Best AI for Orthodontics: Following Up Free Consultations

Three kinds of tool help an orthodontic practice follow up free consultations: scheduled texts and emails that carry a booking link, an assistant that drafts personal replies for staff to approve, and an AI phone agent for inbound calls. Start with the first, and measure consult-to-start before and after.

Most lost starts are not decisions to decline. They are families who meant to call back, waited for a quote that never arrived, or got an unanswered voicemail. That is a follow-up problem, and it is cheaper to fix than lead generation.

Follow me on Instagram@sagnikteaches

Where the families go quiet

A free consultation produces four common endings: the family books on the spot, the family says they will think about it, the family goes elsewhere, or the family drifts. The second and fourth groups are where follow-up works. A useful first step is to write the stages a family passes through and count how many of them leave at each.

Connect on LinkedInSagnik Bhattacharya

1. Enquiry received (call, form, referral).
2. Consultation booked.
3. Consultation attended.
4. Treatment plan and fee presented.
5. Decision made, records or start appointment booked.
6. Treatment started.

Stage 1 and stage 6 counts are usually easy to find. The gaps between them are where AI helps, because every gap needs a message, a reminder or a call that someone must remember to send.

Subscribe on YouTube@sagnikteaches

Four jobs, and the tool type for each

Notice that none of the jobs requires an orthodontic-specific AI. What it requires is a reliable trigger (the consult was attended), clear wording and a status that gets updated. The orthodontic software you already run may do some of this, so check what its messaging module sends before adding anything.

A worked example: the numbers behind a follow-up sequence

Illustration: a two-orthodontist practice sees 60 free consultations a month. Today about 24 start treatment, which is a 40% consult-to-start rate. The team believes about a third of the families who do not start never heard from the practice again after the visit. All figures are invented to show the method.

Of the 36 who do not start, 12 never hear from the practice again. Suppose a follow-up sequence reaches those 12, and 2 of them start. That is two extra starts a month. If the average case is worth, say, $5,500 in revenue, the sequence adds about $11,000 a month. Even if the true figure is half that, the cost of a messaging tool is small beside it. Use your own case value and your own conservative start assumption, and do not promote a result until you have watched three months of real data.

The sequence itself might look like this:

1. Same day: thank-you text with the next step and one link.
2. Day 2: email with a one-page summary of the plan discussed, and a question.
3. Day 5: text from the treatment coordinator by name, offering a five-minute call.
4. Day 10: a short email answering the usual question about cost and payment options.
5. Day 21: a final message saying the offered start date is expiring or the records appointment is available.
6. Day 35: move the family to a quarterly list and stop.

Drafting the personal messages with a general assistant

For messages that need a human voice, an assistant can write the first draft in seconds. Give it the facts and the constraints, and leave out names and any medical detail you do not need to share. Here is an example prompt and an illustrative reply.

Write a short follow-up text from the treatment coordinator to a parent after a
free orthodontic consultation for their 13-year-old. Facts: the plan discussed
was clear aligners over about 18 months; they asked about monthly payments; we
can offer a records appointment next Tuesday or Thursday. Friendly, under 60
words, one question at the end, no prices, no clinical promises. Use [first name]
for the parent and sign off [your name].

A typical reply: "Hi [first name], thanks for coming in with your daughter today. We talked about aligners over roughly 18 months, and I know monthly payments were on your mind. I can talk through the options by phone in five minutes. Would Tuesday or Thursday suit for the next step? [your name]". What to fix: the draft assumed the patient is a daughter, which the prompt did not say. Delete anything the AI added that you did not supply, and keep it to the details you gave it.

That correction is the pattern for every draft: the facts come from you, and the tool supplies the wording. Never allow it to state a price or promise a result.

Replying to the question every family asks

Cost, time in treatment and "can we think about it" are the three recurring questions. Write one approved paragraph for each and reuse it. The point of AI here is to adapt the paragraph to the family's words, not to invent new commitments. The wording for fees and payment plans, including how to phrase it, is covered in fee scripts drafted with AI.

Before and after, illustrative:

Before (staff, rushed):
"Hi, just checking if you've decided about treatment? Let us know."

After (approved, specific):
"Hi [first name], just following up after Tuesday's visit. Many families
want a few days to talk it over, which is completely fine. If it would help,
I can send a one-page summary of the plan and the monthly payment options,
or book a short call. Which would be easier? [your name]"

The second message gives a choice of two easy replies. A question with a yes-or-no answer gets fewer responses than a question offering two ways to say yes.

When an AI phone agent earns its place

Inbound calls are a different job from outbound follow-up. An AI phone agent is useful when families ring back after hours, or when staff cannot answer during treatment. It should book the records appointment or the start visit, take a message about a worry, and hand off emergencies such as a broken bracket or a wire that is poking, to a person. The tutorial on an orthodontic AI receptionist for consults and broken-bracket calls explains the hand-offs in more detail.

A cautious approach is to start with after-hours only. You see how the agent handles real calls without touching daytime patients, and the downside is small, because those calls were going to voicemail anyway. Review the transcripts for two weeks before widening it.

Tracking consult-to-start without special software

You can measure this in a spreadsheet. One row per consultation, with these columns:

• Consult date and source (referrer, search, existing family).
• Treatment type discussed and the fee quoted.
• Status: attended, decided to start, records taken, started, declined, no response.
• Date of each message sent and who sent it.
• Start date, if any.

A free CRM can hold the same fields. HubSpot's free tier, for instance, allows two users and 1,000 contacts, which suits a small practice's consult list, though you should check its current limits before relying on it. On a paid tier, the CRM's advantage is that the follow-up sequence can run from the status field. HubSpot's free tools include only one automated email action and no sequences, so on the free plan you still send most messages by hand, just from better records. The spreadsheet's advantage is that you can start today.

Say the first month's sheet shows 60 consults, 41 with a message sent the same day, 24 starts. Split the starts by "same-day message or not". If 18 of the 41 who got a same-day message started (44%) against 6 of the 19 who did not (32%), the pattern is worth acting on, though with 60 rows it is a pointer, not proof. Keep the sheet running for a quarter before drawing a conclusion.

Why families stall: five reasons and a reply for each

A "we'll think about it" is rarely a single reason. Staff who note the reason in the sheet can send a reply that fits. Here are five common ones with the kind of response that works, illustrated with wording you would adapt to your own practice.

• Cost and payment. Offer a short call about payment options, not a longer email. Link the approved fee script, and do not quote numbers in the text.
• Timing around school, exams or holidays. Offer a start date that suits the calendar and say what a delay does or does not change, using the orthodontist's own wording.
• Comparing with another practice. Do not criticise anyone. Offer a second visit or a call with the orthodontist to answer specific questions.
• Fear or reluctance from the teenager. Suggest a short, no-pressure visit to see the appliances, or a call with the treatment coordinator.
• Unsure whether treatment is needed. Send the clinician-approved summary of what was found and why it was recommended. This one should always come from a person.

Tag the reason when it comes up in conversation, then count the tags monthly. If half the stalls are about payment, the fix may be a clearer plan at the consult, not a better text.

Adults and teenagers need different follow-up

Two patients can leave the same consultation with very different next steps. A parent deciding for a teenager wants clarity on cost, schedule and school disruption, and may speak to a partner before replying. An adult patient deciding for themselves tends to ask about appearance, treatment time and whether it fits around work, and often replies faster by text.

Write two versions of each message in the sequence. Keep the structure identical so you can track them both, and vary the first line and the question. An assistant is handy for producing the second version quickly: paste the approved message, state who the reader is, and ask for a rewrite that keeps every fact unchanged. Check that no new fact crept in.

The free consult no-show

A family who booked and did not come is a follow-up case too. Waiting a week loses them. A sensible routine is a text one hour after the missed time, saying no problem and offering two alternative slots, then one more message the next day. Keep the tone blameless, because someone may have had a school event or a sick child.

"Hi [first name], we missed you today and hope everything is okay. If you'd
still like your free consultation, I have Thursday at 4pm or Saturday at 9am.
Just reply with the one you prefer. [your name]"

Track no-shows separately from attended consults. A high no-show rate points to booking-side fixes, such as confirming by text the day before and again that morning, and has nothing to do with how persuasive the consult was.

A Monday review that takes fifteen minutes

Automation does not remove the need for a person looking at the list. A short weekly meeting keeps the sequence honest:

1. Open the sheet filtered to families seen 5 to 30 days ago with no start date.
2. Read any replies that the system could not classify, and answer them.
3. Pick the five oldest families with no reply, and have the coordinator phone them.
4. Note one reason for a decline or a stall and add it to the tags.
5. Check that no sequence is still running for a family who has booked.

An illustrative Monday: 22 families are in the window, 15 have a scheduled message, 4 have replied and need a person, and 3 have had no contact in ten days. The coordinator makes seven calls and books two records appointments. Fifteen minutes of review has done more than another message would have.

What a monthly report should show

Keep the report to one page so people read it. Every line should be a number you can find in the sheet:

Month: [   ]
Free consultations attended: 60      No-shows: 9
Same-day message sent: 41            Reply rate: 28%
Starts within 90 days of consult: 24 (40%)
Starts from families who replied to a message: 11
Top stall reasons: payment 9, timing 7, comparing 5
Families moved to quarterly list: 12

These numbers are illustrative. The point is to see the same ten lines every month, because a trend over three or four months tells you more than any single figure. If the start rate climbs while the reply rate stays flat, the tone of your messages may not be the driver; look at what else changed, such as a new coordinator or a seasonal pattern.

A realistic mistake: the sequence that ignores the answer

A practice sets up an automatic five-message sequence. A parent replies to message two, saying they have decided to go elsewhere. Messages three, four and five still arrive. The parent posts a complaint, and the practice learns that its tool had no rule to stop the sequence on a reply.

Before turning any sequence on, test three behaviours: a family replies, a family books, and a family texts STOP. Each should end the sequence at once, and a person should see the reply. If the tool cannot do this, do not use it for anything beyond reminders.

Consent, wording and what to leave to people

Marketing messages to families are regulated in many places, and rules on consent for texts and emails differ. Check your local requirements with a qualified adviser, keep a record of how each family agreed to be contacted and make unsubscribing easy. Several rules of thumb help regardless:

• Do not put clinical detail in a text. "Your plan" is fine; the diagnosis is not.
• Do not let an AI quote a fee, a duration or an outcome.
• Address the parent when the patient is a minor, and use the patient's name sparingly.
• Give a real name and a direct line, so replies reach a person.

The staff side matters too. A coordinator who is comfortable on the phone will close more starts than a better template. The practice role-play scripts for phone calls give you a way to rehearse the call that follows the text.

Checking a sequence before it goes live

Run a dry test with a colleague posing as a family. Enter them as a consultation, mark the visit attended and watch what arrives, and when. You are checking four things: the message wording reads as written by a person, the links open the right booking page on a phone, the stop rules work, and a family who books mid-sequence stops receiving prompts. Fix any failure before a real family sees it.

Then run the first live month with only half your consults in the sequence, alternating by week, so you have a comparison group. Alternating weeks is not a perfect experiment, but it is far better than comparing this month with a quieter one last year. When the numbers are clear, extend the sequence to every family.

Keep a note of what you changed and when, so a jump in the start rate can be traced to a cause and not credited to whichever tool happened to be switched on that month.

Matching the tool type to the gap you measured

Pick by the gap you measured. If families vanish before the free consult, start with reminders and confirmation. If they attend but then go quiet, start with the same-day message and the five-step sequence. If your phone is the problem, start with after-hours answering. If you cannot tell where families go quiet because you do not record it, build the spreadsheet first, because every tool decision follows from that data.

Fill the first row of next month's sheet today, run the sequence on one month's consults, and compare the start rate with the previous three months. Referrers and parents who arrive through marketing are covered in marketing for orthodontic practices, and for families who drifted away more than a month ago, win-back messages for lapsed patients uses the same approach.

If you would rather map this with someone, that is what my AI implementation consultation covers. Otherwise, the useful first move costs nothing: write down the six stages above and count how many families sit in each one this month.

Questions orthodontic teams ask about consult follow-up

How soon should the first follow-up go out after a free consult?

Same day is the safest default, ideally within a few hours while the family still remembers the visit. A short message with the next step beats a long one. Test a same-day message against a next-morning one on your own consults for a month and keep whichever starts more cases.

Should AI send the follow-ups without a person checking?

For routine reminders with fixed wording, yes. For anything that discusses fees, treatment advice or a family's worries, have a person approve or write it. A useful split is automatic for booking links and reminders, drafted by AI but approved by staff for everything personal.

How many follow-ups is too many?

Four to six touches over three to four weeks is a sensible starting point, after which you stop and move the family to a longer-term list. The count matters less than giving every message a purpose and an easy way to opt out. Watch your complaint and unsubscribe rate, not just the start rate.

What is a consult-to-start rate?

It is the number of patients who begin treatment divided by the number who attended a consultation in the same group, usually tracked by month of consult. Count to a fixed date, such as 90 days after the visit, so slow decisions are not mistaken for lost cases.

Further reads

• Orthodontic Referral Form Template and an AI Completeness Check — A check for referral forms arriving from other dentists.
• How Dental Practices Use AI for Recalls and Appointment Reminders — Recall and reminder messages that run on the same triggers.
• Explaining Dental Treatment Plans to Patients With AI — Explaining a treatment plan in words a family follows.
• Dental Practice AI Mistakes: Consent, Data, and Over-Automation — Consent, data and over-automation errors to avoid.
• AI Receptionist for a Dental Office: Weave, Zo or a Generic Agent? — Phone agents compared by what they actually book.
• What Does an AI Receptionist Cost a Dental Practice in 2026? — What an AI receptionist costs a dental practice.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: HubSpot's CRM product and pricing pages for the free-tier limits (read 4 November 2026); otherwise practice-side tracking methods and tool types, so check each vendor's own pages for current details.