Playbook · appointment reminder automation ontario
A no-show is a booking you already paid twice for
What a booking reminder sequence costs to build in Ontario, how to count what no-shows cost you before you spend, and the CASL rule that decides the SMS half.
Lasse Pettersen
The most expensive appointment in your diary is the one nobody turns up to. You paid to win it once through advertising, referral or search, and you pay again by holding a slot that could have gone to somebody else. The booking was already bought. The no-show is where it gets thrown away.
This is the automation I would build first in a clinic, a salon, a pest control business, a mobile mechanic or any Ontario firm that runs a diary. Not because it is clever. Because it is the only one on the list where the return is a revenue number rather than a labour saving, and revenue is the larger number in a business of three to twenty people.
Count it before you automate it
Do not buy this on a percentage you read somewhere. Industry no-show figures get quoted with great confidence and none of them were measured in your business. Yours is on your own calendar and it takes ten minutes to pull.
Three numbers:
- How many booked appointments did not happen last month. Cancellations inside 24 hours count. A slot released at 8am for a 9am appointment is not a slot you refilled.
- What one appointment is worth. Not the invoice. The margin, or the invoice if you would rather keep the arithmetic simple and conservative.
- How many of those slots you refilled. Usually close to none, and this is the number owners overestimate.
A clinic with 6 missed appointments a week at $140 each is losing $840 a week. Over a working year that is $43,680. A mobile trade with 3 missed jobs a week at $600 each is losing a larger number than most of its software budget combined.
Now you have the figure the build has to beat. If the annual loss is under about $6,000, the honest answer is that this automation does not pay yet and you should read which jobs to automate first instead. That happens, and it happens most often in businesses under about five people.
What the sequence actually does
The trigger is a new booking in whatever the business already uses. The build does not replace the booking tool. Replacing the booking tool is a different project with a different price and a much worse return.
| Moment | Channel | What it carries |
|---|---|---|
| On booking | Email and text | Date, time, address or meeting link, what to bring, a calendar file, the reschedule link |
| 24 hours before | Email and text | The same details, with the reschedule link first |
| 2 hours before | Text | Time, address or link, reschedule link |
| 15 minutes before | Text | Address or link only |
| 30 minutes after | Thank you, the next step, and one short feedback question | |
| Marked no-show | Text | One reschedule offer, sent once |
The 24 hour message is the one that pays for the build. It is the only message sent while the customer can still move the appointment rather than abandon it. The 2 hour and 15 minute messages reduce lateness, which is worth something, but they do not recover a booking.
The last row matters more than it reads. A no-show handled with one automatic reschedule offer converts a meaningful share of the time, and the alternative in most small businesses is that nobody ever follows up because the person who would have done it is already running late for the next job.
The calendar file is not a detail
Attach a calendar file to the confirmation email. It puts the appointment in the customer’s own calendar, on their own phone, with their own alert. That single attachment does part of the job of the whole reminder sequence and it costs nothing per message.
The CASL question decides the SMS half
This is the part that gets built wrong, and getting it wrong in Canada is more expensive than the automation.
Canada’s Anti-Spam Legislation governs commercial electronic messages, and text messages are commercial electronic messages. The Act treats two things differently:
- A message that confirms or facilitates a transaction the recipient already agreed to. An appointment confirmation and a reminder for a booking the customer made themselves sit here.
- A message that promotes something. “You are due for a checkup”, “we have a spring offer”, “book your next visit”. These are promotions regardless of how they are worded, and they need consent.
The trap is that both messages leave from the same system to the same customer, so a build that does not separate the two streams quietly turns a compliant reminder pipeline into an unconsented marketing list.
What I do on every build of this kind:
- Collect consent at the booking form, with a plain checkbox and a stored timestamp, even where the reminder itself may not require it. Consent you did not need costs nothing. Consent you needed and cannot evidence is the whole problem.
- Put the business name and a way to stop the messages in every message, including the reminders. The exemption is about consent, not about identification, and building an anonymous message saves nothing.
- Keep the promotional stream in a separate list with its own consent record and its own unsubscribe state.
- Record the unsubscribe against the person, not the phone number, because the same customer books under two numbers more often than you would think.
None of this is legal advice. It is the build standard that keeps the question small, and you should confirm the specifics with a lawyer if you are sending at volume.
What it costs to build
$2,500 to $4,500 for the sequence described above, which is the bottom of the $2,500 to $12,000 range published on the workflow automation page. Running costs of roughly $40 to $120 a month at small business volumes, almost all of it the text messages.
What moves the number:
| Driver | Cheap end | Expensive end |
|---|---|---|
| Booking system | One, with a modern API and a real webhook | Two or more, or a system that only exports a file |
| Write back | Reads the booking only | Writes status, notes and tasks into the CRM |
| Message design | One appointment type, one script | Different scripts per service, per location, per language |
| Consent handling | One list, one unsubscribe state | Separate transactional and promotional consent, with an audit trail |
| Rescheduling | A link to your existing booking page | The customer reschedules inside the message thread |
The row that surprises people is the fourth. Consent handling done properly is real engineering, and it is the row I will not cut to hit a price.
What it must not do
- Never let a model compose the appointment details. The date, the time and the address are copied from the booking record. A language model that writes a time that has to be correct will eventually write the wrong one, and the customer will arrive when it said.
- Never send from a different sender each time. One identity, every message. A reminder from an unrecognised number reads as a scam, and in this province it will be reported as one.
- Never send the promotion inside the reminder. It is a different message under CASL and it teaches the customer to ignore the message that recovers the booking.
- Never let it fire after the appointment happened. Check the booking status before every send. A reminder for an appointment somebody already attended undoes the credibility of the whole sequence.
- Never send at 3am. Quiet hours are a build setting, not a nice idea.
Where this sits against the other automations
If your problem is that the phone is not answered and the booking is never made at all, this is the wrong build. That is intake, it is AI agents work, and the first section of the four automations that pay in a trades business explains why it ranks ahead of everything else in a field services business.
If the bookings arrive but the enquiries before them go cold, the build you want is lead response automation, because a reminder sequence cannot save a lead that was never contacted.
If the diary is full and the problem is the paperwork behind it, look at invoice data entry automation instead.
Statistics Canada put Canadian business AI use at 19.2% in the second quarter of 2026, and rural adoption at 9.9% against 21.0% urban. Reminder automation is one of the reasons that gap is worth closing rather than admiring. It is not artificial intelligence in any interesting sense. It is a diary, a phone number and a schedule, wired together once, and it protects revenue in a business in Collingwood or Sudbury exactly as well as it does in one in Toronto.
The honest test before you spend
Pull the three numbers at the top of this page. If the annual loss beats the build inside a year, this is a straightforward yes and it does not need an assessment first. If it does not, or if you cannot get the numbers out of your booking system at all, that second problem is the real one and the $999 assessment is the cheaper place to start.