Established practice

Your reception team can't keep up with the phones.

Better processes and the right AI tools can take the pressure off without losing the personal touch. Here is how to set it up safely.

Is this you? Tap any that are true.

Most people reading this tick at least one.

It only takes one to cost you.

steps: 8; steps
to start: 1 to 2 weeks; to start
scorecard: Free; scorecard

The short answer

Start by counting why patients call, because many calls (often 20 to 40% on our estimate) are avoidable with better confirmations and online booking. Then use AI only for simple, non-clinical tasks such as opening hours, directions and booking requests. Some patients will dislike it, and that is fine if the assistant says it is automated in its first sentence and offers a person in one step: the fair comparison is not AI against your best receptionist, it is AI against an engaged tone or voicemail at 8.45 on a Monday. Do a DPIA, check data hosting and PMS integration, and pilot at your busiest times before rolling out.

What good looks like

Your numbers, before and after

Reception calls

Every call answered · simple requests handled automatically

Team free for patients in clinic

ReportCallsSettings

Illustrative sample, not a real person or result.

8 steps

Do it yourself: step by step

0 of 8 done

Use your phone system's call report or a tally sheet. Record each call by reason (new booking, rebook or cancel, directions or parking, prices, results, clinical question, insurer, billing, other), time of day, whether it was answered, and how many voicemails are returned the same day.
Time
1 to 2 weeks
Typical cost
£0
Who does it
Reception team
Done when: You have call counts by reason and hour, a missed call rate and a voicemail callback time
Add parking, directions, prices and what to bring to booking confirmations. Turn on online booking and rebooking in the PMS. Put the top ten questions on the website. Add a callback option if your phone system offers it.
Time
2 to 4 weeks
Typical cost
£0 to £1,000 (Solvable estimate)
Who does it
Practice manager
Done when: Call volume for the top three avoidable reasons has fallen when you re-count
Start where patients least mind automation: overflow and out-of-hours calls. AI handles opening hours, directions, the price list, booking or rebooking requests and confirmations. A person handles anything clinical, results, complaints, distressed callers, insurer disputes and, where you can, new patient conversations. Write this down.
Time
1 hour
Typical cost
£0
Who does it
Practice manager with lead clinician
Done when: A signed one-page scope with a clear never list
Ask each AI voice supplier where call audio and transcripts are stored, how long they keep them, whether they train models on your data, whether they sign a UK GDPR processing agreement, and how they integrate with your PMS. If the tool assesses symptoms or urgency it may be regulated as a medical device (MHRA), so avoid that scope.
Time
2 to 4 weeks
Typical cost
£0 to £1,000 for DPO review (Solvable estimate)
Who does it
Practice manager, DPO supplier
Done when: DPIA signed off and one supplier chosen
Use the script in the asset. The assistant says it is automated in its first sentence, keeps the greeting under 15 seconds, and offers a person or a callback at any point, triggered by words like 'person' or 'reception' or pressing 0. Update your privacy notice and call recording message.
Time
1 to 2 weeks
Typical cost
Typically a monthly fee plus per-minute charges, varies by supplier
Who does it
Supplier with practice manager
Done when: Test calls by three staff of different ages all reach the right outcome
Keep, narrow or stop. If early hang-ups or complaints are high, shorten the greeting and move the transfer option earlier before giving up. If you keep it, extend hours gradually and review call reasons, transfers and complaints monthly.
Time
1 hour, then 1 hour a month
Typical cost
As above
Who does it
Clinic owner and practice manager
Done when: A recorded decision with the figures behind it, then three monthly reviews held

This problem is Solvable.

Copy this table

Call audit sheet, AI scope and patient-friendly call script

Free to copy, adapt and use. No sign-up.

Table

Call audit: [Clinic name], week of [date]

Call reason	Avoidable?	AI suitable?
New booking	No	Booking request only
Rebook or cancel	Yes, online	Yes
Directions, parking	Yes, in confirmation	Yes
Prices	Yes, on website	Yes
Opening hours	Yes	Yes
Results	No	Never
Clinical question	No	Never
Insurer authorisation	Partly	No
Billing query	Partly	Take message
Complaint	No	Never
Missed or abandoned

Busiest hours

[List the three busiest hour slots]

AI scope (sign off)

AI will: answer opening hours, directions, parking and price list; take booking and rebooking requests; confirm appointments. AI will never: discuss symptoms, results or medication; handle complaints; decide urgency. Transfer: caller says 'speak to someone' or asks anything outside scope, transfer to reception or take a callback within [2] working hours. Emergency line: 'If this is a medical emergency, please hang up and call 999.'

Supplier questions

Where are audio and transcripts stored, and for how long?
Do you use our call data to train your models?
Will you sign a UK GDPR data processing agreement?
Which PMS systems do you integrate with, and is booking live or a request?
What does it cost per month and per minute at our volume?
How quickly can we switch it off?

Signed: [Practice manager] [Clinical lead] Date: [ ]

---

AI assistant call script: [Clinic name]

Greeting (keep under 15 seconds)

"Thank you for calling [Clinic name]. I'm the clinic's automated assistant. I can help with booking requests, changing an appointment, opening hours and directions. To speak to a person at any time, just say 'reception' or press 0."

If the caller wants a person

In hours: "Of course. I'm putting you through now." If the line is busy or out of hours: "Our team is with patients at the moment. Can I take your name and number so someone calls you back by [time]?"

Rebooking

"I can help with that. Can I take your full name and date of birth? ... Thank you. Your appointment is on [date] at [time] with [consultant]. Would you like to change it or cancel it?"

New booking request

"I'll take a few details and a member of our team will call you back by [time] to confirm a time that suits you. Is this for yourself? Are you paying yourself or using health insurance?"

Opening hours and directions

"We're open [days and hours]. We're at [address]. Parking is [details]. I can text you a map link if you'd like."

Anything clinical

"I'm not able to help with medical questions, but I'll make sure a member of our clinical team gets your message. If this is urgent, please call NHS 111, or 999 in an emergency."

Ending

"Is there anything else I can help with? Thank you for calling [Clinic name]."

Trigger words for transfer

person, human, reception, someone, help, complaint, results, pain, urgent, worried, 0

Avoid these

Common mistakes

Buying an AI tool before counting call reasons, so it automates the wrong calls.
Letting the AI answer anything that sounds clinical.
Hiding that it is automated, which undermines trust when callers find out.
Making callers work through menus before they can reach a person.
Using a long, chatty greeting that older callers abandon.
Skipping the DPIA and not asking where recordings are stored.
Going live everywhere at once instead of on overflow calls at peak times.
Judging the pilot on staff anecdotes instead of transfers, hang-ups and complaints.

Free personal check

Done by Paul Robinson, not software

Answer eight quick questions and I'll show you which calls to remove and which AI can take

What I need from you

Eight tap answers about your call volume and reasons
Your phone system name, if known

What you get

How many calls you could remove at source
What AI should and should not handle for you
The three steps to take before you buy anything

Within 2 working days

No sign-in, no sales call. I read every request myself and reply on WhatsApp.

Free tools

Check where you stand

Is your reception ready for help?

You know your call volumes by reason and time
Common questions are answered online
You've decided what an AI assistant should and shouldn't handle
You've checked data protection with any supplier
Any new tool is piloted before full roll-out

0 of 5 answered

Answer them all to see your score.

Your answers aren't linked to you. We keep anonymous totals to improve these checks.

Your illustrative estimate

£42,000

Solvable estimate. Rounded to the nearest £500. Updates as you type.

Calls per day

Starting figure 60. Change it to match your practice.

Missed rate

Starting figure 15%. Change it to match your practice.

Booking enquiry share

Starting figure 30%. Change it to match your practice.

Lost for good

Starting figure 50%. Change it to match your practice.

Share recovered by removing avoidable calls and overflow cover

Starting figure 50%. Change it to match your practice.

Initial consultation fee

Starting figure £250 (midpoint of £250 (Default: £250, a typical self-pay initial consultation fee (Solvable estimate). Change it to your own fee.)). Change it to match your practice.

Working days

Starting figure 250 (from 250. All Solvable estimates; replace with your week of call data.). Change it to match your practice.

Formula

Annual recoverable value = calls per day x missed rate x share that are booking enquiries x share lost for good x share recovered x initial consultation fee x working days

Variables and defaults

calls per day: 60
missed rate: 15%
booking enquiry share: 30%
lost for good: 50%
share recovered by removing avoidable calls and overflow cover: 50%
initial consultation fee: £250 (Default: £250, a typical self-pay initial consultation fee (Solvable estimate). Change it to your own fee.)
working days: 250. All Solvable estimates; replace with your week of call data.

Defaults only. Change them in the calculator to use your own numbers.

Assumptions and defaults

calls per day = 60
missed rate = 15%
booking enquiry share = 30%
lost for good = 50%
share recovered by removing avoidable calls and overflow cover = 50%
initial consultation fee = £250 (Default: £250, a typical self-pay initial consultation fee (Solvable estimate). Change it to your own fee.)
working days = 250. All Solvable estimates; replace with your week of call data.

Solvable estimate. Gross loss at these defaults is about £84,000 a year. The share the steps can recover is roughly £42,000 a year, range £20,000 to £55,000, before the cost of any tool and before counting reception time.

Benchmarks

20 to 40% of calls avoidable
Share of inbound calls about directions, hours, prices or rebooking that better confirmations and online booking could remove (Solvable estimate; your one-week call log gives your own figure)
Solvable estimate
Software that only signposts (for example 'call 999') is unlikely to be a medical device; software that filters by red flag or severity may be one
Why the AI must never assess symptoms or urgency
Source: MHRA guidance on stand-alone software, Appendix 1: symptom checkers

Every supplier I know who solves this. No one pays to be listed. Where I've worked with them, I say so.

Lila's Desk
Reception and secretarial
Remote medical secretaries and admin assistants for small clinics and private practices, booked by the hour.
Remote medical secretaries for small clinics, DBS-checked and GDPR-trained, with pay-as-you-go and monthly options.
Best for: Small clinics and private practices
Price on request
No commercial relationship
Moneypenny
Reception and secretarial
Telephone answering, live chat and AI receptionist for clinics and consultants, handled by a dedicated healthcare team.
AI receptionist for healthcare practices, with optional escalation to your own staff or Moneypenny's healthcare reception team.
Best for: Clinics and consultants wanting 24/7 cover
Price on request
No commercial relationship
Semble
Practice management software
Practice management and patient records software for private clinics and consultants, covering scheduling, notes, telehealth and payments.
Clinic PMS covering bookings, notes, invoicing, online payments and a patient portal in one place.
Best for: Private consultants, clinics and clinic groups
Price on request
No commercial relationship

Know a supplier I have missed? Tell me Suggest yourself

The call type audit has not been done to identify which calls are automatable There is no internal resource with the technical knowledge to evaluate and implement AI call tools No PMS integration has been scoped, which is a prerequisite for automated booking via phone

The volume and type of calls received is not formally tracked
There is no classification of calls by complexity or automation suitability
PMS integration requirements for AI telephony have not been assessed
No supplier shortlist or evaluation criteria has been developed
The implementation is not sequenced, making it feel too complex to start

If reception handles 100 calls per day at an average of 4 minutes per call, automating 60% saves 240 minutes of daily staff time. At £20 per hour, that is £32,000 per year in recoverable staff time.

Reception staff are diverted from high-value patient interactions by repetitive queries
Call wait times are long, affecting patient experience and conversion
After-hours callers go to voicemail because reception cannot stay on to cover
Staff turnover risk increases as admin pressure builds
The operational case for growth is limited by reception capacity

What it looks like in practice

A clinic where reception handles 100 calls per day at four minutes per call uses 400 minutes of staff time on phone handling alone, 55 to 65% of which is for bookings, FAQs, and status queries that AI can handle
A one-week call audit categorising 20 call types by volume and complexity takes approximately two hours of management time to analyse and immediately reveals where AI could be deployed first for maximum return
AI telephony with PMS integration, where the system can book appointments directly without staff involvement, is available from specialist healthcare telephony providers at a monthly cost typically under £300 for a single-site clinic

Paul Robinson

14 years in UK private healthcare

I solve this one for clinic and hospital teams

Want a hand? I'll look at your situation and tell you the three things I'd tackle first.

A few quick questions, about two minutes. I read every one myself and usually reply the same day.

No sales call unless you ask for one. Just a straight answer from me.

I reply personally, usually the same day.

Questions

Frequently asked questions

Will AI replace our receptionists?

It should not. Its job is to take repetitive calls so your team can handle patients in front of them and complex calls.

Will older patients refuse to use it?

Some will. Give them a one-word route to a person and a callback, and measure transfers by time of day rather than assuming.

Should we tell patients they are speaking to AI?

Yes, in the first sentence. It is honest, supports transparency under UK GDPR, and a clear route to a person avoids complaints.

Is it safe with patient data?

Only if the supplier signs a processing agreement, stores data appropriately and you have done a DPIA before go-live. Callers often mention symptoms, so treat recordings and transcripts as health data. Ask the six supplier questions in the asset. This is general information, not legal advice. Speak to a data protection specialist before you go live.

Can it book appointments directly?

Some integrate with the PMS for live booking, others take a request. Test this in the demo.

What if a patient complains or the AI gets something wrong?

Log it, listen to the recording and change the script. Listen to recordings weekly during the pilot and keep the scope narrow. One complaint is information, not a verdict.

Should it ever handle clinical calls?

No. Keep clinical questions, results and anything urgent with people. Software that grades symptoms by severity or red flags may count as a medical device.

Is a callback service or remote reception cover better than AI?

For some clinics, yes. A promised callback within two hours from a person, or a remote secretary covering peaks, may solve the problem without AI. Compare both against voicemail.

Last reviewed: 30 October 2026

Sources

MHRA guidance on stand-alone software, Appendix 1: symptom checkers

The Solvable community

You're not the only one with this problem.

One problem solved every week, on WhatsApp.

Join the community

One message a week. No sales messages. Reply STOP any time.

The short answer

Your numbers, before and after

Do it yourself: step by step

1Log a week of calls1 to 2 weeks · £0

2Ask patients directly2 weeks, alongside step 1 · £0

3Remove avoidable calls first2 to 4 weeks · £0 to £1,000 (Solvable estimate)

4Decide what AI handles and what it never handles1 hour · £0

5Complete a DPIA and supplier checks2 to 4 weeks · £0 to £1,000 for DPO review (Solvable estimate)

6Script it around disclosure and an easy exit1 to 2 weeks · Typically a monthly fee plus per-minute charges, varies by supplier

7Pilot at peak times and measure4 weeks · As above

8Decide on evidence and review monthly1 hour, then 1 hour a month · As above