Run a two-location optometry group long enough and the front desk stops being a staffing question and becomes an arithmetic problem. Both offices sell glasses and contacts, both bill medical for dry eye, foreign body removal, and diabetic exams, and both phones ring from the moment the door unlocks. When one receptionist quits or a second office opens, the reflex is to post a job. But before you write that offer letter, it is worth running the real cost of hiring vs virtual receptionist for clinics like yours, because the in-house number is bigger than the salary line, and the virtual number is not as small as the sales pitch implies. For an optical-plus-medical practice spread across two sites, the model you pick decides whether you are paying for coverage or just paying for a chair.
This is not a case against people. It is a case for doing the math with your actual call volume and your actual optical-versus-medical booking split, because the three models — in-house, human virtual, and AI — fail in completely different places, and the failure point is where your money leaks.
What a Loaded In-House Seat Actually Costs Across Two Sites
Start with the honest number for one front-desk hire. A receptionist at $38,000 base in most markets is not a $38,000 cost. Add employer payroll taxes, workers' comp, a health-insurance contribution, paid time off, and the weeks of training before they can confidently tell an optical refill from a medical complaint, and the fully loaded figure lands between $52,000 and $62,000 a year. That is one seat, at one site.
Now scale it to your reality. Two locations means two front desks, minimum. But a single person per site cannot answer the phone while checking a patient in, cannot cover their own lunch, and cannot take a sick day without the line going dark. Real coverage for a busy optical-plus-medical office is closer to three or four bodies across the group once you account for overlap, PTO, and the 4:30-to-5:30 rush when the phone and the front counter both peak. Suddenly the front-desk labor line is $150,000 to $220,000 loaded, and you still have gaps.
Then there is turnover. Front-desk roles churn every 18 to 24 months, and each replacement costs you the posting, the interviewing, the four to six weeks of half-productive training, and the stretch where the remaining staff cover two jobs and answer fewer calls. In optometry that gap is expensive in a specific way: a missed call is not just a lost booking, it is a lost frame sale, a lost contact-lens annual supply, and a lost medical exam, often from the same patient.
Where the Human Virtual Receptionist Math Turns
The virtual receptionist pitch is seductive because it deletes the parts of the in-house number that hurt most: no benefits, no PTO, no turnover, no desk. You pay for talk time. For a low-volume specialty office, that is genuinely cheaper. For a two-site optometry group that sells optical and bills medical, the volume works against you.
Most human virtual services bill per minute, commonly $1.10 to $1.40, or a per-call rate in the $1.50-to-$3.00 range. Optometry calls are not short. A patient asking about their contact-lens prescription, their vision-plan coverage, whether their exam is billed to VSP or to their medical insurance, and then booking around a work schedule is a four-to-six-minute call. Multiply an average of five real minutes per booking-worthy call by the volume a busy optical-plus-medical office generates, and a single location can cross $2,000 a month. Across two sites you are looking at $3,000 to $4,000, which quietly lands in the same neighborhood as one loaded in-house wage — except now a stranger with a script is representing both your offices.
flowchart LR
A[Patient calls two site optometry group] --> B{Which front desk model}
B -->|In house| C[Loaded seat 52K to 62K per site]
B -->|Human virtual| D[Per minute billing rises with volume]
B -->|AI front desk| E[Flat fee answers both sites at once]
C --> F[Coverage gaps at lunch and 5 PM]
D --> G[Script cannot see live optical vs exam slots]
E --> H[Reads live availability books at source]
F --> I[Missed frame and medical revenue]
G --> I
H --> J[Every call answered and booked]The Optical-Versus-Medical Booking Problem No Script Solves
Here is the failure that pure cost comparison hides. An optometry front desk is not answering one kind of call. A patient who wants to reorder contacts needs a short optical lane. A patient with a scratched cornea needs a medical slot, often same-day, and needs it billed to their medical insurance, not their vision plan. A diabetic annual exam has its own coding and its own time block. These are three different products routed to three different places on the schedule.
A human virtual receptionist working off a script almost never sees your live schedule. They cannot tell that Dr. Okafor's medical slots at the north office are full but her optical lane has a 3:15 opening, so they take a message or promise a callback. Your staff then plays phone tag, and the same red-eye patient who could have been booked in ninety seconds ends up at an urgent care that bills nothing back to you. Even a trained in-house person struggles with this across two sites, because they can only see the schedule they are logged into.
This is the specific pain point where the model matters more than the price. Cheap coverage that mis-routes an optical refill into a medical exam slot, or fumbles a same-day medical complaint into next week, costs you far more than the few dollars per minute you saved. The booking logic has to live somewhere reliable, and a script inside a temp's headset is the least reliable place to put it.
Coverage Math: Why Two People Still Can't Answer the Phone
Run the clock on a real day. Both offices open at 8, both take lunch in a staggered window between noon and 1:30, and both hit a wall of check-outs, callbacks, and new-patient calls between 4:30 and 5:30. A single receptionist per site is, by definition, unable to answer the phone while standing at the counter helping the patient in front of them. Two people across two sites cannot cover each other's lunches without leaving one desk unattended.
The result is that even a fully staffed in-house model drops calls at exactly the moments optometry patients call most: over the lunch hour when working patients step away from their own desks, and at end of day when they are leaving the office. A human virtual service pooled across both locations helps at the margins, but a shared pool still queues callers during a synchronized rush, and a patient on hold for four minutes about a frame purchase simply hangs up and drives to the retail optical down the street.
This is the structural reason the head-to-head cost table is misleading. You are not really comparing three prices for the same coverage. You are comparing three prices for three very different levels of coverage, and the two human models both leave the phone unanswered during your highest-value hours no matter how the salary or per-minute math shakes out.
Where a Flat-Fee AI Front Desk Changes the Comparison
An AI front desk resets the arithmetic because its cost does not scale with either headcount or minutes. It answers 100% of calls at both locations at the same time — no lunch gap, no 5 PM queue, no sick day — for a flat monthly fee that sits outside payroll entirely. There is no 25-40% load, no turnover replacement bill, and no per-minute meter running during a six-minute vision-plan conversation.
More importantly, it solves the routing problem the human models cannot. Because it is connected to your practice management system, it reads live availability at each site and applies your own rules: a contact-lens reorder goes to the optical lane, a red-eye complaint goes to a same-day medical slot with the right insurance flagged, a diabetic exam gets its correct time block. It writes the booking directly rather than leaving a message, and it works the recall list and sends multilingual reminders on top. The branching that a temp would have to memorize and still get wrong lives in software that applies it identically on every call. You can see how that booking logic and the two-site coverage fit together on the /features page, and the flat monthly line is laid out on /pricing.
For a two-location group, the practical upshot is this: the AI model is usually cheaper than three or four loaded in-house seats, competitive with or below a busy two-site human virtual bill, and — unlike both — it actually answers every call and routes every optical-versus-medical booking correctly. The cost comparison and the coverage comparison finally point the same direction.
Running Your Own Numbers Before You Post the Job
Before you hire, spend an afternoon with three figures from your own practice. Pull your monthly call volume per site from your phone system, note your average call length for a booking-worthy call, and estimate what share of your calls are optical versus medical. Those three numbers turn every vague pitch into a real comparison: the in-house cost is your loaded seats times the bodies real coverage requires, the human virtual cost is your volume times minutes times the per-minute rate, and the AI cost is a flat line you can read off a page.
When you lay them side by side with your own inputs, the pattern that shows up for most two-site optometry groups is that the cheapest-looking option on the surface is rarely the one that answers the most calls or books the most exams. The right question is not which front desk costs the least, but which one turns your 5 PM rush and your same-day red-eye into revenue instead of a missed call. Run the three numbers against that standard, and the model that wins is usually the one that never sends a patient to voicemail in the first place.