
Place a test patient in the recall queue and then book the appointment through reception before the next call is due. Check whether the outstanding recall contact still occurs. Repeat with a cancellation and a new booking for a different purpose.
Next use two people sharing a phone number. Inspect how the workflow avoids disclosing or updating the wrong person’s appointment. Ask staff to approve the exact identification and shared-phone procedure used in the proposed setup.
Measure the invitation through to attendance
Track eligible records, contacts attempted, explicit responses, accepted bookings and completed visits separately. Do not count everyone who answers as a successful recall, and do not credit all routine attendance to the campaign.
Review staff time, incorrect contacts, duplicate invitations and unresolved requests alongside attendance. Keep the pilot small enough for reception to reconcile each result with the practice system. Expansion should follow an accurate administrative process with a clearly owned exception queue.
Questions and answers
Should an AI calling script choose the recall interval?
No. The recall list and clinical decisions belong to the practice’s approved process; the trial should handle only authorised administrative outreach and scheduling.
Designing the Recall List Before Any Call Is Made
The administrative work that determines whether a dental recall campaign is worth running happens entirely before the first outbound contact. A staff-approved list is not simply a data export; it is a filtered, reviewed and authorised set of records that the practice has positively decided to act on. Building that list with deliberate criteria protects patients, protects the practice and gives any evaluation of AI-assisted calling a meaningful baseline to work against.
Begin by agreeing internally on what makes a record eligible for outreach at this moment. That is a clinical and administrative judgement for the practice, not a default inherited from a software field. Staff involved in preparing the list should be able to describe, for any record on it, why that person appears, who authorised inclusion and what contact preference applies. If those three questions cannot be answered consistently, the list needs further review before any outreach proceeds.
Eligibility criteria worth documenting in writing before the campaign begins include: whether the patient has an active relationship with the practice, whether a responsible clinician has indicated recall is appropriate, whether there are any flags in the practice management system that should pause contact, and whether contact preferences have been checked against the most recent record rather than an older export. None of these are questions an AI calling system can resolve from conversation; they require a human review step with access to the full patient record.
Once the list is prepared, version it. Save a copy with a clear timestamp so that, after the campaign runs, you can compare the original authorised list against the contacts that were actually attempted. Discrepancies between those two sets are among the most revealing findings from a pilot, and they are invisible without the original reference.
Separating the Administrative Invitation From the Clinical Relationship
An outbound recall call, whether made by a staff member or an AI assistant, is an administrative invitation. Its purpose is to offer a booking opportunity to someone the practice has already determined is eligible. It is not an assessment of whether that person needs treatment, a reminder of a clinical recommendation, or an opportunity to update health information. Keeping that boundary clear in the script, in staff briefings and in evaluation criteria matters for how the practice manages patient trust and how it reviews the results.
A well-framed invitation script communicates that the practice is reaching out to offer an appointment, that booking is voluntary, and that any clinical questions will be handled by the clinical team separately. Consider drafting a short, plain-language version of the invitation and asking two or three staff members to read it and identify anything that might be heard as clinical advice or as a statement of health urgency. Their reactions complement the practice’s compliance review by testing whether the wording is understandable to someone who was not involved in drafting it.
The script should also make clear what the patient can do if they have already booked, if they want to decline further contact, or if they have a question the caller cannot answer. Each of those paths needs a defined next step, not an open-ended response. Where an AI assistant handles the call, those paths need to be tested explicitly rather than assumed from a product description.
A Worked Scenario: From List Entry to Appointment Confirmation
Consider a fictional patient, referred to here as Patient A, whose record appears on the authorised recall list. The responsible staff member has confirmed that Patient A has no contact-preference restrictions, that the record is current and that no appointment is already in the system. Patient A is assigned to the outbound queue.
The AI assistant attempts contact and reaches Patient A directly. Patient A says they would like to arrange a visit. At this point, the assistant's role is to offer the approved booking method: either a transfer to reception, a callback from a named staff member, or a link to the practice's own booking tool if that is the agreed workflow. The assistant does not confirm a specific appointment slot unless the practice has explicitly set up and tested real-time slot access. The response to Patient A is logged, and the record is flagged as awaiting booking confirmation.
Reception receives the callback flag, calls Patient A, and books an appointment through the practice management system. The appointment confirmation comes from the practice system, not from the campaign tool. Only at that point is Patient A's record marked as resulting in a booked appointment. If Patient A does not attend, the booking is recorded as a non-attendance, not as a campaign failure or success.
Now consider Patient B, who shares a phone number with another household member who is also on the practice list. The outbound call is answered by the household member, not Patient B. A robust workflow addresses this by not disclosing which patient prompted the call until the correct person has been identified, and by routing the call to reception if identification cannot be confirmed. Staff should be asked, before the campaign begins, to approve the exact identification procedure for shared-phone scenarios and to confirm it has been tested with a dummy record pair. This is not a detail to resolve after a live call goes wrong.
Criteria for Evaluating the Booking Handover
The handover between an AI-assisted call and an actual booked appointment is where many administrative gaps appear. Evaluating that handover requires specific observable evidence rather than a general sense that things went smoothly. The following comparison criteria are worth applying to any pilot before deciding whether to expand the campaign.
Logging completeness
After each contact attempt, a reviewable record should show: the patient identifier, the outcome the patient expressed, the time of the attempt, and the next action assigned. A record that shows only a call outcome code without those fields gives reception staff nothing to act on and gives the practice nothing to audit. Ask to see a sample log entry before the campaign begins and check whether every field is populated in a way staff can use.
Duplicate contact prevention
If a patient books an appointment through a channel the campaign tool does not monitor, a second outbound contact attempt should not occur. Testing this requires deliberately booking a patient through reception and then checking whether the system suppresses the next scheduled call. This test should be run with at least one cancellation-and-rebook scenario as well, because cancellations can reopen eligibility in ways that differ from an initial booking.
Response routing
When a patient raises something outside the scope of the invitation, the workflow should route that response to a named person or queue rather than leaving it unresolved. Check whether a clinical question, a complaint or an ambiguous response produces a clear next action in the log, and confirm who owns that queue and how often it is reviewed during the campaign period.
Preference recording
When a patient asks not to be contacted further, that preference should be recorded in a way that the practice management system can act on, not only in the campaign tool. Verify the exact mechanism by which a declined-contact response travels from the campaign log to the record that reception staff use when booking.
Designing the Pilot Size and Reconciliation Process
A small pilot is not a compromise; it is a method. Running outbound recall calls on a limited list gives reception staff a realistic chance to reconcile every result against the practice management system during the campaign period. That reconciliation is the only way to find out whether the administrative process works before it runs at a scale where errors multiply.
A useful pilot size is one where a single staff member can review every logged outcome within their normal working hours across the campaign window, without that review becoming a project in itself. If the pilot is too large for that, it is too large to evaluate accurately. Resize it before starting rather than discovering the problem mid-campaign.
Set the reconciliation schedule before the campaign begins. Decide how often a staff member will compare the campaign log against the practice management system, what discrepancies they are looking for, and what they will do when they find one. Discrepancies worth tracking include: contacts attempted for patients who were not on the authorised list, patients who booked independently but still received a call, patients who responded but whose response was not actioned, and patients who declined contact but received a follow-up attempt.
Each of those discrepancies has a different cause and a different resolution. A contact attempted for an ineligible patient suggests a list-preparation problem. A call to a patient who already booked suggests a synchronisation problem. An unactioned response suggests a routing or ownership problem. Conflating them into a single error count gives no useful information for improving the process.
Handling Exceptions During the Campaign
Even a well-designed recall campaign will produce responses that the workflow was not explicitly built for. A patient who asks about a family member's appointment, a patient who wants to discuss a previous visit before booking, a patient who is unhappy about being called at all, or a patient whose details have changed since the list was prepared: each of these requires a human response, and each should have a named owner before the campaign begins.
An exception queue that is reviewed once a day is more useful than one that is reviewed weekly, because some exceptions carry time sensitivity. A patient who is unhappy about being contacted needs a response within a timeframe that reflects the practice's own communication standards, not the campaign's reporting cycle. A patient whose phone number has changed needs that update reflected before a second contact attempt is made.
Staff who own the exception queue should know, at the start of the campaign, what decisions they can make independently and what decisions need escalation. That boundary should be documented, not assumed. A call that raises a clinical concern, even indirectly, should go to a clinician rather than being resolved administratively. A call that raises a data accuracy concern should go to whoever manages patient records. Ownership clarity at the start of the pilot is what makes the exception queue usable rather than a place where unresolved items accumulate.
Reviewing the Campaign Before Deciding on Continuation
At the end of the pilot, the decision about whether to continue, adjust or stop the campaign should rest on the reconciled administrative record, not on the campaign tool's own reporting. Those two sources may agree, and if they do, that agreement is itself a useful finding. If they disagree, understanding why matters more than the aggregate numbers either source produces.
A structured review should ask: whether the authorised list was contacted as intended without additions or omissions; whether responses were routed and actioned within the timeframes the practice set; whether booking handovers produced confirmed appointments in the practice system; whether exception handling resolved every open item before the review; and whether staff found the reconciliation process manageable or found it introduced more administrative work than the campaign offset. The answers to those questions tell the practice whether the workflow is operationally sound, which is a prerequisite for any decision about expanding it.
Bringing those findings to a team meeting that includes reception staff, the person responsible for the patient list and the clinician or manager who authorised the pilot produces a richer review than any single-role perspective. Reception staff will notice friction that is invisible from a data report. The list owner will identify whether eligibility criteria produced the expected contacts. The authorising clinician or manager will assess whether the patient experience matched the practice's communication standards. That combined review is the practical output of a well-run pilot, and it is the starting point for any decision about what comes next.
Sources and next steps
This independent Sysevo guide proposes evaluation steps, not hands-on Zanus AI results. Check the official Zanus AI Front Office overview and ask the vendor to demonstrate your exact workflow and confirm its current availability.
For related context, read Zanus AI dental receptionist booking. To compare providers using the same requirements, explore Sysevo outbound campaigns or book a Sysevo demo. Sysevo publishes this guide and sells its own business software.
Related workflow guides
- Zanus AI Clinic Booking ROI: Count Additional Completed Visits
- Zanus AI and HIPAA: Review the BAA and Actual Data Flow
For current Sysevo subscriptions and usage options, compare plans and pricing.



