Key takeaways
- Most requests for another receptionist are about a two- or three-hour peak. Compare talk minutes needed in the busiest hour with phone minutes actually available.
- Sort a sample of calls into three buckets: needs the chart, needs only approved information, should never reach the desk.
- Price a hire with benefits: the $19.01 median hourly wage for receptionists in physician offices is roughly $27.67 an hour loaded[2][3].
- Automation today takes approved-information calls and overflow, not scheduling or eligibility work, so the answer is often a part-time hire plus an answering layer.
- Run a 30-day overflow test before posting the job and compare abandoned calls in the peak hours.
The request is usually about two hours, not eight
The case for another receptionist usually arrives as a complaint about mornings. The lines light up at eight, the waiting room fills, and the two people at the desk are checking in patients with a caller on hold. That is a real problem. It is not always a full-time-job-sized problem, and the right answer depends on numbers most practices already have but rarely pull.
Practices are weighing the same choice under cost pressure. In a June 2026 MGMA poll, 84% of medical groups said operating costs were running higher than at the same point a year earlier[4]. Asked about their top planned cost-cutting move for 2026, practice leaders put automation first at 36%, ahead of process fixes at 23%, hiring freezes or cuts at 18% and outsourcing at 14%[1]. Intent has run ahead of action, though: 68% said they had not redesigned a role or adjusted staffing with the help of AI in the past year[5].
The method below takes about a week of evenings: four numbers from your phone system, a sort of calls by who has to take them, an honest price for each option, and a 30-day test before anyone writes a job ad.
Find the peak hour, not the daily average
Pull four ordinary weeks of call reports by hour of day: calls offered, calls answered, calls abandoned and average talk time. Leave out holiday weeks and flu-season spikes for the baseline; plan for those separately. Daily averages hide the problem. A desk handling 80 calls a day looks fine on paper until you see that 28 of them land between 8 and 10 a.m.
Turn the peak into talk minutes per hour: calls in the busiest hour multiplied by average talk time. Then compare it with phone time actually available. Someone who is also checking in patients, scanning cards and taking copays does not have 60 phone minutes in an hour, and in many offices it is closer to 30. If you are unsure, time it for a week, as set out in measuring front desk workload before you hire.
Example: 400 calls a week, with 35% of each day's calls between 8 and 10 a.m. That is 28 calls in two hours, or 14 an hour. At 3.5 minutes each, the peak needs 49 talk minutes an hour. Two staff with about 30 phone minutes each have 60, so on paper it fits. Add one 10-minute insurance call or a patient at the window with a form question and the queue builds. The abandoned-call column will show exactly where.
If abandoned calls cluster in two hours and the rest of the day is quiet, a full-time hire buys six slack hours to fix two busy ones. If abandonment is spread across the day, or the line at the window is the real bottleneck, the case for another person gets much stronger.
Sort calls by who has to take them
Next, take 100 calls from the reports or recordings and put each one in one of three buckets. The split decides what automation can absorb today and what only a person with system access can do.
| Bucket | Typical calls | Who should take them |
|---|---|---|
| Needs the chart or judgment | Scheduling with visit-type rules, eligibility and prior authorization, results, refills, billing disputes | Staff with access to the schedule, payer portals and chart |
| Needs only approved information | Hours, holiday closures, directions and parking, fax number, portal help, whether you take new patients, which plans you accept | Anyone or anything with the approved answers: a trained float, an answering service, an SI Employee |
| Should not reach the desk | Vendor pitches, wrong numbers, robocalls, calls meant for the building manager | A phone menu option, a blocked-number list or a corrected listing |
The first bucket is where the hours go. Practice leaders polled by MGMA in March 2026 named eligibility and prior authorization (45%) and scheduling (31%) as the most time-consuming phone tasks for staff, far ahead of intake (9%) and refills (6%)[6]. Those calls need someone who can see the schedule and the payer portal. Do not expect online booking to drain them either: only 11% of medical group leaders said most of their patients self-schedule[7].
The second bucket is small in minutes but large in interruptions. A thirty-second βwhat time do you close?β costs far more than thirty seconds when it pulls a receptionist out of a prior authorization form. That bucket is what automation handles well today. The third bucket is a process fix, not a staffing decision; stopping vendor and sales calls from clogging lines covers it.
Price the hire properly
The median wage for receptionists in physician offices was $19.01 an hour in May 2025[2]. Wages are only part of the cost: for private-industry office and administrative support workers, benefits made up 31.3% of total compensation in June 2026[3]. Grossing the median wage up at that ratio gives about $27.67 an hour, or roughly $57,550 a year for 2,080 paid hours. That is before recruiting, the training hours a new person needs, and the colleague time spent covering while they learn.
Then add the churn risk. Front desk and entry-level administrative roles remain among the most common churn points in medical practices[8], and the BLS projects about 105,100 receptionist openings a year through 2035 even as total employment edges down, mostly to replace people who leave[9]. Assume you will refill the seat at some point; the cost of replacing a front desk employee walks through that figure.
A hire buys things nothing else does: a second pair of hands at the window, copays collected, forms checked, cover when someone is off sick, and cross-training into scheduling or referrals in the quiet hours. If the bottleneck is physical, meaning the line at the window rather than the phone, hire.
Example: a part-time receptionist covering 8 a.m. to noon, five days a week, works 20 hours a week, or 1,040 hours a year. At the median physician-office wage that is about $19,770 in wages[2], plus whatever benefits the practice offers part-time staff. It puts a second person at the window for exactly the hours the peak report flagged, and leaves the afternoon as it is.
Ask the desk as well as the logs. Staff usually know which call type breaks their concentration, and their answer often differs from the one that takes the most minutes. Both matter: minutes tell you how many hours to buy, interruptions tell you which work to move.
Price what automation actually takes off the desk
Automation has to be priced against the bucket it can take, not the whole phone load. The calculator below uses the example desk: 80 calls a weekday, three in ten needing only approved information, three minutes each once the interruption is counted.
| Input | Example |
|---|---|
| Inbound calls per weekday | 80 |
| Minutes per call, including the interruption | 3 |
| Share that need only approved information | 30% |
| Loaded hourly cost of a receptionist | $27.67 |
| Calls a month | 1,736 |
| Staff hours on these calls | 87 |
| Hours handed off | 26 |
| Monthly value of that staff time | $721 |
How it is worked out: Calls a day Γ 21.7 working days Γ minutes per call Γ· 60 Γ share handled = hours; hours Γ hourly cost.
Assumptions
- Hourly cost of $27.67: the $19.01 median receptionist wage in physician offices, grossed up for benefits at 31.3% of total compensation.[3]
- Calls needing the chart, the schedule or a payer portal stay with staff and are not counted here.[6]
- 80 calls a weekday and a 30% share are example inputs. Replace them with your own four-week sort.
Example only. The result is staff time on calls that need nothing but approved answers, not a revenue figure.
About 26 hours a month in this example, or roughly six hours a week, returned in hundreds of small pieces. That is worth having, and it is not a person. Arithmetic like this is why the choice is rarely one or the other.
Be precise about what the automated layer does today. For a medical office, an SI Employee answers office and new-patient questions from the information the practice approves, gives the approved 911 instruction when a caller describes an emergency, and hands results, refills, bills and appointments to the staff line you choose without taking health details. It can cover lunch, overflow when the lines are busy, and after hours. Booking into a practice management system is planned rather than live, so it relieves the second bucket, not the first. The full list is on the what it does page.
Decision rules
- Hire when the window is the bottleneck, when abandoned calls are spread across the whole day, or when most peak calls fall in the first bucket and need the chart.
- Automate the information calls when they are a large share of the peak and staff lose time to interruptions rather than to long calls.
- Do both when the morning peak is mixed: cover overflow and lunch with an answering layer, and hire part-time for the window in the busiest hours instead of a full-time seat. Float and part-time staff covers the scheduling side.
- Do neither, and fix a process, when one call type dominates for a fixable reason, such as refill status calls caused by a slow refill workflow.
- Your team
- System
01Your team
Pull four weeks of logs
Calls offered, answered and abandoned by hour, plus average talk time.
02Your team
Find the peak hour
Talk minutes needed in the busiest hour against phone minutes available.
03Your team
Sort 100 calls
Needs the chart, needs approved information only, or should not reach the desk.
04Your team
Price both options
Loaded cost of a hire against the desk hours an answering layer would free.
05System
Test for 30 days
Forward overflow in the peak hours only and compare abandoned calls.
Example sequence. Steps one to four take about a week; the test takes a month.
Test for 30 days before posting the job
Week 1
Forward overflow during the two peak hours and over lunch only. Staff still answer first; the answering layer takes what rings through.
Weeks 2 and 3
Read ten call summaries a day. Note any call that should have reached a person sooner and tighten the handoff rules.
Week 4
Compare abandoned calls in the peak hours with the baseline weeks, and ask the desk what changed at the window.
Decide
If the window is still the bottleneck, post a part-time role for the peak. If the queue cleared, revisit the numbers in a quarter.
The family medicine morning-rush example models this kind of setup for a two-clinician practice taking about 400 calls a week, most of them before ten. For the money side in more depth, see calculating ROI on front desk automation and setting a competitive receptionist rate. If you land on a mixed answer, hybrid front desk coverage shows how to draw the layers.
Frequently asked questions
How many calls justify a second receptionist?
There is no single number, because the same 80 calls a day can be easy or impossible depending on how many land in one hour and how long they take. Compare talk minutes needed in your busiest hour with phone minutes actually available, and look at where abandoned calls cluster.
Is it cheaper to automate than to hire?
It depends on what the automation can take. In the example above, approved-information calls add up to about 26 staff hours a month, while a full-time receptionist at the median physician-office wage comes to roughly $57,550 a year with benefits[2][3]. Automation that covers part of the work does not replace a person, but it can make a part-time hire enough.
Will patients who want a person be put off?
Keep patient-specific calls going to people, and use automation for questions with fixed answers and for overflow that would otherwise ring out or reach voicemail. Listen to recordings in the first month to hear how callers respond.
What should never be automated at the front desk?
Anything that needs clinical judgment, the chart or a decision about disclosure: triage, results, refill decisions and requests from third parties. Payments at the window and identity checks need a person in the room.
Should we hire part-time instead of full-time?
Often, when the problem is a two- or three-hour peak. A part-time person for the morning window plus cover for overflow and lunch can fit the actual load better than a full-time seat that is quiet all afternoon.
References
Medical Group Management Association (MGMA) (2026). Automation, process fixes top list of cost-cutting moves for medical practices in 2026 (opens in a new tab) β
U.S. Bureau of Labor Statistics (2025). OEWS series OEUN000000062110043417108: Receptionists and information clerks in offices of physicians, hourly median wage (opens in a new tab) β
U.S. Bureau of Labor Statistics (2026). Employer Costs for Employee Compensation, Table 4: private industry workers by occupational and industry group, June 2026 (opens in a new tab) β
Medical Group Management Association (MGMA) (2026). Operating costs keep climbing for medical practices in 2026 (opens in a new tab) β
Medical Group Management Association (MGMA) (2026). AI is slowly redesigning work in medical practices rather than replacing workers (opens in a new tab) β
Medical Group Management Association (MGMA) (2026). Phones are still a bottleneck costing medical practices time they can't afford (opens in a new tab) β
Medical Group Management Association (MGMA) (2024). Putting the power of scheduling into patients' hands (opens in a new tab) β
Medical Group Management Association (MGMA) (2026). Stabilized but not solved: Staff turnover in medical practices looking no better, no worse in 2026 (opens in a new tab) β
U.S. Bureau of Labor Statistics (2025). Occupational Outlook Handbook: Receptionists (opens in a new tab) β