Good isn't good enough: what an AI receptionist has to know about a physio clinic
Mann Physio isn't real. It's a composite of the MSK clinics we work with and speak to every week. The name is an inside joke from earlier days at Motics, when w...
Published
Meet Mann Physio & Performance
Mann Physio isn't real. It's a composite of the MSK clinics we work with and speak to every week. The name is an inside joke from earlier days at Motics, when we were describing a fictional clinic to work out our ideal customer profile for a trade show.
Mann Physio has three sites: a main clinic, a smaller city-centre room and a space inside a local gym. It has 11 clinicians. Six are physios, two of them part-time. One physio only sees post-op patients and only works at the main site on Tuesdays and Thursdays. There's a sports doctor two days a month, two sports massage therapists and a Pilates instructor.
About a third of patients are insured through Bupa, AXA or Vitality. The rest pay themselves. The phone rings around 1,000 times a month, and the clinic also opens on Saturday mornings.
Mann is not an unusual physio clinic. That is why we built it as we did.
What does Mann Physio offer its patients?
We asked our team to list what a clinic like Mann offers. This is what came back:
Initial assessment, self-pay (45 min)
Initial assessment, insured (45 min, needs an authorisation code)
Follow-up (30 min)
Extended follow-up (45 min)
Post-op ACL rehab review
Post-op knee replacement rehab
Sports medicine consultation
Sports massage (30 or 60 min)
Shockwave therapy
Acupuncture or dry needling
Advanced running analysis
Return-to-run assessment
Clinical Pilates 1:1
Pilates group class
Pre-season screening
That's 15, and Mann is a small group. One multi-site group we spoke to has 30 appointment types, and 60 to 70 billing codes on its sports medicine side alone, different for insured and private patients.
Every one of these has a length, a price, a list of clinicians who can see it, sites where it runs and rules about who can book it. A receptionist learns these over months. An AI receptionist has to know them on day one.
Five calls on a Saturday morning
It's Saturday morning at Mann. The front desk opens at 9. These calls come in before then.
7:40am. "I've done my back in." A new patient. The right booking is an initial assessment with a physio: 45 minutes, not a 30-minute follow-up and not a sports massage. Book the wrong slot and Monday's first clinician starts 15 minutes behind, and every patient after them has to wait.
7:55am. "I'm six weeks after my ACL surgery and need my next session." This patient can only see the post-op physio, at the main site, on a Tuesday or Thursday. The risk: they're booked with whoever is free at the gym site, and they turn up to someone who doesn't know their protocol or where they are in their recovery.
8:10am. "I'm with Bupa." Insured initial assessments need an authorisation code and a different appointment type. The risk: the patient is booked as self-pay, the invoice is wrong, the insurer won't pay it and the patient gets a bill they weren't expecting. That's more admin for the team, which is the problem the AI was meant to solve.
8:25am. "Can I move Monday's appointment?" A follow-up, as part of a course with one clinician. The right answer is a new time with the same person, and a mention of the cancellation policy if it's inside 24 hours. The risk: the patient is moved to a different clinician and the course falls apart.
8:40am. "My back's got worse and my legs feel numb." This isn't a booking at all. The right answer is to tell the patient to call 999 straight away and flag the call to the team. The risk is the AI tries to assess it or hands out medical advice. That delays the urgent care the patient needs, and it's irresponsible.
Four of these five calls can be booked by an AI. The fifth should never be. Knowing the difference matters as much as getting the bookings right.
Why 90% doesn't work in a clinic
Say Mann's AI receptionist gets 90% of calls right. On 1,000 calls a month, that's 100 calls that need fixing.
Each one means someone spots it, calls the patient back and rebooks. Some won't be spotted until the patient is in the waiting room. If each takes 10 minutes, that's more than 16 hours of staff time a month, plus the slots that go empty and the patients who don't come back. At 95% it's still 50 calls and over 8 hours a month.
In most businesses, a wrong booking is an annoyance. In a clinic, it's a patient in pain booked into the wrong slot, a post-op patient seen by the wrong person or an insured patient billed by mistake. One clinic owner told us their last AI receptionist got about 80% right, and that 80% was a catastrophe for them.
Good isn't good enough here. Every mistake means more admin for the team and a worse experience for the patient.
What to ask before you buy
If you're looking at an AI receptionist for out of hours or overflow calls, these are the questions to ask:
Does it know the difference between an initial assessment and a follow-up?
Does it know which clinicians see which patients, and at which sites?
Does it handle insured patients differently from self-pay?
Does it integrate with the systems you already use?
Can you choose which appointment types it books, and which it passes to a person?
What does it do when it isn't sure?
Can you trace every call it took and every booking it made?
If the answer to any of these is "no" or "not yet", it will create admin work rather than reduce it.
How we've built ours
Our Phone Agent is built around these questions. We set it up with each clinic, appointment type by appointment type. One clinic we work with has 69 appointment types in its practice software. Together we planned to start with about 15 for the agent to book, with the rest passed to the team.
It books straight into Cliniko, Semble and Nookal, with more EHR integrations on the roadmap. When it isn't sure, or a call needs escalating, it hands over to a team member. In one London clinic's first month, it handled 65% of calls from start to finish and passed about a quarter to the team when needed.
We're on a mission to help patients reach their clinic faster while reducing the admin load on staff. To see what missed and misbooked calls are costing your clinic, try our missed call calculator.