Article
A medical office answering service answers the calls your practice cannot: after hours, at weekends, during lunch, or when every line is busy. A live operator picks up, follows your protocol, takes a message and routes anything urgent to the on-call provider.
From published price lists, that costs roughly $1.30 to $1.43 per minute. A managed AI receptionist costs $0.75 to $1.00 per minute and, instead of taking a message, books the appointment directly into your system.
That comparison is the whole decision, and most guides to this category avoid it because they are published by answering services. This one works through what each model actually costs, what each actually delivers, and where the traditional model still wins, because it does, in specific situations.
Key Takeaways
The per-minute cost of a human answering service and a managed AI agent are now close. MAP Communications publishes 1.30–1.43 per minute. MedReception.ai publishes 0.75–1.00. The gap is no longer the argument.
The real difference is what you hold afterwards. An answering service hands you a message your staff must action tomorrow. An AI receptionist hands you a booked appointment. One defers the work; the other removes it.
Most providers will not publish a price. Of the medical answering services reviewed, one published a full rate card and the other quoted only after a consultation, plus a $75 setup fee.
After-hours clinical triage is where human services still earn their rate. If your need is a trained operator reaching an on-call physician at 3am, that is a different purchase from front-desk overflow.
What a medical answering service actually does
The traditional model is message-taking, and it is worth being precise about that because it defines the value.
A call comes in outside hours. A live operator answers in your practice's name, follows a script you supplied, identifies the caller and the reason, and then does one of two things: takes a message for your staff to handle when the office opens, or, if the call meets your urgency criteria, reaches the on-call provider by phone, text or pager.
Standard capabilities across the category:
24/7/365 live answering, including nights, weekends and holidays
Protocol-based call routing by caller type, urgency and provider on call
Message delivery via secure portal, SMS, email or app
On-call escalation to the right clinician under your rules
Bilingual support, usually English and Spanish. Dental practices buy the same service under a different label
Appointment requests. Captured as a request, rarely booked directly
That last point is the one to hold onto. notifyMD states its agents "schedule appointments, answer commonly asked questions, take messages, and direct urgent calls to the on-call provider." Read carefully, scheduling in this context usually means capturing the request. Few traditional services write into your EHR's live availability. Confirm which yours does.
What it costs
MAP Communications is one of the few in this category publishing a full rate card, and it is HIPAA, PCI-DSS, HITRUST and SOC 2 compliant with no long-term contract.
Plan | Monthly | Included minutes | Overage | Effective cost per minute |
|---|---|---|---|---|
Pay As You Go | $49 | 0 | $1.37/min | n/a |
Business | $179 | 125 | $1.30/min | $1.43 |
Enterprise | $339 | 250 | $1.28/min | $1.36 |
Premium | $649 | 500 | $1.28/min | $1.30 |
Now place that against the AI options and against a fuller-service virtual receptionist.

Service | Type | Cost per minute | What you hold after the call |
|---|---|---|---|
Ruby, 500 min | Human virtual receptionist | $3.45 | Call handled, message sent |
MAP Communications, 125 min | Human answering service | $1.43 | Message relayed to staff |
MAP Communications, 500 min | Human answering service | $1.30 | Message relayed to staff |
MedReception.ai, 1,000 min | AI, managed | $1.00 | Booked into the EMR |
MedReception.ai, 2,000 min | AI, managed | $0.75 | Booked into the EMR |
AI, self-built | 0.07–0.31 | Booked, if you build it |
Two observations.
Price is no longer the differentiator between a human answering service and managed AI. At 500 minutes, MAP costs $1.30 a minute. At 1,000 minutes, MedReception.ai costs $1.00. Those are the same order of magnitude.
A full-service human virtual receptionist is a different price class entirely. Ruby works out at $3.45 a minute at its largest published plan, more than double the answering service. You are buying more capability, and paying accordingly.
The billing details that move the real number
Published rates are a starting point. Ask every provider:
What is the billing increment? Per-second, 30-second or per-minute rounding changes the effective rate substantially on short calls.
What counts as billable? Hold time, wrong numbers, hang-ups, solicitation calls.
Is there a setup fee? PatientCalls charges $75 on the first invoice. MAP lists none.
Are there surcharges for holidays, after-hours or bilingual calls?
What is the overage rate, and does it apply from the first minute over?
Note how hard this is to do by comparison shopping: of the two medical answering services reviewed here, only MAP publishes rates at all. PatientCalls offers per-minute, per-call or flat-rate models but quotes only after a consultation.
The cost that never appears on the invoice
Here is the part that decides whether an answering service is good value, and it is not on any rate card.
What happens to the message?
A patient calls at 7pm wanting to move Thursday's appointment. The service answers, takes the details, and delivers a message. The next morning your front desk reads it, opens the schedule, finds a slot, calls the patient back, gets voicemail, leaves a message, and waits. The patient calls back at lunchtime, when the desk is busy.
You have paid roughly $1.30 a minute for the original call and then spent your own staff's time twice more, which is the same gap scheduling software alone cannot close. The work did not disappear; it was deferred and duplicated, and the patient waited a day for something a booking system could have done in ninety seconds.
For context, that staff time is not free either: the median US receptionist earns $38,010 a year, about $18.27 an hour, per the Bureau of Labor Statistics.
This is the structural reason the per-minute comparison understates the gap. A message-taking service and a booking service are not the same product at a similar price. They are different products, and one leaves the task on your desk.
The honest counter-argument: message-taking is the right design when the call genuinely needs human judgement, and when the alternative, an AI that mis-books, is worse. Which brings us to where the traditional model is still clearly correct.
Where a human answering service still wins
After-hours clinical triage. If the job is a trained operator applying urgency criteria and reaching an on-call physician at 3am, that is a clinical workflow with real liability attached. Specialist providers such as TriageLogic build specifically for this, and it is not front-desk overflow.
Low volume. At 50–125 minutes a month, absolute cost is small and the difference between $179 and $99 does not justify an implementation project.
No integration appetite. Forward your line and you are live. No EHR project, no conversation design, no escalation rules to configure.
Genuinely unpredictable calls. Practices whose after-hours volume is mostly complex, distressed or clinical get more from a person than from a booking engine.
Where AI has taken over
Appointment volume. Booking, rescheduling and cancellations are rule-based, verifiable against live availability, and the highest-frequency reason patients call. This is what an AI medical receptionist is built for.
Concurrency. Three patients calling at once are all answered. Human services queue.
Consistent cost at volume. Above roughly 1,000 minutes a month, the per-minute gap compounds into real money.
Daytime overflow. Calls that currently ring out while your desk is with a patient. The baseline there is zero, which makes it the lowest-risk place to start.
Which model fits
Your situation | Likely right answer |
|---|---|
Under ~125 minutes/month, mostly after hours | Human answering service |
After-hours clinical triage to on-call providers | Specialist triage service, bought deliberately |
Daytime overflow, abandoned calls | AI receptionist, piloted on missed calls |
High appointment-booking volume | AI with real-time EHR write-back |
Mixed clinical and administrative after-hours | Hybrid: AI first, human escalation |
HIPAA: the same rules apply to both
An answering service hearing a patient's name and reason for calling is handling PHI, which makes it a business associate, with the same obligations as any AI vendor. HHS requires satisfactory assurances in the form of a signed BAA before you disclose PHI, and that applies identically whether the operator is a person or software.
Ask both kinds of provider:
Will you sign a BAA covering every service we enable?
Where are your operators physically located, and what controls apply if they work from home?
How are messages transmitted, and is SMS or email containing PHI encrypted?
How long are call recordings retained, and can we require deletion?
Who are your subcontractors? For AI vendors this means the speech, language-model and telephony providers; each requires its own BAA before PHI reaches it.
MAP publishes HIPAA, PCI-DSS, HITRUST and SOC 2 compliance, which is a stronger position than a HIPAA badge alone. Ask for the current attestation rather than the logo.
Switching: what to ask before you move
What is our actual after-hours call volume and mix? Pull three months of call data before shopping. Most practices overestimate the clinical share.
What proportion of calls could be resolved rather than relayed? That number is the value of switching.
Can the new provider match our on-call escalation rules exactly? Have them demonstrate it, including the awkward cases.
What is the minimum term and notice period? MAP advertises no long-term contracts; not all competitors match that.
Can we run both in parallel for a month? Sully.ai's AI Receptionist can take daytime overflow while your incumbent keeps after-hours. Route after-hours to the incumbent and daytime overflow to the new system. The comparison is then real rather than theoretical.
Frequently Asked Questions
How much does a medical answering service cost?
From published rates, 1.28–1.43 per minute. MAP Communications lists $49/month pay-as-you-go with $1.37 per minute, up to $649/month for 500 included minutes at an effective $1.30 per minute. Most providers in this category do not publish rates and quote after a consultation, sometimes with a setup fee: PatientCalls charges $75.
What is the average cost for an answering service?
Published medical rates cluster around $1.30 per minute, with a fuller-service virtual receptionist such as Ruby at $3.45 per minute and managed AI at 0.75–1.00. Treat any quoted "average" with caution, because the headline rate matters less than the billing increment, what counts as billable, and the overage rate.
What is an answering service for a doctor's office?
A third-party service whose live operators answer calls your practice cannot take, following protocols you supply. They identify the caller and reason, take a message for your staff, and escalate urgent calls to the on-call provider. The defining characteristic is that they relay rather than resolve.
Is there a free medical answering service?
No, not one that can lawfully handle patient calls. Any service touching PHI must sign a BAA and maintain safeguards, and that has a cost. Free consumer voicemail and auto-attendants exist but do not meet the requirement, and free AI tools generally will not sign a BAA.
Can an answering service book appointments directly?
Some capture appointment requests; relatively few write a confirmed booking into your EHR's live availability. This is the single most important question to ask, because it determines whether you have removed work or moved it to tomorrow morning. AI receptionists are more likely to offer real-time write-back.
Do we need an answering service if we have an AI receptionist?
Often not, but many practices keep a human path for after-hours clinical calls while AI handles administrative volume. The hybrid arrangement, with AI first and human escalation under defined rules, is common and sensible.
Is a medical answering service HIPAA compliant?
Only with a signed BAA covering the service. HHS does not certify vendors, so compliance is contractual. Look for an independent certification such as SOC 2 or HITRUST as evidence, and ask where operators are located and how messages containing PHI are transmitted.
Sources
MAP Communications: Pricing. Published plans: 49pay-as-you-go(1.37/min), $179/125 min, $339/250 min, $649/500 min, overage 1.28–1.37/min; HIPAA, PCI-DSS, HITRUST and SOC 2 compliance; no long-term contracts.
PatientCalls: Pricing. Per-minute, per-call and flat-rate models quoted on consultation; $75 one-time setup fee.
Ruby: Pricing. Virtual Receptionist plans, 500 minutes at $1,725.
MedReception.ai: Pricing. Professional $995/1,000 min; Elite $1,495/2,000 min.
Retell AI: Pricing. 0.07–0.31/min for voice agents plus $0.015/min telephony.
notifyMD: Medical answering service. Stated agent capabilities.
TriageLogic: After-hours answering service for medical offices. Clinical triage positioning.
U.S. Bureau of Labor Statistics: Occupational Outlook Handbook: Receptionists. Median pay $38,010/year, $18.27/hour.
U.S. Department of Health & Human Services: Business Associates guidance. BAA requirement before disclosure of PHI.
All pricing checked against published price lists on 13 October 2026.
Article
Keep reading
Article
/
Dental Answering Service: What It Costs, and Where Missed Calls End Up
What a dental answering service costs, how after-hours dental calls are handled, and where the calls you miss actually end up, with federal ED data.
Article
/
After-Hours Medical Answering Service: Coverage Models and On-Call Routing
What happens to your patients' calls after 5pm, the four coverage models compared, and how to design on-call routing that does not end at the emergency department.
Article
/
Patient Self-Scheduling: What the Evidence Shows, and the Guardrails That Decide It
Self-scheduling halves no-shows, then loses more appointments to cancellation than it saves. What two million appointments reveal, and the guardrails that decide it.




