Should a practice add front-desk staff or automate?

This decision is nearly always framed as a headcount question and nearly always settled by whichever queue is easiest to observe. A practice has two patient-facing queues running at once: the people standing at the desk, and the people holding on the line. They compete for the same coordinators. Moving somebody from one to the other does not add capacity — it changes which queue is allowed to be long, and the queue that gets shortened is almost always the one management can see from the corridor.

That visibility asymmetry is not a soft observation, it is structural. A patient who waits at the desk is in the building, is counted, and complains to someone who works there. A patient who hangs up after ninety seconds on hold has left no trace that anyone in the practice will encounter, because in most phone configurations the record is created when a call is answered. The loss is real, it is recurring, and it does not appear in any report the decision is made from.

There is a measurement that would settle it, and the practice already owns it. Call abandonment — the share of calls that entered a queue or rang, and were ended by the caller before a person picked up — is reported by essentially every business phone system. What does not exist is an aggregate: no body publishes a cross-practice abandonment figure with a stated sample, which is why a vendor's deflection claim cannot be benchmarked against anything, and why the honest comparison is against the practice's own baseline rather than against an industry number.

The cost side has its own trap. The obvious source for what a front-desk hire costs is the county wage series for the industry, and it is the wrong instrument for this. It divides total quarterly wages by average monthly employment across every employee in the industry code — clinicians included — so in an ambulatory setting it produces a figure well above any actual front-desk wage and well below what a clinical hire costs. It is a fine measure of an industry and a poor proxy for a role.

The practical resolution is usually not either-or. Automation earns its place on the queue that produces no record: an always-on first response that acknowledges the enquiry, holds the conversation open and captures the callback details, so that an abandoned call becomes a recorded one. A person earns their place on the interaction that requires judgement, which is the one at the desk. Framed that way the question stops being whether to hire and becomes which queue each resource is genuinely better at.

The failure mode to avoid is buying automation and pointing it at the visible queue, because that is the queue where the automation will be measured and found unnecessary. Deflecting check-in produces a kiosk nobody uses and a coordinator who is still on the phone.