Case Study · Healthcare · AI Receptionist
How a Long Island healthcare practice stopped losing its after-hours calls
1,365 calls answered in five months. Two out of every three ended with the patient connected, called back, or booked.
- Industry
- Healthcare
- Market
- Long Island, NY
- Period
- May to September 2026
- Scope
- AI phone receptionist · Direct booking
Results at a glance
May through September 2026
Holding steady from the first quarter to September
All but one after the office had closed
In its first month of self-booking
The problem
The practice is busy. Its phones don't stop at 5pm just because the front desk does. Before this project, any call that came in after hours went to voicemail, and staff returned those calls the next morning.
That sounds fine until you look at when people actually call. In the first three months we tracked, the single busiest hour of the day for calls was 5pm. 136 calls came in during that hour, and all but one arrived after the office had closed. These are people who just got out of work, finally had a minute to deal with their pain, and picked up the phone. They got a voicemail.
Some of them leave a message. A lot of them don't. They call the next practice on the list.
What we built
We set up an AI phone receptionist trained on the practice's own information: hours, locations, insurance, services, and how to tell a new patient from an existing one. It answers every call it's routed, during and after business hours, and it handles each one based on what the caller needs:
- Office open: it transfers the caller straight to the front desk.
- Office closed: it takes the caller's details and reason for calling so staff can call back first thing, with everything they need already in hand.
- Urgent: it flags the call so it doesn't sit in a queue with routine requests.
- New patients who want to book: it now books them directly into open appointment slots.
That last piece took a detour. The original plan was to connect the receptionist directly to the practice's EHR scheduling system. When that integration didn't pan out, we didn't wait on the vendor. We built our own booking flow on a shared calendar with a set of open slots the practice controls, and the receptionist books into those. It went live in September and booked two new patients in its first month.
What changed
The first three months (May to July 2026)
- 732 calls answered, 332 of them while the office was closed
- 488 calls (67%) ended in a live transfer, a captured callback or a message
- 58 callers identified as new patients, 22 of them outside office hours
- 29 calls flagged as urgent
August 2026
- 335 calls answered
- 78 callbacks captured with full details
- 10 calls flagged as urgent
September 2026
- 298 calls answered, averaging 47 seconds each
- 132 transferred live to the front desk (44%)
- 67 callbacks captured
- 199 calls (67%) ended in a transfer or a callback
- 2 new patients booked directly into the calendar
- Only 1 call in 298 scored as negative by the system's sentiment rating, down from 11 in August
The number that matters most is the one that didn't move. Two out of every three calls ended with something the practice could act on, in the first quarter and again in September. That isn't a good month. It's how the system performs.
What this means for your practice
If your phones go to voicemail at 5pm, you're losing patients you'll never hear about. They don't show up in any report because they never became patients. An AI receptionist won't replace your front desk, and it isn't meant to. It covers the hours your front desk can't, sends urgent calls to the top of the pile, and gives new patients a way to book when they're ready instead of when you're open.
More visibility brings more calls, and this is how we make sure those calls get answered.
Want to see how many calls your practice is missing?
We'll look at your call patterns and show you where patients are falling through.
Book a free consultation