Every managing partner eventually asks the same question at a budget review: are we carrying too much front-of-house overhead, or is this just what it costs to run the practice? The honest answer is that most groups have no idea, because the number they quote is the payroll line, and the payroll line is wrong by roughly 40 percent. Getting the administrative cost per physician right is the first real step toward controlling it, and once you see the fully-loaded figure, the case for changing how the work gets done stops being abstract.
This piece is written for the person who signs off on the P&L: the CFO or the managing partner who has to defend overhead to the other owners. We will put a defensible range on the number, pull apart the costs hiding behind each admin FTE, and show where AI resets the baseline rather than just trimming around the edges.
The Benchmark Nobody Quotes Correctly
Start with what the market data actually shows. For most outpatient single-specialty and primary care groups, total support staff run somewhere between 3.5 and 5.0 FTE per physician across all non-clinical and clinical-support roles. Narrow that to the front-of-house and revenue-cycle functions we are talking about here, schedulers, front desk, phone triage, billing, prior auth, and you are typically looking at 1.8 to 2.6 administrative FTE per physician.
Put dollars on it. If your average administrative wage is 22 dollars an hour, a single FTE costs about 45,760 in base pay. Load it properly and you are near 68,000. Multiply by 2.2 FTE per physician and the administrative cost per physician lands around 150,000 a year at the high end, and closer to 90,000 for a lean, well-run group in a lower-wage market. That is the honest range. When a partner says overhead is 55,000 a physician, they are quoting base salary for a single role and ignoring the rest of the roster and every load factor.
Two things drive where a specific practice falls in that range. The first is payer mix and specialty complexity, because a practice drowning in prior authorizations needs more revenue-cycle hands. The second, and the one you can actually move, is how much of the routine work still requires a human to be sitting at a desk with a phone.
Unpacking the Fully-Loaded Cost of a Front Desk Hire
The gap between base salary and true cost is where budgets quietly break. Here is what actually sits on top of that 45,760 base for a front desk role, and why the fully-loaded cost of a front desk hire runs 1.35 to 1.6 times the number on the offer letter.
- Payroll taxes and benefits: FICA, unemployment, workers comp, health coverage, and retirement match add 25 to 35 percent. That alone moves 45,760 to roughly 60,000.
- Turnover: Front desk roles turn over at 25 to 40 percent a year. Each departure costs 4,000 to 7,000 in recruiting, onboarding, and the productivity lost while the seat is empty or half-trained. Averaged across the team, budget 2,000 to 3,000 per FTE per year just for churn.
- Ramp time: A new hire takes 60 to 90 days to reach full output. During that window you are paying full wage for partial work, and often paying a second person to train them.
- Coverage and overtime: Lunches, sick days, PTO, and the inevitable no-call-no-show mean you either pay overtime or leave phones unanswered. Neither is free.
- Management overhead: A supervisor or office manager spends real hours hiring, scheduling, correcting, and refereeing. That time is part of the cost of the desk even though it lives on a different line.
Stack those and a 45,760 base becomes 62,000 to 73,000 in true annual cost. The diagram below traces how one advertised salary balloons into the real number.
flowchart LR A[Base salary<br/>45760] --> B[Add taxes<br/>and benefits] B --> C[Add turnover<br/>and ramp] C --> D[Add coverage<br/>and overtime] D --> E[Add management<br/>time] E --> F[Fully loaded<br/>62k to 73k]
The practical lesson for a CFO: never model a new hire at base pay. When you evaluate whether you need another scheduler, the decision is against 65,000-plus, not 46,000. That single correction changes a lot of hiring decisions.
Why Support Staff Per Physician Creeps Up As You Grow
Here is the trap that catches growing groups. You would expect economies of scale, that adding a fourth or fifth physician lets your existing admin team spread across more providers. In practice the opposite usually happens. Support staff per physician tends to creep up 0.2 to 0.4 FTE with each provider added, because more physicians mean more phone lines, more schedules to juggle, more claims, more prior auths, and more coordination between desks that used to just talk across the room.
The math is unforgiving. If you run 2.2 admin FTE per physician at three physicians, that is 6.6 FTE. Add two physicians and let the ratio drift to 2.5, and you are at 12.5 FTE for five physicians. You added 67 percent more providers but 89 percent more admin cost. Overhead outran revenue, and it did it quietly, one justified hire at a time. Each hire looked reasonable in isolation. The aggregate is a margin problem.
Breaking that curve is the entire game. If you can hold the ratio flat as you grow, or bend it down, the incremental physician drops far more to the bottom line. That is why the question is not really "how do we cut a position," it is "how do we stop the ratio from climbing." Those are different problems with different solutions, and only the second one scales.
Where AI Resets the Administrative Cost Baseline
The reason the ratio climbs is that the highest-volume administrative work, answering calls, booking, rescheduling, chasing no-shows, sending reminders, is linear in patient volume and effectively uncapped. It is also the work with the least clinical judgment, which is exactly the work that automates cleanly.
An AI front desk answers 100 percent of inbound calls, day or night, and books directly into your schedule without a human picking up. Self-filling scheduling pulls from a waitlist to auto-refill a cancellation before the slot ever shows up as a gap, and multi-channel reminders cut the no-show rate that otherwise forces you to overbook and over-staff. When that volume gets absorbed, each remaining admin FTE stops spending their day on the phone and starts covering more providers with the same headcount. The ratio bends down instead of up. You can see exactly which capabilities map to which cost drivers on the /features page.
flowchart TD
A[Rising call and<br/>scheduling volume] --> B{Human desk<br/>only}
B -->|Yes| C[Add admin FTE<br/>per provider]
C --> D[Cost per physician<br/>climbs]
A --> E{AI front desk<br/>plus scheduling}
E --> F[Calls booked<br/>slots refilled]
F --> G[Same FTE covers<br/>more providers]
G --> H[Cost per physician<br/>falls]The financial framing that matters to a CFO is this: the savings almost never show up as a layoff, and you should not model it that way. It shows up as the hire you do not make. When physician four joins and the phones do not force a new scheduler, that avoided 65,000 fully-loaded cost is the return. Run it against a platform subscription and the math is straightforward. At the kind of monthly pricing on the /pricing page, a single avoided admin hire typically pays for the platform several times over, and the recovered no-show revenue is on top of that.
Building Your Own Per-Physician Overhead Model
Do not benchmark against an article, including this one. Build the number for your own practice, because the range only tells you whether you should be worried, not what to do. Here is a model a CFO can assemble in an afternoon.
First, count every administrative and support FTE that touches scheduling, phones, front desk, and revenue cycle, including partial allocations of your office manager. Second, apply the load factor honestly: multiply base wages by 1.4 as a starting point, then adjust up if your turnover is above 30 percent. Third, divide by physician count to get your current administrative cost per physician. Fourth, and this is the step most groups skip, model the next two providers and ask whether the ratio holds or climbs. If it climbs, you have found your leak.
Then run the counterfactual. Take your inbound call volume and estimate what share is routine booking, rescheduling, and reminders, usually 60 to 75 percent. That is the slice AI can absorb. Convert it into FTE-hours, and you will typically find it equals somewhere between half and a full FTE of pure phone-and-schedule labor per two physicians. That is the hire you avoid, and it is the difference between a ratio that climbs and one that falls. A related breakdown of the front desk staff per provider ratio is worth pairing with this exercise, and if you are weighing a specific expansion, the real cost of adding a provider to the practice uses the same load logic.
Putting a Real Target on the Number
A well-run outpatient group should be able to hold administrative cost per physician in the 90,000 to 120,000 range and keep it flat as it adds providers. If you are above 150,000, or if your ratio climbs every time you grow, the problem is not that your team is lazy. It is that too much of your overhead is tied to work that no longer requires a person at a desk.
The move is not to cut the people doing skilled, judgment-heavy work. It is to stop the routine call and schedule volume from dictating your headcount, so the next physician you add lands mostly as margin instead of triggering the next hire. Build the model, load the costs honestly, find the ratio, and decide whether your baseline is one you chose or one that just accumulated.