Booking appointments by text: how it actually works
What happens between 'do you have anything Thursday' and a confirmed slot on the calendar: availability, the offer, the pick, the hold, the confirmation, and what can go wrong at each step.
Frontiva · · 4 min read
Booking by text works in six steps. The customer asks for a time. The system reads real availability for the right service and staff member. It offers two or three concrete slots. The customer picks one. The slot is held and written to the calendar immediately. A confirmation goes out with the details and a way to change it. The whole exchange is four or five short messages, and the calendar it writes to is the one your team actually uses.
Step 1: understanding the request
"Do you have anything Thursday for a cleaning" contains a day, a service and, implicitly, "for me". The system needs to know the service (to get the duration and who can do it), the customer (new or existing), and the preferred time. A good AI asks only for what is missing: if they said Thursday and cleaning, it asks nothing and offers times.
Step 2: reading real availability
This is where most booking tools quietly fail. Availability must come from the live calendar, filtered by the service's duration, the staff members who provide it, their working hours at that location, existing appointments, buffers, and time off. A copy of the calendar that syncs every fifteen minutes will eventually offer a slot that was just taken. Our post on two-way calendar sync explains why live matters.
Step 3: offering slots, not a link
"I have Thursday at 10:00 or 2:30 with Maria. Either work?" Two or three options, with the person's name if that matters to the customer. Not a link to a booking page, which asks the customer to start over in a different interface. Offering slots keeps the exchange in the thread and takes one reply to complete.
Step 4: the pick
"2:30" or "the second one" or "yes Thursday afternoon please". The system has to read all three. If the reply is ambiguous ("either is fine"), it picks the first and confirms it, rather than asking again.
Step 5: the hold and the write
The moment the customer picks, the slot is held and written to the calendar so nobody else can take it in the seconds before confirmation. If a deposit is required, the hold has an expiry and the payment link goes out; the slot releases automatically if the deposit does not arrive.
Step 6: the confirmation
Date, time, service, who with, where (with the address for multi-location businesses), what to bring, and how to reschedule: "Reply CHANGE to move it." Followed by a reminder the day before with the same option. The confirmation is also where a new customer gets the intake form link.
What goes wrong, and where
The slot was already taken. Stale availability. Fix: live sync.
Wrong duration. The service was booked at the default length. Fix: durations per service and, where needed, per staff member.
Wrong location. Multi-location business, customer assumed their usual branch. Fix: the confirmation always states the address, and the system asks when the customer has been to more than one.
No reminder, no-show. Fix: reminders with a reschedule option, on a schedule tuned for your kind of appointment.
Frequently asked questions
Can the customer book a specific staff member?
Yes. The system knows who provides each service and each person's hours, and offers only their real availability.
What if the customer wants a time that is not available?
The system says so and offers the nearest alternatives. It does not book over someone else's appointment or outside working hours.
Does the customer need an app?
No. It is a text conversation. The same works on web chat, Instagram and WhatsApp when those channels are connected.
What Frontiva does here
Frontiva's bookings compute live availability per service, staff member and location, and a database constraint refuses the second of two overlapping appointments. Customers book themselves on a public page per location, your team books and moves appointments on the calendar, and SMS reminders follow. The AI does not book inside a text conversation yet; it answers from your knowledge, which can include the link to your booking page. For the AI side, read can an AI receptionist book appointments.