When Coastal Dental Group's office manager pulled the call report for the first time, she stopped scrolling halfway down the page. Out of roughly 1,400 inbound calls the previous month, the practice had answered 812 of them. The other 588 had rung out, hit voicemail, or been dropped into a hold queue nobody ever came back to.
That's a 42 percent miss rate. And it was costing them somewhere between $18,000 and $30,000 a month in lost production — not in theory, but in appointments that never got booked because the person on the other end gave up and called the practice down the street.
This is the story of what changed in their first 30 days with an AI receptionist, with the real numbers attached. The practice name and a few identifying details have been generalized, but the metrics and the timeline are representative of what a busy three-doctor practice actually experiences.
The problem: a front desk that was never the bottleneck on paper
Coastal Dental Group is a three-doctor general and cosmetic practice in Coral Springs, Florida. Two front-desk staff. A healthy patient base, a decent online review profile, and a hygiene schedule that stayed mostly full. By every metric the owner looked at on a normal Tuesday, the practice was doing fine.
The problem was the metric nobody was looking at: the calls that never connected.
Here's what a typical day looked like at the front desk. Mornings were a wall — patients arriving for 8 a.m. and 9 a.m. appointments, insurance questions, the first wave of "I woke up with a toothache" calls. Both staff members were heads-down checking people in. The phone rang anyway. Lunch was worse: one person covered the desk from noon to 1, sometimes nobody did. End of day, both staff were reconciling the schedule and chasing down tomorrow's confirmations. And after 5 p.m., the phone simply rolled to a voicemail box that, by the office manager's own admission, got checked "when someone remembered."
None of this was negligence. It's the structural reality of a front desk: the moments when the phone rings most are the exact moments your staff is least able to pick it up. You cannot answer a call and check in a patient at the same time, and no amount of "just be faster" fixes that.
The practice had also been quietly losing a specific kind of caller. Coral Springs and the surrounding Broward County area have a large Spanish-speaking population. Neither front-desk staff member was a confident Spanish speaker. When a Spanish-preference caller got through, the interaction was slow and awkward; when they hit voicemail, they almost never left a message and almost never called back. The owner suspected this was a real leak but had no way to measure it.
So the practice had three overlapping problems: a high overall miss rate, a total dead zone after hours, and a bilingual gap with no instrumentation on it. What they did not have was budget for two more full-time front-desk hires, which at Florida market rates would have run $7,000–$9,000 a month fully loaded — and still wouldn't have covered nights and weekends.
What they did: a five-day setup, not a six-month project
The owner's first question was the right one: "How long is this going to take, and how much is it going to disrupt my front desk?" Practice owners have been burned by software rollouts that promise two weeks and take two quarters.
The honest answer was five business days, and the work was mostly about information, not installation.
Day 1 — discovery. The practice handed over the things an AI receptionist actually needs to sound like it works there: hours, doctor names, the services offered, new-patient vs. existing-patient flows, the insurance plans accepted, parking instructions, and the handful of questions the front desk answers fifty times a week. ("Do you take my insurance?" "Do you do same-day crowns?" "Where do I park?")
Days 2–3 — build and call routing. The AI receptionist was configured to answer in English or Spanish based on what the caller spoke — no "press 2" menu, just a natural conversation in the caller's language. It was set up to book appointments directly into the practice's existing scheduling software, capture new-patient details, answer the routine FAQ list, and — critically — recognize a dental emergency and escalate it to a human or an on-call protocol instead of treating it like a routine booking.
Day 4 — testing. The practice called its own new line a few dozen times, trying to break it. Booking a cleaning. Asking about insurance. Switching to Spanish mid-call. Pretending to have a broken tooth on a Saturday. The escalation rules and the booking flow got tuned based on what those test calls surfaced.
Day 5 — go live. The AI receptionist was set to catch overflow first — calls that rang more than four times, calls during lunch, and every call after hours and on weekends. The front desk kept first crack at the phone during staffed hours. Nothing about the staff's workflow changed on day one. The AI simply caught what was already falling on the floor.
That overflow-first approach mattered. The owner didn't want to gamble the practice's entire phone line on new software in week one. Starting with "answer only the calls we're currently losing" meant the downside was capped at zero — those calls were already going to voicemail — and the upside started showing up immediately.
The results: 30 days, measured
Here is the comparison the office manager pulled at the end of the first full month, against the baseline month before go-live.
Missed-call rate: 42% → 4.6%. The headline number. Of roughly 1,450 inbound calls in the first 30 days, 1,383 were answered live — by a staff member or by the AI. The 4.6 percent that still slipped through were mostly simultaneous-ring situations during peak morning load, and even those callers got a callback prompt rather than a dead voicemail box.
Appointments booked by the AI receptionist: 61. These are appointments that the practice can attribute specifically to calls the front desk would not have caught — lunch-hour calls, after-hours calls, weekend calls, and overflow during the morning rush. Not every one is fully incremental; a handful of those callers might have called back later. But conservatively, the practice counted roughly 40 net-new appointments it would not have booked under the old setup.
After-hours capture. 23 of those 61 bookings came from calls that arrived after 5 p.m. or on weekends — a window that previously produced almost nothing. The most memorable was a Sunday-evening call from a patient with a cracked molar; the AI booked her into a Monday-morning emergency slot and flagged it for the team. That single visit turned into a same-week crown.
Spanish-language calls. Once the practice could finally see this number, it was bigger than the owner guessed: 19 percent of inbound calls in month one were handled in Spanish. Under the old setup, a meaningful share of those callers had been hanging up. The bilingual capability wasn't a nice-to-have — it was recovering a patient segment the practice had been silently losing for years.
Revenue math. The practice's average value for a new patient — first visit plus the typical follow-on treatment over the following months — sits in the $600–$1,200 range, with cosmetic and restorative cases running well above that. Taking the conservative 40 net-new appointments and a conservative $650 average, that's about $26,000 in production attributable to month one. The AI receptionist costs a fraction of one front-desk salary. The office manager's words: "It paid for itself before the second week."
Staff effect. This one didn't show up in a report, but the owner flagged it anyway. The front desk stopped eating lunch at the desk. The low-grade stress of a phone ringing during a check-in rush dropped, because staff knew an unanswered call wasn't a lost patient anymore — it had a backstop. Turnover at the front desk is expensive and disruptive; anything that makes the job less frantic is worth real money.
A fair note on what didn't go perfectly: the first week produced a few escalations that didn't need escalating — the AI was tuned conservatively, so it bumped some routine calls to staff out of caution. That's the correct direction to err, and the routing was tightened within days. By week three the escalation logic was landing where the practice wanted it.
Would this work for your practice?
Coastal Dental Group is not a special case. A three-doctor practice missing 40-plus percent of its calls is the norm, not the exception — industry call-tracking data has been saying so for years. If anything makes a practice a strong fit for this, it's a few specific conditions:
You're a strong fit if you have call volume your front desk visibly can't fully cover — if voicemail is doing real work in your practice, that's lost revenue, not a safety net. You're a strong fit if you serve a bilingual community and your front desk can't, because that gap is almost certainly costing you patients you can't even see leaving. And you're a strong fit if your after-hours and weekend calls go nowhere, because that's the easiest, lowest-risk place to start — every call answered there is pure upside.
The honest counter-case: if your practice is small enough that your front desk genuinely catches nearly every call, the math is less dramatic. The value here scales with the size of the leak. The way to find out is to pull your own call report — answered vs. total inbound, for one ordinary month — and look at the miss rate. Most owners are surprised, and not in a good way.
Apex Tools AI builds bilingual AI receptionists specifically for US dental practices and med spas. Plans run $400 and $450 a month for the main tiers, with a $100 add-on option, and the standard setup is the same five-business-day process described above. You can see the pricing breakdown here and exactly how the setup and call handling work here.
The fastest way to judge it is to hear it. Call the demo line at (954) 475-6922 and put it through the same test Coastal Dental Group did — book an appointment, ask about insurance, switch to Spanish, describe an emergency. If it handles your hardest calls the way you'd want your best receptionist to, the rest is just math on your own miss rate.
Apex Tools AI is based in Hollywood, Florida, and works with practices across the US.