"Is it working?" is the wrong question to ask about an AI receptionist, because the honest answer is always some version of "sort of, probably." The right question is a number: what does it cost you per appointment actually booked?
Cost per booked appointment turns a vague feeling into something you can track month to month, compare against what a booking is worth, and use to test whether a change helped. It's one of the few metrics that connects what you spend directly to the outcome you care about.
Most businesses either don't calculate it or calculate it in a way that doesn't hold up. They count conversations instead of bookings, forget half the cost, or compare a slow month against a busy one and draw the wrong conclusion.
The one idea worth taking away, if you read nothing else, is that a plan-plus-usage bill needs two versions of this number, and the section explaining why sits right after the formula.
What is cost per booked appointment?
Cost per booked appointment is the total you spent on a booking channel in a period, divided by the number of appointments that channel actually created in the same period.
Cost per booked appointment = total spend ÷ appointments actually created
With Aiva Chat, total spend has two parts. There's your plan, which is $49, $99, or $199 a month. Then there's usage billed directly to you by OpenAI for the AI and Twilio for calls and texts, on your own accounts, with no Aiva Chat markup. Add both.
The two words that matter are "actually created." Not conversations. Not calls answered. Not calls that sounded like they were going well. Appointments that landed on your calendar. Everything else is a leading indicator, not the number.
It's closely related to customer acquisition cost, just scoped to one channel and one kind of outcome rather than a whole marketing budget. It's useful for any appointment-based business: clinics, salons, consultants, home services, studios, and anyone else whose revenue starts with a booking.
How to calculate cost per booked appointment
Step 1: Pick a period long enough to mean something
Use a full calendar month at minimum. A week is too short; a few unusual days can swing the result. Seasonal businesses should look at several months before drawing conclusions.
Step 2: Add up total spend
Take your Aiva Chat plan price for the month. Add your OpenAI bill and your Twilio bill for the same dates. Aiva Chat's OpenAI usage card shows estimated spend by type and by assistant, and the call log can show each call's cost, split into the Twilio part and the AI part. Your exact bill is on each provider's own site. The usage costs guide explains where each number lives.
Step 3: Count appointments actually created
Use the Appointments Booked figure on Aiva Chat's Analytics page, or count appointments on your calendar that came through Aiva Chat. Exclude anything your staff booked by hand.
Step 4: Divide, twice
Divide total spend by bookings for full cost per booking. Then divide usage only by bookings for usage per booking. Write both down.
Step 5: Compare against value
Put the result next to what an average booking is worth to you. That comparison is the real answer to "is this worth it?"
Why you need two versions of the number
If a flat fee were the only cost, the formula would still work, but it would tell you little. In a slow month, a fixed number divided by few bookings spikes. In a busy month, the same fixed number looks great. The metric would track volume instead of performance.
Aiva Chat's cost is split: a fixed plan plus usage that moves with volume. So keep two numbers.
Full cost per booking answers "is this worth what I'm paying?" It drops in busy months because the plan is spread across more bookings.
Usage per booking is the performance number. If bookings double, usage roughly doubles too, so usage per booking stays stable across slow and busy months. A real change in it is telling you something true about how well conversations convert.
Once you have a real cost-per-booking number, the next question answers itself: what's a booked appointment actually worth to you? If it's worth many times what it cost, the spend is behaving more like margin than overhead.
Benefits of tracking cost per booking
It replaces a feeling with a number
"The phones seem quieter" isn't something you can act on. A number you calculate the same way every month shows trends clearly, and it settles debates about whether the assistant is pulling its weight.
It ties spend to revenue
Because the denominator is bookings, the metric connects directly to revenue. Multiply bookings by average booking value and you can see return on spend at a glance, which our guide to AI receptionist ROI expands on.
It shows whether a change helped
Updated your FAQ answers or changed the booking flow? Compare usage per booking for a comparable window before and after. A real improvement shows up there, separate from whatever happened to overall volume.
It reveals which channel converts best
Splitting by phone calls and website chat shows where conversations turn into bookings most efficiently. That points setup effort at the channel that will pay back most.
It keeps vendor comparisons honest
Any tool's pricing can be expressed as cost per booking. Comparing on that basis, rather than on monthly price alone, accounts for how well each one actually converts.
Worked examples at different scales
The businesses and figures below are invented and rounded to keep the arithmetic easy; your own numbers come from your dashboard and your OpenAI and Twilio bills.
A busy appointment-based business
Situation: The $49 Starter plan plus $101 across OpenAI and Twilio bills, for $150 total. 165 new appointments came through Aiva Chat. Calculation: $150 ÷ 165 = about $0.91 full cost per booking. Usage per booking: $101 ÷ 165 = about $0.61. Reading it: If an average booking is worth $80, the spend is a small fraction of the value it produced.
A solo consultant
Situation: The $49 plan plus $11 of usage, mostly website chat, for $60 total. 18 bookings. Calculation: $60 ÷ 18 = about $3.33 per booking. Reading it: Higher, mainly because the plan is spread across fewer bookings. Not automatically worse; it depends on what an 18-booking month is worth to a consultant.
A seasonal caterer
Situation: In a slow month: $49 plan plus $30 usage, 12 bookings. In peak season: $49 plus $145 usage, 58 bookings. Calculation: Slow month: $79 ÷ 12 = about $6.58 full, $2.50 usage per booking. Peak: $194 ÷ 58 = about $3.34 full, still $2.50 usage per booking. Reading it: Full cost per booking halved while usage per booking didn't move. That's the split doing its job.
The zero-booking month
Situation: A brand-new account with real spend but no bookings yet, or a calendar connection that isn't wired up. Calculation: You can't divide by zero, and you shouldn't force a number. Reading it: Treat it as a signal. Either the sample is too small, or something upstream needs checking before the metric is worth trusting.
Cost per booking vs. other metrics
| Metric | What it measures | Strength | Weakness |
|---|---|---|---|
| Cost per booked appointment | Spend per confirmed booking | Ties spend to revenue | Needs a reliable booking count |
| Usage per booking | Variable spend per booking | Stable across busy and slow months | Ignores the fixed plan cost |
| Cost per call answered | Spend per conversation | Easy to calculate | Rewards volume, not outcomes |
| Resolved without a person | Share of conversations handled fully | Shows conversation quality | Doesn't show revenue |
| Handoff rate | Share passed to staff | Shows where the AI hits limits | Some handoffs are exactly right |
| Cost per booking with a receptionist | Wages per booking handled | Useful benchmark | Leaves out everything else the person does |
Cost per call answered is the easiest number to get and the most misleading. A tool that answers lots of calls without booking any looks efficient on that measure.
Resolution and handoff rates tell you about conversation quality, which explains why cost per booking moves. Watch them together: rising cost per booking with falling resolution points to a conversation problem; rising cost with steady resolution is usually seasonal mix.
The receptionist benchmark is worth doing once. The US Bureau of Labor Statistics reports that the median hourly wage for receptionists "was $18.27 in May 2025." At 40 hours a week, that's roughly $3,170 a month in wages alone. If that person handled 220 bookings in a month alongside everything else, wages work out to about $14 per booking, before benefits and with no coverage after hours. Receptionists do far more than book, so treat this as a sanity check rather than a case against hiring: it compares Aiva Chat with what was answering the phone before instead of with free.
How to set up cost-per-booking tracking
Step 1: Make sure every booking is counted
Connect Google Calendar so Aiva Chat bookings are written there, and confirm bookings appear on the Analytics page. If your booking count is wrong, everything downstream is wrong.
Step 2: Record manual bookings separately
Add bookings staff take by hand to Aiva Chat's Appointments page so the calendar is complete, but keep them out of the Aiva Chat booking count for this metric.
Step 3: Turn on the cost column
In the call log, turn on the Cost column to see each call's Twilio and AI cost. It helps explain spikes.
Step 4: Keep a simple monthly sheet
Columns: month, plan, OpenAI, Twilio, total, bookings from calls, bookings from chat, bookings from texts (if text booking is on), full cost per booking, usage per booking, average booking value. Five minutes a month.
Step 5: Add value per booking
Use average first-visit value to start. For repeat-visit businesses, our breakdowns of what a clinic can save in a year and what a salon can save in a year show how lifetime value changes the picture.
Step 6: Review quarterly and after changes
Compare quarters, and run before-and-after comparisons whenever you change answers, rules, or the booking flow.
Common cost-per-booking mistakes
Counting conversations as bookings
A conversation that ended with "I'll think about it" is not a booking. Including them makes the number look far better than reality and hides conversion problems.
Forgetting half the cost
Plan price alone understates spend; usage alone ignores the plan. Add both, using the same dates for each bill.
Judging from one month
A single slow or busy month distorts full cost per booking. Seasonal businesses especially should look at a full season before deciding anything.
Comparing your number to someone else's
The same cost per booking means different things for a clinic booking modest consultations and a boutique booking large custom orders. Compare against your own booking value and your own history.
Editing the past for cancellations
Removing cancelled bookings after the fact makes the metric inconsistent. Count bookings when created, and track cancellations separately. If cancellations spike, look at the path from missed call to booked appointment and your confirmation process.
Best practices for cost per booked appointment
- Calculate monthly, review quarterly. Consistency matters more than precision.
- Always keep both numbers: full cost per booking and usage per booking.
- Split by channel: calls, website chat, and texts if Booking by text is on.
- Use the same date range for plan, OpenAI, and Twilio figures.
- Compare against booking value, not against zero.
- Run before-and-after checks whenever you change the assistant's answers or flow.
- Watch resolution and handoff rates alongside it to explain movements.
Conclusion
If you track one number for an AI receptionist, make it usage per booking, with full cost per booking beside it. The first tells you whether conversations are converting better or worse; the second tells you whether the whole thing is worth what you pay. Reading only the full figure will have you celebrating busy months and panicking in slow ones for no reason.
The caveat is the denominator. The metric is only as honest as your booking count, so make sure staff-made bookings stay out of it and cancellations are tracked separately rather than subtracted later.
The decision point comes after two or three months of clean numbers. If a booking is worth many times what it costs to produce, the useful question becomes which other calls or channels the assistant should be taking on.
Ready to see your own cost per booking?
Set up the monthly sheet from the tracking section before you start, so the first month counts. Then run the assistant on your calls and website chat and fill in the columns from Analytics and your two provider bills.
The 14-day trial needs no card, which gives you half a month of real bookings to divide into. After that, Starter is $49 a month; the plans and their limits page shows when a higher tier would make sense. Call +1 (970) 972-0974 to hear an assistant book an appointment.
Start your free trial, or book a demo and we'll help you wire up the tracking.
Posts from the people building Aiva Chat in Rajkot — how the product works, what we have learned building it, and guides for getting the most out of an AI receptionist.