DEV Community

RingBot AI
RingBot AI

Posted on

Clinics: The Phone Is Where Triage Goes Wrong, Not Just Where Bookings Get Lost

Most writing about clinic phones is about missed appointments. That is the second most expensive problem.

The first is triage. A clinic's phone is the only place where a patient can describe symptoms and be routed somewhere appropriate — and when it rings out, they either call an emergency number, turn up unannounced, or wait.

Why clinics are different from other appointment businesses

A dental practice has one service line and a fairly uniform appointment type. A clinic usually has:

  • multiple practitioners with different specialisms
  • appointment types with different durations and prerequisites
  • referral requirements that vary by payer
  • triage rules that are clinical judgements, not scheduling preferences

That last point is the constraint that makes generic "AI receptionist" advice wrong for clinics. A system that improvises a clinical answer is a liability, not a convenience.

Where the calls actually leak

1. The morning rush. 08:00–10:00 is when results, prescription requests and same-day appointment demands arrive, and when the front desk is also checking in patients.

2. The lunch hour. Phones covered by one person who also needs a break.

3. After the last appointment. 17:30–20:00, when working patients call about their own symptoms or a family member's.

4. The recall and results line. Patients ringing about results that are not ready, or about a letter they received. These calls are long, repetitive, and mostly answerable from a policy.

Sorting the calls: three buckets

Before evaluating any system, write your own list into these three buckets. This is the single most useful exercise, and it is free.

Bucket A — answer instantly, from policy. No clinical judgement involved:

  • opening hours, location, parking, accessibility
  • what to bring to a first appointment
  • whether a referral is needed for a given service
  • how to request a repeat prescription, and the turnaround
  • whether results are ready, and what the process is
  • fees, cancellation policy, late-arrival policy

Bucket B — collect, structure, route to a human. The system should capture, not decide:

  • new patient enquiries with symptoms described in general terms
  • requests for a same-day appointment
  • insurance and coverage questions that depend on the patient's plan
  • complaints and administrative disputes

Bucket C — never automated. Route to a clinician or emergency guidance immediately:

  • chest pain, breathing difficulty, severe bleeding, signs of stroke
  • anything involving a child's high fever or a pregnancy complication
  • medication side effects and dosage questions
  • mental health crisis
  • post-operative concerns

Bucket C is why a clinic cannot use a general-purpose answering bot. The agent must recognise these and hand off — and the handoff has to be tested, not assumed.

The arithmetic, done carefully

Clinics should be cautious with revenue arithmetic, because the ethical framing matters more than the optimisation framing. A useful version:

`` calls ringing out per week: 40 share that are scheduling-related (bucket A or B): 60% -> 24 share that would have resulted in a booked appointment: 50% -> 12 average revenue per visit: 90 = weekly scheduling revenue at risk: ~1,080 ``

The number is not the point. The point is that the triage calls in bucket C are not a revenue calculation at all — they are a safety question, and they should be handled correctly regardless of what they are worth.

What a system must do to be worth it in a clinic

  • Never improvise clinical guidance. If it does not know, it routes.
  • Recognise red-flag language and route immediately, with an escalation path that is tested monthly.
  • Know the appointment types and their prerequisites, so a booking is not created that the clinician has to undo.
  • Answer administrative questions from your own written policy, with the version controlled.
  • Collect structured intake — reason for call, duration, payer, preferred times — so the callback takes two minutes, not ten.
  • Work in the languages your patient population actually speaks. In multilingual cities this changes both access and safety.
  • Log everything. Who called, what was said, what was routed, and when.

Governance before technology

This is where clinics differ from every other business in this series. Before deploying:

  • Decide what the system may record and where transcripts live.
  • Decide retention and who can access them.
  • Decide what it must never say — and write that as a hard rule, not a prompt suggestion.
  • Test the escalation path with real scenarios, and re-test after any change.
  • Tell patients they are speaking to an automated system.

None of this is a feature you buy. It is a policy you write, and the vendor should be able to support it rather than work around it.

Where this is the wrong purchase

  • Small practices with low call volume — a shared inbox and a written policy sheet will do more.
  • No ability to maintain the knowledge base. Stale opening hours or a wrong fee schedule in a clinic is worse than a voicemail.
  • A patient population that relies on the phone for reassurance. Some clinics serve patients who need a human voice first; automating the first touch can reduce access.
  • No clinical sign-off on the routing rules. Do not let an operations team invent triage.

What to measure

  • Time to answer, by hour of day
  • Share of bucket A calls resolved without a human
  • Share of bucket C calls correctly escalated — audit a sample monthly, this is the safety metric
  • No-show rate before and after
  • Staff time returned per day

The third one is the only number that matters if the others conflict.


Disclosure: I work on RingBot, an Agent-as-a-Service platform whose voice agents are built with an AI Builder, bound to your own knowledge base, and can escalate to a person with a structured summary. It is a general platform, not a clinical system, and the governance questions above are the practice's to answer. See also the same arithmetic for dental practices, restaurants, letting agents and gyms, and the cost comparison of the four options.

Top comments (0)