The situation
Meridian Dental Group runs twelve general dentistry practices across two states. Each practice has its own front desk, its own Dentrix install, and, until this project, its own way of reminding patients about hygiene recall appointments. Some offices called every patient two days out. Some sent a postcard. One office manager kept a paper list.
Across the group the no-show rate on recall appointments was 18.4%. On a typical day that meant four or five empty hygienist chairs per location, and hygienist time is the most expensive thing to leave idle in a dental practice. Front desks were spending about two hours a day on reminder calls, most of which went to voicemail.
Marcus, the COO, had a specific ask. One workflow for all twelve locations, live before the next quarter, with no new software for the front desks to learn. Dentrix runs on a server in each office and was not going to be replaced. The group's compliance lead had one condition of her own: messages could carry a first name, the practice name, a date and time, and a link. Nothing else.
What we built
Each night, data moves from each practice's Dentrix into a messaging flow. Reminders go out by text at seven days, two days, and the morning of the appointment. Patients confirm or reschedule by replying. Each front desk gets a short list at 7:30 every morning of who is coming, who is not, and which slots just opened.
The Dentrix side is a scheduled report, the kind every office already ran for end-of-day. Each practice exports the next fourteen days of recall appointments to a shared folder at 11pm. A Make scenario picks up the twelve files, standardizes phone numbers, drops duplicates and patients who have opted out, and writes the rows into a per-location tab in one Google Sheet. That sheet is the main record for the workflow. It holds the appointment, the reminder stage, and the last reply.
Twilio sends the messages from one number per practice, so replies land at the right office. A reply of C marks the row confirmed. A reply of R marks it as needing a reschedule and sends the patient a link to the practice's existing online booking page. The front desk also sees the request on the morning list and calls if the patient has not rebooked by noon. Any other reply is forwarded to the office as a text so a person can read it.
At 7:30 each morning the scenario builds the list for that day: confirmed, unconfirmed, cancelled, and open slots, with the waitlist patients who asked for an earlier appointment. It goes to the office manager by email and prints to the front desk.
The rollout
Days one and two. Discovery at two practices, one large and one small. We sat with the front desks through a morning of calls and mapped what a reminder needed to say and what a reply needed to do. The compliance lead reviewed the data flow and the message templates on day two.
Days three to five. Build. The Dentrix report took one afternoon per location to schedule, done over a screen share with each office manager. The Make scenario, the sheet structure, and the Twilio numbers were finished by end of day five and tested against the two pilot exports.
Day six. Pilot at the two discovery practices. Real reminders, real replies. We adjusted the morning send time from 7am to 7:30 after the first office asked for it, and shortened the two-day message.
Days seven and eight. All twelve locations, then handoff. The runbook is four pages: what runs when, what to do if the export is missing, how to opt a patient out, how to add a thirteenth location.
Results
| Measure | Before | After 90 days |
|---|---|---|
| Recall no-show rate, all locations | 18.4% | 12.7% (−31%) |
| Front-desk time on reminder calls, per day | ~2 hrs | ~25 min |
| Appointments confirmed within 24 hours of the 7-day text | n/a | 74% |
| Opened slots refilled the same week | not tracked | 38% |
| Locations on the workflow | 0 | 12 |
The no-show number is the one the group cares about, and it dropped at every location, from a 22% improvement at the smallest office to 39% at the largest. The refilled-slot number is new: before, an opening two days out usually stayed open, because nobody had time to work the waitlist. Now it is the first thing on the morning list.
What's next
Meridian took the workflow fully in-house after the 30-day support window, which was the plan from the start. Their office managers have since added a thirteenth location on their own using the runbook. They have asked us to scope a second automation: the same nightly export, used to flag patients who are overdue for recall and have no appointment booked at all.