The real cost of your radiology front desk

At 8:05 on a Monday your reception phone has four callers holding and one patient at the counter. Two of those callers hang up before anyone speaks to them. One was a GP practice trying to book an urgent CT.

Nobody logs that. It does not appear in your RIS, your daily list, or your revenue report. It surfaces six weeks later as a quiet drop in referral volume from a practice that used to send you eight studies a week.

Start with the calls you never answered

Across imaging centres, roughly 18% of inbound calls are lost to engaged tones and voicemail. That is the number worth arguing about before anyone discusses staffing levels.

Every one of those calls was a booking, a referral, or a referrer who now thinks of you as hard to reach. Some of them go straight back to Google and search for somebody nearer.

What a receptionist actually costs

The salary is the smallest part

In any Australian capital city, medical receptionists earn roughly $60,000 to $75,000 a year. That figure is where most staffing conversations start and stop.

The real number includes four weeks annual leave and ten days personal leave (the phones keep ringing while the seat is empty, so someone backfills it), workers compensation and on-costs, then recruitment, training, rostering and supervision.

Fully loaded, one desk seat costs materially more than the salary line in your budget. Multiply that by the four seats a mid-volume site needs just to hold a Monday morning, and reception becomes one of the largest controllable line items in the practice.

Why “replace the front desk” is the wrong goal

This is where most AI vendors overreach, and where the operational logic breaks.

Referring doctors must reach a person. A GP calling about an urgent finding, a specialist chasing a report, a registrar asking whether contrast is appropriate: none of those are ticket-shaped, and none of them should end at an AI.

Referrer calls run at about 12% of inbound volume. Our own cost model deliberately keeps receptionists for that 12% and does not pretend otherwise.

The actual target

The goal is not fewer humans on the phone. The goal is humans on the calls where a human changes the outcome, and a machine on the routine traffic where one does not.

Where the volume actually goes

TriageVoice answers every inbound call within two rings, on the phone line and on the website voice button, 24/7. It identifies intent on the first exchange: booking, FAQ, referrer escalation, emergency, or human handoff.

The routine traffic resolves without touching a person: bookings, reschedules, preparation questions, opening hours. The rest lands at reception with context already captured, so the patient does not repeat themselves and your receptionist does not re-key anything. How large each share is depends on your call mix and the workflows you configure, which is why we measure it against your own baseline rather than publish a percentage.

Concurrency matters more than answer speed. Ten simultaneous callers at 8am are all answered at once, not queued one receptionist at a time. The Monday surge stops being a staffing problem and becomes a non-event.

After hours is smaller than you think, and worth more

Calls landing outside desk hours run at roughly 4% of business-hours volume, measured across a full month of live Triagents phone data.

4% sounds trivial until you price it. Those calls currently reach voicemail, and a patient who hits voicemail at 7pm on a Thursday books with whoever answers first the next morning.

An after-hours booking is not incremental admin. It is a scan that would otherwise have gone to the practice down the road.

What changes for the referrer

Referrer recognition happens on the first exchange, not after a menu tree. A doctor is routed on a different path from a patient asking about parking.

By your rules, that path is a warm transfer to reception or the on-call radiologist, with the transcript and clinical context on screen before the call connects.

Out of hours, the referrer still reaches a person or your nominated escalation path, never a mailbox. Referrer loyalty is built on answered calls, and it decays the same way.

What changes for the patient

Two rings, no hold music, no menu tree. The agent asks only what the booking genuinely requires, because modality, body part, contrast rules and funding pathway are modelled per machine and per site.

The confirmation SMS goes out before the call ends.

The patient standing at your counter benefits too, and this is the change practice managers notice first. Your staff stop splitting attention between a handset and a person in front of them.

Where the RIS fits

None of this helps if bookings live in a parallel system. Standard RIS and PMS connectors are part of the core offering; which are enabled at your sites, and any bespoke integration for a site running something unusual, is set out in the quotation. Support for a given RIS version is confirmed during scoping rather than assumed, and the integration scope is published in one place.

Bookings, referral details and call outcomes write back into the systems your radiographers and reception already work in. The intake layer does not become a second source of truth.

Australian clinics run on sovereign Australian infrastructure, which matters when the layer handling first contact is handling identified patient information.

What this does not do

It does not triage clinically. It does not offer an opinion on whether a study is appropriate. It does not keep an emergency caller talking: an emergency is recognised on the first exchange and the caller is told to hang up and call 000 (or the emergency number for the clinic’s country), immediately and at every site.

And it does not remove your front desk. It removes the phone from your front desk, which is a different claim, and a more defensible one.

Run your own numbers

Everything above depends on your call volume, your billing mix and current staffing. Those differ sharply between a single suburban site and a nine-site network.