53% of med spa practices haven't adopted AI at all for content creation, patient communication, analytics, or targeting, according to Growth99's 2026 State of Aesthetic & Elective Wellness Marketing Report (AmSpa, January 2026). Among the 47% that have, most are using it for content and analytics, not front-desk automation. The AI receptionist is the higher-leverage application. Every missed inbound call from a new patient is a $527 first-visit opportunity from a lead that already cost $132 to generate, per the same report. Choosing the right vendor, and knowing which questions to ask before signing, is the harder part.
The med spa AI receptionist market is crowded and moving fast. Vendors range from general-purpose voice AI platforms configured for healthcare to purpose-built systems designed for aesthetic practices. Feature claims cover a wide range, and most demo environments don't reflect the edge cases your front desk deals with daily.
What a med spa AI receptionist does (and doesn't do)
An AI receptionist is software that handles inbound calls, texts, or web chat on behalf of the practice. The core applications are call answering, appointment qualification, and booking. Beyond that, implementations vary widely.
Voice AI, text automation, and hybrid systems
Voice AI handles inbound phone calls. It answers, identifies the caller's intent, asks qualification questions, and books into the practice's calendar. A well-configured system sounds human. Callers have no way to know they're speaking to software. A poorly configured system, with wrong escalation rules, no language handling, or no after-hours logic, loses callers to the next practice that picks up.
Text and chat automation handles inbound SMS and web chat. A prospective patient texts the number listed in your Google Business Profile, or starts a chat on the website. The system qualifies them and surfaces booking options. Response speed drives text conversion. Research on healthcare scheduling puts the threshold at five minutes: leads who wait longer rarely book.
Hybrid systems handle both channels through a unified platform. Some practices route voice to AI and text separately; others run a single vendor across both. The right choice depends on where your inbound volume comes from.
What the system should not handle on its own
Three call types must escalate to a human without exception: clinical questions about treatment suitability for a specific medical history, any mention of complications or adverse reactions, and callers who sound distressed or confused. Any system worth buying has hard escalation rules for all three. If a vendor demo doesn't show you what happens when a caller says "I had a treatment yesterday and something doesn't look right," that gap in the demo is a gap in the product.
How to evaluate any AI receptionist vendor
Most vendor demos show the system at its best. The questions below surface the edge cases that determine whether the product works in production.
| Evaluation criterion | Question to ask | What the answer reveals |
|---|---|---|
| Booking depth | "Does it book directly into my calendar system, or collect a lead for my team to follow up?" | Lead-only systems automate the call but not the conversion. Your front desk still handles follow-up. You've added a layer, not removed one. |
| HIPAA compliance | "Will you sign a HIPAA Business Associate Agreement before we go live?" | A caller's name, phone number, and appointment reason are Protected Health Information. No BAA means you cannot use the product compliantly. |
| Language handling | "Does it auto-detect Spanish from the first sentence, or does the caller press a number to select?" | In most metro markets, a meaningful percentage of aesthetic patients prefer Spanish. Press-1 menus create friction that voice AI is supposed to eliminate. |
| After-hours behavior | "What does it do at 9pm when we're closed — offer booking, or tell callers to call back?" | After-hours calls are a disproportionate share of new patient inquiries. A system that defers loses those leads to the next practice with 24/7 coverage. |
| Clinical escalation | "Show me exactly what happens when a caller mentions a complication or reaction." | Escalation must be immediate and unconditional. Any hesitation from the vendor on this question is a hard disqualifier. |
Six questions to ask before you sign
The evaluation table covers what to look for. These steps are how to test it before you commit.
- Test the integration live, not in a sandbox. Ask the vendor to demonstrate a real booking into your actual calendar — Mindbody, Vagaro, Boulevard, Jane, or Zenoti. Most demos use test environments that don't reflect your specific configuration. Integration behavior in a sandbox is not predictive of integration behavior in production. If the vendor can't demo live, the integration isn't ready.
- Ask for the HIPAA Business Associate Agreement before signing anything. Not after the contract. Not "in process." Before you give them access to any patient-facing channel. If the vendor redirects to a general privacy policy or says they're still getting HIPAA-ready, that is a no. A BAA is a legal requirement for any vendor that processes Protected Health Information on your behalf, not a feature they add later.
- Call the demo number speaking Spanish from the first word. Don't say "hola" and switch to English. Open in Spanish and stay there for three sentences. A system that detects Spanish from the opening sentence and responds in Spanish has the detection right. A system that ignores the language signal, defaults to English, or routes to a press-1 menu has a dropout problem with a large patient segment in most urban markets.
- Call the demo number at 9pm. See what the system actually does. Does it offer to schedule? Does it recite business hours and hang up? Does it capture your name and number and promise a morning callback? After-hours behavior is often the most revealing test. It tells you what the vendor prioritized, and whether their team thought through what happens to a patient calling at 9pm.
- Call with a complication scenario. Say: "I had a treatment two days ago and I'm concerned about how it looks." The system should route immediately to a human or capture contact information for an urgent callback. It should not attempt to reassure you, ask follow-up questions about the treatment, or suggest rescheduling. If it does any of those things, the system is misconfigured in a way that creates clinical and legal risk for your practice.
- Ask what data it writes back to your marketing stack. A system that books appointments but doesn't tag the source channel (Google Ads, Instagram, organic search, referral) breaks your patient acquisition cost calculation. You've automated the booking but broken the attribution. For how this feeds into your full CAC calculation, see the med spa patient acquisition cost guide. And for whether AI receptionist fees belong in your marketing budget line, see the med spa marketing budget benchmark guide.
What one unconverted new patient call actually costs
Most practices frame the AI receptionist decision around the monthly software fee. The cost that matters more is a missed call. The numbers below use verified figures from Growth99's 2026 report.
The break-even calculation
A solo-provider practice with a two-person front desk misses four or five inbound new patient calls per month. That's not unusual. At those numbers, adoption pays for itself at most price points. A system that answers but doesn't convert doesn't justify the fee.
The HIPAA compliance trap
Many practices assume any cloud-based software tool is HIPAA compliant. It isn't. HIPAA compliance requires a signed Business Associate Agreement (BAA) between your practice and any vendor that stores, processes, or transmits Protected Health Information. A caller's name, phone number, and appointment type all constitute PHI. A vendor that can't produce a BAA creates a compliance exposure for your practice.
Before signing any AI receptionist contract, ask: "Will you send me your HIPAA Business Associate Agreement today?" A vendor with a real compliance program sends it the same day. A vendor without one deflects to their privacy policy or says it's "in process."
Generic voice AI platforms built on consumer infrastructure, tools not built for healthcare, cannot offer a BAA. Some well-known sales automation tools fall into this category. Check before you sign, not after six months of patient calls have gone through the system.
The AI receptionist handles the call. Spa Ledger handles the numbers behind it.
An AI receptionist books the appointment. Spa Ledger tells you whether that appointment came from Google Ads, Instagram, or referral, and whether the service booked is profitable. Weekly P&L by service line, injectable COGS by provider, and patient acquisition cost by channel. Every week, not at year-end.
Get Started